cluster/ec: Add support for hardware acceleration

This patch implements functionalities for fast encoding/decoding
using hardware support. Currently optimized x86_64, SSE and AVX is
added.

Additionally this patch implements a caching mecanism for inverse
matrices to reduce computation time, as well as a new method for
computing the inverse that takes quadratic time instead of cubic.

Finally some unnecessary memory copies have been eliminated to
further increase performance.

Change-Id: I26c75f26fb4201bd22b51335448ea4357235065a
BUG: 1289922
Signed-off-by: Xavier Hernandez <xhernandez@datalab.es>
Reviewed-on: http://review.gluster.org/12837
Tested-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
Smoke: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Pranith Kumar Karampuri <pkarampu@redhat.com>

38 files changed, 21106 insertions, 12302 deletions

diff --git a/xlators/cluster/ec/src/Makefile.am b/xlators/cluster/ec/src/Makefile.am
index cbdceefdbe0..c5d9ab1812b 100644
--- a/xlators/cluster/ec/src/Makefile.am
+++ b/xlators/cluster/ec/src/Makefile.am
@@ -12,8 +12,11 @@ ec_sources += ec-dir-write.c
 ec_sources += ec-inode-read.c
 ec_sources += ec-inode-write.c
 ec_sources += ec-combine.c
-ec_sources += ec-gf.c
 ec_sources += ec-method.c
+ec_sources += ec-galois.c
+ec_sources += ec-code.c
+ec_sources += ec-code-c.c
+ec_sources += ec-gf8.c
 ec_sources += ec-heal.c
 ec_sources += ec-heald.c
 
@@ -24,10 +27,34 @@ ec_headers += ec-data.h
 ec_headers += ec-fops.h
 ec_headers += ec-common.h
 ec_headers += ec-combine.h
-ec_headers += ec-gf.h
 ec_headers += ec-method.h
+ec_headers += ec-galois.h
+ec_headers += ec-code.h
+ec_headers += ec-code-c.h
+ec_headers += ec-gf8.h
 ec_headers += ec-heald.h
 ec_headers += ec-messages.h
+ec_headers += ec-types.h
+
+if ENABLE_EC_DYNAMIC_INTEL
+  ec_sources += ec-code-intel.c
+  ec_headers += ec-code-intel.h
+endif
+
+if ENABLE_EC_DYNAMIC_X64
+  ec_sources += ec-code-x64.c
+  ec_headers += ec-code-x64.h
+endif
+
+if ENABLE_EC_DYNAMIC_SSE
+  ec_sources += ec-code-sse.c
+  ec_headers += ec-code-sse.h
+endif
+
+if ENABLE_EC_DYNAMIC_AVX
+  ec_sources += ec-code-avx.c
+  ec_headers += ec-code-avx.h
+endif
 
 ec_ext_sources = $(top_builddir)/xlators/lib/src/libxlator.c
 
diff --git a/xlators/cluster/ec/src/ec-code-avx.c b/xlators/cluster/ec/src/ec-code-avx.c
new file mode 100644
index 00000000000..92bd3e83c5e
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-avx.c
@@ -0,0 +1,116 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#include <errno.h>
+
+#include "ec-code-intel.h"
+
+static void
+ec_code_avx_prolog(ec_code_builder_t *builder)
+{
+    builder->loop = builder->address;
+}
+
+static void
+ec_code_avx_epilog(ec_code_builder_t *builder)
+{
+    ec_code_intel_op_add_i2r(builder, 32, REG_DX);
+    ec_code_intel_op_add_i2r(builder, 32, REG_DI);
+    ec_code_intel_op_test_i2r(builder, builder->width - 1, REG_DX);
+    ec_code_intel_op_jne(builder, builder->loop);
+
+    ec_code_intel_op_ret(builder, 0);
+}
+
+static void
+ec_code_avx_load(ec_code_builder_t *builder, uint32_t dst, uint32_t idx,
+                 uint32_t bit)
+{
+    if (builder->linear) {
+        ec_code_intel_op_mov_m2avx(builder, REG_SI, REG_DX, 1,
+                                   idx * builder->width * builder->bits +
+                                   bit * builder->width,
+                                   dst);
+    } else {
+        if (builder->base != idx) {
+            ec_code_intel_op_mov_m2r(builder, REG_SI, REG_NULL, 0, idx * 8,
+                                     REG_AX);
+            builder->base = idx;
+        }
+        ec_code_intel_op_mov_m2avx(builder, REG_AX, REG_DX, 1,
+                                   bit * builder->width, dst);
+    }
+}
+
+static void
+ec_code_avx_store(ec_code_builder_t *builder, uint32_t src, uint32_t bit)
+{
+    ec_code_intel_op_mov_avx2m(builder, src, REG_DI, REG_NULL, 0,
+                               bit * builder->width);
+}
+
+static void
+ec_code_avx_copy(ec_code_builder_t *builder, uint32_t dst, uint32_t src)
+{
+    ec_code_intel_op_mov_avx2avx(builder, src, dst);
+}
+
+static void
+ec_code_avx_xor2(ec_code_builder_t *builder, uint32_t dst, uint32_t src)
+{
+    ec_code_intel_op_xor_avx2avx(builder, src, dst);
+}
+
+static void
+ec_code_avx_xor3(ec_code_builder_t *builder, uint32_t dst, uint32_t src1,
+                 uint32_t src2)
+{
+    ec_code_intel_op_mov_avx2avx(builder, src1, dst);
+    ec_code_intel_op_xor_avx2avx(builder, src2, dst);
+}
+
+static void
+ec_code_avx_xorm(ec_code_builder_t *builder, uint32_t dst, uint32_t idx,
+                 uint32_t bit)
+{
+    if (builder->linear) {
+        ec_code_intel_op_xor_m2avx(builder, REG_SI, REG_DX, 1,
+                                   idx * builder->width * builder->bits +
+                                   bit * builder->width,
+                                   dst);
+    } else {
+        if (builder->base != idx) {
+            ec_code_intel_op_mov_m2r(builder, REG_SI, REG_NULL, 0, idx * 8,
+                                     REG_AX);
+            builder->base = idx;
+        }
+        ec_code_intel_op_xor_m2avx(builder, REG_AX, REG_DX, 1,
+                                   bit * builder->width, dst);
+    }
+}
+
+static char *ec_code_avx_needed_flags[] = {
+    "avx2",
+    NULL
+};
+
+ec_code_gen_t ec_code_gen_avx = {
+    .name   = "avx",
+    .flags  = ec_code_avx_needed_flags,
+    .width  = 32,
+    .prolog = ec_code_avx_prolog,
+    .epilog = ec_code_avx_epilog,
+    .load   = ec_code_avx_load,
+    .store  = ec_code_avx_store,
+    .copy   = ec_code_avx_copy,
+    .xor2   = ec_code_avx_xor2,
+    .xor3   = ec_code_avx_xor3,
+    .xorm   = ec_code_avx_xorm
+};
diff --git a/xlators/cluster/ec/src/ec-code-avx.h b/xlators/cluster/ec/src/ec-code-avx.h
new file mode 100644
index 00000000000..fdca4ad2c8f
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-avx.h
@@ -0,0 +1,18 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#ifndef __EC_CODE_AVX_H__
+#define __EC_CODE_AVX_H__
+
+#include "ec-code.h"
+
+extern ec_code_gen_t ec_code_gen_avx;
+
+#endif /* __EC_CODE_AVX_H__ */
diff --git a/xlators/cluster/ec/src/ec-code-c.c b/xlators/cluster/ec/src/ec-code-c.c
new file mode 100644
index 00000000000..7387f3ea435
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-c.c
@@ -0,0 +1,11431 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#include <inttypes.h>
+#include <string.h>
+
+#include "ec-method.h"
+#include "ec-code-c.h"
+
+#define WIDTH (EC_METHOD_WORD_SIZE / sizeof(uint64_t))
+
+static void gf8_muladd_00(void *out, void *in)
+{
+    memcpy(out, in, EC_METHOD_WORD_SIZE * 8);
+}
+
+static void gf8_muladd_01(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        out_ptr[0] ^= in_ptr[0];
+        out_ptr[WIDTH] ^= in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] ^= in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] ^= in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] ^= in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] ^= in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] ^= in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] ^= in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_02(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in7;
+        out1 = in0;
+        out7 = in6;
+        out5 = in4;
+        out6 = in5;
+        out3 = in2 ^ in7;
+        out4 = in3 ^ in7;
+        out2 = in1 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_03(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in0 ^ in7;
+        tmp0 = in2 ^ in7;
+        out1 = in0 ^ in1;
+        out7 = in6 ^ in7;
+        out5 = in4 ^ in5;
+        out6 = in5 ^ in6;
+        out4 = in3 ^ in4 ^ in7;
+        out2 = tmp0 ^ in1;
+        out3 = tmp0 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_04(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in6;
+        out1 = in7;
+        out7 = in5;
+        out6 = in4;
+        tmp0 = in6 ^ in7;
+        out2 = in0 ^ in6;
+        out5 = in3 ^ in7;
+        out3 = tmp0 ^ in1;
+        out4 = tmp0 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_05(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in0 ^ in6;
+        out1 = in1 ^ in7;
+        out7 = in5 ^ in7;
+        out6 = in4 ^ in6;
+        out2 = out0 ^ in2;
+        out3 = out1 ^ in3 ^ in6;
+        out5 = out7 ^ in3;
+        out4 = out6 ^ in2 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_06(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in6 ^ in7;
+        tmp0 = in1 ^ in6;
+        out1 = in0 ^ in7;
+        out7 = in5 ^ in6;
+        out6 = in4 ^ in5;
+        out4 = in2 ^ in3 ^ in6;
+        out5 = in3 ^ in4 ^ in7;
+        out3 = tmp0 ^ in2;
+        out2 = tmp0 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_07(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in6;
+        tmp1 = in5 ^ in6;
+        tmp2 = in0 ^ in7;
+        tmp3 = tmp0 ^ in3;
+        out6 = tmp1 ^ in4;
+        out7 = tmp1 ^ in7;
+        out0 = tmp2 ^ in6;
+        out1 = tmp2 ^ in1;
+        out3 = tmp3 ^ in1;
+        out4 = tmp3 ^ in4;
+        out5 = out4 ^ out7 ^ in2;
+        out2 = tmp0 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_08(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in5;
+        out1 = in6;
+        out7 = in4;
+        out6 = in3 ^ in7;
+        out3 = in0 ^ in5 ^ in6;
+        out5 = in2 ^ in6 ^ in7;
+        out2 = in5 ^ in7;
+        out4 = out2 ^ in1 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_09(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in0 ^ in5;
+        tmp0 = in3 ^ in6;
+        out1 = in1 ^ in6;
+        out7 = in4 ^ in7;
+        out2 = in2 ^ in5 ^ in7;
+        out3 = tmp0 ^ out0;
+        out6 = tmp0 ^ in7;
+        out4 = out1 ^ out7 ^ in5;
+        out5 = out2 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_0A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in5 ^ in7;
+        out1 = in0 ^ in6;
+        out7 = in4 ^ in6;
+        out2 = in1 ^ in5;
+        out6 = out0 ^ in3;
+        out3 = out0 ^ out1 ^ in2;
+        out5 = out7 ^ in2 ^ in7;
+        out4 = out2 ^ in3 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_0B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in5;
+        tmp1 = in0 ^ in6;
+        tmp2 = in4 ^ in7;
+        out0 = in0 ^ in5 ^ in7;
+        out2 = tmp0 ^ in1;
+        out1 = tmp1 ^ in1;
+        out6 = tmp1 ^ out0 ^ in3;
+        out7 = tmp2 ^ in6;
+        out4 = tmp2 ^ out6 ^ in1;
+        out3 = out6 ^ in0 ^ in2;
+        out5 = tmp0 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_0C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in5 ^ in6;
+        out1 = in6 ^ in7;
+        out7 = in4 ^ in5;
+        tmp0 = in1 ^ in5;
+        tmp1 = in0 ^ in7;
+        out5 = in2 ^ in3 ^ in6;
+        out6 = in3 ^ in4 ^ in7;
+        out2 = tmp1 ^ out0;
+        out4 = tmp0 ^ in2;
+        out3 = tmp0 ^ tmp1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_0D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in4 ^ in5;
+        tmp1 = in5 ^ in6;
+        out1 = in1 ^ in6 ^ in7;
+        out7 = tmp0 ^ in7;
+        out4 = tmp0 ^ in1 ^ in2;
+        out0 = tmp1 ^ in0;
+        tmp2 = tmp1 ^ in3;
+        out6 = tmp2 ^ out7;
+        out2 = out0 ^ in2 ^ in7;
+        out3 = out0 ^ out1 ^ in3;
+        out5 = tmp2 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_0E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        tmp1 = in2 ^ in5;
+        tmp2 = in5 ^ in6;
+        out1 = in0 ^ in6 ^ in7;
+        out3 = tmp0 ^ tmp1;
+        out2 = tmp0 ^ tmp2;
+        tmp3 = tmp1 ^ in3;
+        out7 = tmp2 ^ in4;
+        out0 = tmp2 ^ in7;
+        out4 = tmp3 ^ in1 ^ in7;
+        out5 = tmp3 ^ out7;
+        out6 = out0 ^ out5 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_0F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in6 ^ in7;
+        tmp1 = tmp0 ^ in1;
+        tmp2 = tmp0 ^ in5;
+        out1 = tmp1 ^ in0;
+        out7 = tmp2 ^ in4;
+        out0 = tmp2 ^ in0;
+        out6 = out7 ^ in3;
+        out5 = out6 ^ in2 ^ in7;
+        tmp3 = tmp1 ^ out0 ^ in2;
+        out4 = tmp1 ^ out5;
+        out2 = tmp3 ^ in6;
+        out3 = tmp3 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_10(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in4;
+        out1 = in5;
+        out7 = in3 ^ in7;
+        tmp0 = in6 ^ in7;
+        out2 = in4 ^ in6;
+        tmp1 = out2 ^ in5;
+        out6 = tmp0 ^ in2;
+        out3 = tmp0 ^ tmp1;
+        out5 = out2 ^ out3 ^ in1;
+        out4 = tmp1 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_11(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out7 = in3;
+        out0 = in0 ^ in4;
+        out1 = in1 ^ in5;
+        out6 = in2 ^ in7;
+        out4 = in0 ^ in5 ^ in6;
+        out5 = in1 ^ in6 ^ in7;
+        out2 = in2 ^ in4 ^ in6;
+        out3 = in3 ^ in4 ^ in5 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_12(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in4 ^ in7;
+        out1 = in0 ^ in5;
+        out3 = in2 ^ in4 ^ in5;
+        tmp0 = out0 ^ in6;
+        out2 = tmp0 ^ in1;
+        tmp1 = tmp0 ^ in3;
+        out6 = tmp0 ^ out3;
+        out5 = out2 ^ in5;
+        out7 = tmp1 ^ in4;
+        out4 = tmp1 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_13(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out7 = in3 ^ in6;
+        tmp0 = in0 ^ in5;
+        tmp1 = in4 ^ in7;
+        out6 = in2 ^ in5 ^ in7;
+        out4 = tmp0 ^ out7 ^ in7;
+        out1 = tmp0 ^ in1;
+        out0 = tmp1 ^ in0;
+        out5 = tmp1 ^ in1 ^ in6;
+        out3 = tmp1 ^ out6 ^ in3;
+        out2 = out5 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_14(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in4 ^ in6;
+        out1 = in5 ^ in7;
+        out2 = in0 ^ in4;
+        tmp0 = out0 ^ in5;
+        out7 = out1 ^ in3;
+        tmp1 = out1 ^ in2;
+        out3 = tmp0 ^ in1;
+        out6 = tmp0 ^ tmp1;
+        out4 = tmp1 ^ out2;
+        out5 = out3 ^ in3 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_15(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out7 = in3 ^ in5;
+        tmp0 = in0 ^ in4;
+        out1 = in1 ^ in5 ^ in7;
+        out5 = in1 ^ in3 ^ in6;
+        out0 = tmp0 ^ in6;
+        out2 = tmp0 ^ in2;
+        out3 = out5 ^ in4 ^ in5;
+        out6 = out2 ^ in0 ^ in7;
+        out4 = tmp0 ^ out6 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_16(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in5;
+        tmp1 = in4 ^ in7;
+        tmp2 = in2 ^ in3 ^ in4;
+        out1 = tmp0 ^ in7;
+        out4 = tmp0 ^ tmp2;
+        out0 = tmp1 ^ in6;
+        tmp3 = tmp1 ^ in1;
+        out6 = out0 ^ in2 ^ in5;
+        out2 = tmp3 ^ in0;
+        out3 = out6 ^ in1;
+        out7 = tmp2 ^ out6;
+        out5 = tmp3 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_17(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in5;
+        tmp1 = in3 ^ in6;
+        tmp2 = tmp0 ^ in4;
+        out4 = tmp0 ^ in0 ^ in3;
+        out7 = tmp1 ^ in5;
+        tmp3 = tmp1 ^ in1;
+        out6 = tmp2 ^ in7;
+        out5 = tmp3 ^ in4;
+        out3 = tmp3 ^ out6;
+        out0 = out3 ^ out4 ^ in1;
+        out2 = out3 ^ out7 ^ in0;
+        out1 = tmp2 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_18(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in4 ^ in5;
+        out1 = in5 ^ in6;
+        tmp0 = in4 ^ in7;
+        out5 = in1 ^ in2 ^ in5;
+        out6 = in2 ^ in3 ^ in6;
+        out2 = tmp0 ^ out1;
+        out7 = tmp0 ^ in3;
+        tmp1 = tmp0 ^ in0;
+        out3 = tmp1 ^ in6;
+        out4 = tmp1 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_19(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out5 = in1 ^ in2;
+        out7 = in3 ^ in4;
+        tmp0 = in0 ^ in7;
+        out6 = in2 ^ in3;
+        out1 = in1 ^ in5 ^ in6;
+        out0 = in0 ^ in4 ^ in5;
+        out4 = tmp0 ^ in1;
+        tmp1 = tmp0 ^ in6;
+        out2 = tmp1 ^ out0 ^ in2;
+        out3 = tmp1 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_1A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in4 ^ in5;
+        tmp1 = in5 ^ in6;
+        tmp2 = tmp0 ^ in1;
+        out0 = tmp0 ^ in7;
+        out1 = tmp1 ^ in0;
+        tmp3 = tmp1 ^ in3;
+        out5 = tmp2 ^ in2;
+        out2 = tmp2 ^ in6;
+        out7 = tmp3 ^ out0;
+        out6 = tmp3 ^ in2;
+        out4 = tmp3 ^ out2 ^ in0;
+        out3 = tmp0 ^ out1 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_1B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in4;
+        tmp1 = in2 ^ in5;
+        tmp2 = in3 ^ in6;
+        out5 = tmp0 ^ in1;
+        tmp3 = tmp0 ^ in0;
+        out6 = tmp1 ^ in3;
+        out0 = tmp1 ^ tmp3 ^ in7;
+        out7 = tmp2 ^ in4;
+        tmp4 = out5 ^ in6;
+        out3 = tmp2 ^ tmp3;
+        out2 = tmp4 ^ in5;
+        out4 = tmp4 ^ out3;
+        out1 = tmp3 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_1C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in3;
+        tmp1 = in4 ^ in6;
+        tmp2 = in5 ^ in7;
+        out6 = tmp0 ^ tmp1;
+        out0 = tmp1 ^ in5;
+        out1 = tmp2 ^ in6;
+        tmp3 = tmp2 ^ in1;
+        tmp4 = tmp2 ^ in4;
+        out2 = tmp4 ^ in0;
+        out7 = tmp4 ^ in3;
+        out5 = tmp0 ^ tmp3;
+        out3 = tmp3 ^ out2;
+        out4 = out3 ^ in2 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_1D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in3;
+        tmp1 = in0 ^ in4;
+        tmp2 = in3 ^ in4;
+        tmp3 = in2 ^ in7;
+        out3 = tmp0 ^ tmp1;
+        out5 = tmp0 ^ tmp3;
+        tmp4 = tmp1 ^ in5;
+        out6 = tmp2 ^ in2;
+        out7 = tmp2 ^ in5;
+        out2 = tmp3 ^ tmp4;
+        out4 = out3 ^ out6 ^ in6;
+        out0 = tmp4 ^ in6;
+        out1 = out2 ^ out4 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_1E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in4;
+        tmp1 = in2 ^ in7;
+        tmp2 = tmp0 ^ in1;
+        out3 = tmp1 ^ tmp2;
+        out2 = tmp2 ^ in5;
+        out4 = out3 ^ in3 ^ in6;
+        tmp3 = out4 ^ in7;
+        out6 = tmp3 ^ out2 ^ in4;
+        out7 = tmp1 ^ out6;
+        out0 = out7 ^ in3;
+        out1 = tmp0 ^ out0;
+        out5 = tmp3 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_1F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in4 ^ in6;
+        tmp1 = tmp0 ^ in5;
+        out7 = tmp1 ^ in3;
+        out0 = tmp1 ^ in0 ^ in7;
+        out6 = out7 ^ in2 ^ in6;
+        out1 = out0 ^ in1 ^ in4;
+        out4 = out0 ^ out6 ^ in1;
+        out3 = tmp0 ^ out4;
+        out2 = out4 ^ out7 ^ in7;
+        out5 = out3 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_20(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in4;
+        out0 = in3 ^ in7;
+        tmp0 = in3 ^ in4;
+        tmp1 = in6 ^ in7;
+        out2 = out0 ^ in5;
+        out4 = tmp0 ^ in5;
+        out3 = tmp0 ^ tmp1;
+        out7 = tmp1 ^ in2;
+        out6 = tmp1 ^ in1 ^ in5;
+        out5 = out2 ^ out3 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_21(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in1 ^ in4;
+        tmp0 = in4 ^ in6;
+        out4 = in3 ^ in5;
+        out7 = in2 ^ in6;
+        out0 = in0 ^ in3 ^ in7;
+        out6 = in1 ^ in5 ^ in7;
+        out3 = tmp0 ^ in7;
+        out5 = tmp0 ^ in0;
+        out2 = out4 ^ in2 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_22(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in3;
+        out1 = in0 ^ in4;
+        out7 = in2 ^ in7;
+        out4 = in4 ^ in5 ^ in7;
+        out5 = in0 ^ in5 ^ in6;
+        out6 = in1 ^ in6 ^ in7;
+        out3 = in2 ^ in3 ^ in4 ^ in6;
+        out2 = in1 ^ in3 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_23(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out7 = in2;
+        out0 = in0 ^ in3;
+        out4 = in5 ^ in7;
+        out5 = in0 ^ in6;
+        out6 = in1 ^ in7;
+        out3 = in2 ^ in4 ^ in6;
+        out1 = in0 ^ in1 ^ in4;
+        out2 = out4 ^ out6 ^ in2 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_24(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in4 ^ in7;
+        tmp0 = in3 ^ in4;
+        out0 = in3 ^ in6 ^ in7;
+        out3 = tmp0 ^ in1;
+        tmp1 = out0 ^ in5;
+        out6 = tmp1 ^ out3;
+        out2 = tmp1 ^ in0;
+        out7 = tmp1 ^ in2 ^ in3;
+        out5 = out2 ^ in4;
+        out4 = tmp0 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_25(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in1 ^ in4;
+        tmp0 = in2 ^ in5;
+        out1 = out3 ^ in7;
+        out7 = tmp0 ^ in6;
+        out6 = out1 ^ in5;
+        out4 = out7 ^ in3 ^ in7;
+        out2 = out4 ^ in0;
+        out0 = tmp0 ^ out2;
+        out5 = out0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_26(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in3 ^ in6;
+        tmp0 = in4 ^ in7;
+        out7 = in2 ^ in5 ^ in7;
+        tmp1 = out0 ^ in0 ^ in5;
+        out1 = tmp0 ^ in0;
+        tmp2 = tmp0 ^ in6;
+        out2 = tmp1 ^ in1;
+        out5 = tmp1 ^ in7;
+        out6 = tmp2 ^ in1;
+        out4 = tmp2 ^ out7;
+        out3 = out0 ^ out6 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_27(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out7 = in2 ^ in5;
+        out0 = in0 ^ in3 ^ in6;
+        out6 = in1 ^ in4 ^ in7;
+        out4 = out7 ^ in6;
+        out2 = out0 ^ out7 ^ in1;
+        out5 = out0 ^ in7;
+        out1 = out6 ^ in0;
+        out3 = out6 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_28(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in3;
+        out1 = in4 ^ in6;
+        out0 = in3 ^ in5 ^ in7;
+        tmp0 = out1 ^ in7;
+        tmp1 = out0 ^ in4;
+        out7 = tmp0 ^ in2;
+        tmp2 = tmp0 ^ in1;
+        out3 = tmp1 ^ in0;
+        out6 = tmp1 ^ tmp2;
+        out4 = tmp2 ^ in3;
+        out5 = out3 ^ in2 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_29(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in2 ^ in3;
+        tmp0 = in1 ^ in3;
+        tmp1 = in4 ^ in6;
+        tmp2 = in0 ^ in4 ^ in7;
+        out6 = tmp0 ^ in5;
+        out4 = tmp0 ^ in6 ^ in7;
+        out1 = tmp1 ^ in1;
+        out7 = tmp1 ^ in2;
+        out3 = tmp2 ^ in5;
+        out5 = tmp2 ^ in2;
+        out0 = out3 ^ in3 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_2A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in3 ^ in5;
+        tmp0 = in1 ^ in3;
+        tmp1 = in0 ^ in4;
+        out7 = in2 ^ in4 ^ in7;
+        out3 = tmp1 ^ out0 ^ in2;
+        out2 = tmp0 ^ in7;
+        out6 = tmp0 ^ in6;
+        out1 = tmp1 ^ in6;
+        out5 = tmp1 ^ out7 ^ in5;
+        out4 = out1 ^ in0 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_2B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in1 ^ in6;
+        out7 = in2 ^ in4;
+        tmp0 = in0 ^ in5;
+        tmp1 = in2 ^ in7;
+        out6 = in1 ^ in3;
+        out1 = out4 ^ in0 ^ in4;
+        out3 = tmp0 ^ out7;
+        out0 = tmp0 ^ in3;
+        out5 = tmp1 ^ in0;
+        out2 = tmp1 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_2C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in5;
+        tmp1 = in2 ^ in3 ^ in4;
+        tmp2 = tmp0 ^ in6;
+        out4 = tmp1 ^ in1;
+        out5 = tmp1 ^ in0 ^ in5;
+        tmp3 = tmp2 ^ in4;
+        out6 = tmp2 ^ out4;
+        out7 = tmp3 ^ in7;
+        out2 = tmp3 ^ out5;
+        out3 = out6 ^ in0;
+        out0 = tmp1 ^ out7;
+        out1 = tmp0 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_2D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in3;
+        out4 = tmp0 ^ in1;
+        tmp1 = tmp0 ^ in0;
+        out2 = tmp1 ^ in6;
+        out5 = tmp1 ^ in4;
+        tmp2 = out2 ^ in2;
+        tmp3 = tmp2 ^ in5;
+        out0 = tmp3 ^ in7;
+        out7 = tmp3 ^ out5;
+        out6 = out4 ^ out7 ^ in6;
+        out3 = tmp2 ^ out6;
+        out1 = out0 ^ out6 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_2E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in4 ^ in7;
+        out0 = in3 ^ in5 ^ in6;
+        tmp1 = tmp0 ^ in0;
+        tmp2 = tmp0 ^ in2;
+        out1 = tmp1 ^ in6;
+        out4 = tmp2 ^ in1;
+        out7 = tmp2 ^ in5;
+        out3 = out0 ^ out4 ^ in0;
+        out2 = out3 ^ out7 ^ in7;
+        out6 = tmp1 ^ out2;
+        out5 = tmp1 ^ out7 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_2F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3;
+        tmp1 = in2 ^ in5;
+        out4 = in1 ^ in2 ^ in7;
+        out6 = in1 ^ in3 ^ in4;
+        out5 = tmp0 ^ in2;
+        tmp2 = tmp0 ^ in6;
+        out7 = tmp1 ^ in4;
+        out0 = tmp2 ^ in5;
+        out2 = tmp2 ^ out4;
+        out1 = tmp2 ^ out6 ^ in7;
+        out3 = tmp1 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_30(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in4 ^ in5;
+        tmp0 = in3 ^ in6;
+        tmp1 = in4 ^ in7;
+        out6 = in1 ^ in2 ^ in5;
+        out3 = tmp0 ^ in5;
+        out4 = tmp0 ^ in0;
+        out7 = tmp0 ^ in2;
+        out0 = tmp1 ^ in3;
+        out2 = tmp1 ^ out3;
+        out5 = tmp1 ^ in0 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_31(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in5 ^ in6;
+        tmp0 = in4 ^ in5;
+        tmp1 = in0 ^ in3 ^ in4;
+        tmp2 = out3 ^ in2;
+        out1 = tmp0 ^ in1;
+        out0 = tmp1 ^ in7;
+        out4 = tmp1 ^ in6;
+        out6 = tmp2 ^ in1;
+        out2 = tmp2 ^ out0 ^ in0;
+        out5 = out1 ^ in0 ^ in7;
+        out7 = tmp0 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_32(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in3 ^ in4;
+        out7 = in2 ^ in3;
+        tmp0 = in5 ^ in6;
+        tmp1 = in0 ^ in7;
+        out6 = in1 ^ in2;
+        out1 = in0 ^ in4 ^ in5;
+        out2 = tmp0 ^ out0 ^ in1;
+        out3 = tmp0 ^ out7 ^ in7;
+        out4 = tmp1 ^ in6;
+        out5 = tmp1 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_33(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in3;
+        tmp1 = in0 ^ in4;
+        tmp2 = in1 ^ in5;
+        out6 = in1 ^ in2 ^ in6;
+        out7 = tmp0 ^ in7;
+        out0 = tmp1 ^ in3;
+        out1 = tmp1 ^ tmp2;
+        tmp3 = tmp2 ^ in7;
+        tmp4 = tmp2 ^ in4 ^ in6;
+        out5 = tmp3 ^ in0;
+        out3 = tmp3 ^ out6;
+        out4 = tmp4 ^ out5;
+        out2 = tmp0 ^ tmp4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_34(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in4;
+        tmp1 = in4 ^ in5;
+        tmp2 = tmp0 ^ in1;
+        tmp3 = tmp0 ^ in6;
+        out1 = tmp1 ^ in7;
+        tmp4 = tmp1 ^ in2;
+        out5 = tmp2 ^ in0;
+        out3 = tmp2 ^ out1;
+        out0 = tmp3 ^ in7;
+        out7 = tmp3 ^ tmp4;
+        out6 = tmp4 ^ in1;
+        out2 = out3 ^ out5 ^ in3;
+        out4 = tmp4 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_35(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in6;
+        tmp1 = in5 ^ in7;
+        out7 = tmp0 ^ tmp1 ^ in3;
+        out3 = tmp1 ^ in1;
+        out1 = out3 ^ in4;
+        tmp2 = out1 ^ in7;
+        out5 = tmp2 ^ in0 ^ in3;
+        out6 = tmp0 ^ tmp2;
+        out0 = out3 ^ out5 ^ in6;
+        out4 = tmp0 ^ out0;
+        out2 = out4 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_36(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in0 ^ in2;
+        tmp0 = in1 ^ in3;
+        out0 = in3 ^ in4 ^ in6;
+        out6 = in1 ^ in2 ^ in4;
+        out5 = tmp0 ^ in0;
+        tmp1 = out5 ^ in5;
+        out2 = tmp1 ^ in4;
+        out3 = tmp1 ^ out4;
+        out1 = tmp0 ^ out2 ^ in7;
+        out7 = out3 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_37(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in2;
+        tmp1 = in2 ^ in4;
+        tmp2 = tmp0 ^ in6;
+        out3 = tmp0 ^ in5;
+        out4 = tmp1 ^ in0;
+        out6 = tmp2 ^ in4;
+        out1 = out3 ^ out4 ^ in7;
+        tmp3 = out4 ^ in1 ^ in3;
+        out7 = tmp3 ^ out1;
+        out2 = tmp3 ^ in5;
+        out5 = tmp1 ^ out2;
+        out0 = tmp2 ^ tmp3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_38(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in0 ^ in3;
+        tmp0 = in3 ^ in4;
+        tmp1 = in5 ^ in7;
+        tmp2 = out3 ^ in1;
+        out2 = tmp0 ^ in6;
+        out0 = tmp0 ^ tmp1;
+        out4 = tmp1 ^ tmp2;
+        out7 = out2 ^ in2;
+        out1 = out2 ^ in3 ^ in5;
+        out6 = out4 ^ in0 ^ in2;
+        out5 = tmp2 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_39(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in0;
+        tmp0 = in1 ^ in5;
+        tmp1 = tmp0 ^ in4;
+        out1 = tmp1 ^ in6;
+        out5 = out1 ^ in0 ^ in2;
+        tmp2 = tmp0 ^ out5;
+        out2 = tmp2 ^ in0 ^ in3;
+        out7 = out2 ^ in7;
+        out6 = tmp1 ^ out7;
+        out4 = tmp2 ^ out6;
+        out0 = out4 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_3A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        tmp1 = in0 ^ in2;
+        tmp2 = in3 ^ in4;
+        tmp3 = in1 ^ in6;
+        tmp4 = in3 ^ in7;
+        out4 = tmp0 ^ in5;
+        out5 = tmp1 ^ tmp3;
+        out3 = tmp1 ^ tmp4;
+        out0 = tmp2 ^ in5;
+        out7 = tmp2 ^ in2;
+        tmp5 = tmp3 ^ in4;
+        out2 = tmp4 ^ tmp5;
+        out1 = tmp5 ^ out4;
+        out6 = tmp0 ^ out3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_3B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in6;
+        tmp1 = in2 ^ in7;
+        tmp2 = tmp0 ^ in3;
+        out3 = tmp1 ^ in0;
+        out6 = tmp1 ^ tmp2;
+        out2 = out6 ^ in4;
+        out7 = tmp0 ^ out2;
+        out0 = out3 ^ out7 ^ in5;
+        out5 = out0 ^ out2 ^ in7;
+        out1 = tmp2 ^ out0;
+        out4 = out1 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_3C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3;
+        tmp1 = in2 ^ in7;
+        tmp2 = in1 ^ in6 ^ in7;
+        out2 = tmp0 ^ in4;
+        out3 = tmp0 ^ tmp2;
+        out4 = tmp1 ^ out3 ^ in5;
+        out5 = tmp2 ^ out2 ^ in2;
+        out1 = out4 ^ out5 ^ in6;
+        out0 = out1 ^ in3;
+        out7 = tmp1 ^ out0;
+        out6 = tmp2 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_3D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in2;
+        tmp1 = tmp0 ^ in3;
+        out2 = tmp1 ^ in4;
+        tmp2 = out2 ^ in5;
+        out4 = tmp2 ^ in1 ^ in6;
+        out5 = out4 ^ in7;
+        out6 = out5 ^ in0;
+        out7 = out6 ^ in1;
+        out0 = tmp0 ^ out7;
+        out1 = tmp1 ^ out5;
+        out3 = tmp2 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_3E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in5;
+        tmp1 = tmp0 ^ in4;
+        out0 = tmp1 ^ in6;
+        out7 = tmp1 ^ in2;
+        out6 = out7 ^ in1 ^ in5 ^ in7;
+        out2 = out6 ^ in0 ^ in2;
+        out4 = out0 ^ out6 ^ in0;
+        out5 = tmp0 ^ out4;
+        out3 = out5 ^ in7;
+        out1 = out3 ^ out6 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_3F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        out3 = tmp0 ^ in2 ^ in6;
+        tmp1 = out3 ^ in5 ^ in7;
+        out4 = tmp1 ^ in4;
+        out5 = tmp1 ^ in3;
+        out1 = out4 ^ in2;
+        out7 = out1 ^ out3 ^ in3;
+        out2 = tmp0 ^ out7 ^ in5;
+        tmp2 = out2 ^ in0;
+        out6 = tmp2 ^ in6;
+        out0 = tmp1 ^ tmp2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_40(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in3 ^ in7;
+        tmp0 = in3 ^ in4;
+        tmp1 = in6 ^ in7;
+        out4 = tmp0 ^ in2;
+        out5 = tmp0 ^ in5;
+        out0 = tmp1 ^ in2;
+        out7 = tmp1 ^ in1 ^ in5;
+        out2 = out0 ^ in4;
+        out3 = out2 ^ out5 ^ in7;
+        out6 = out3 ^ out4 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_41(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in2 ^ in3;
+        tmp0 = in5 ^ in6;
+        tmp1 = in6 ^ in7;
+        out5 = in3 ^ in4;
+        out1 = in1 ^ in3 ^ in7;
+        out6 = in0 ^ in4 ^ in5;
+        out3 = tmp0 ^ in2;
+        out7 = tmp0 ^ in1;
+        out2 = tmp1 ^ in4;
+        out0 = tmp1 ^ in0 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_42(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in2 ^ in6;
+        out5 = in3 ^ in5;
+        out1 = in0 ^ in3 ^ in7;
+        out7 = in1 ^ in5 ^ in7;
+        out4 = in2 ^ in4 ^ in7;
+        out6 = in0 ^ in4 ^ in6;
+        out2 = out0 ^ in1 ^ in4;
+        out3 = out5 ^ in6 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_43(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out5 = in3;
+        out7 = in1 ^ in5;
+        out4 = in2 ^ in7;
+        out6 = in0 ^ in4;
+        out0 = in0 ^ in2 ^ in6;
+        out3 = in5 ^ in6 ^ in7;
+        out2 = in1 ^ in4 ^ in6;
+        out1 = in0 ^ in1 ^ in3 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_44(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in3;
+        out0 = in2 ^ in7;
+        tmp0 = in4 ^ in7;
+        out7 = in1 ^ in6 ^ in7;
+        out6 = in0 ^ in5 ^ in6;
+        out4 = tmp0 ^ in3 ^ in6;
+        out3 = out0 ^ in1 ^ in3 ^ in5;
+        out2 = out0 ^ in0 ^ in4;
+        out5 = tmp0 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_45(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in1 ^ in3;
+        out7 = in1 ^ in6;
+        out5 = in4 ^ in7;
+        out6 = in0 ^ in5;
+        out0 = in0 ^ in2 ^ in7;
+        out4 = in3 ^ in6 ^ in7;
+        out2 = out5 ^ in0;
+        out3 = out0 ^ out6 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_46(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in2;
+        out1 = in0 ^ in3;
+        out7 = in1 ^ in7;
+        out4 = in4 ^ in6;
+        out5 = in5 ^ in7;
+        out6 = in0 ^ in6;
+        out3 = in1 ^ in3 ^ in5;
+        out2 = out4 ^ out6 ^ in1 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_47(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in6;
+        out7 = in1;
+        out5 = in7;
+        out6 = in0;
+        tmp0 = in0 ^ in1;
+        out3 = in1 ^ in5;
+        out0 = in0 ^ in2;
+        out1 = tmp0 ^ in3;
+        out2 = tmp0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_48(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in3;
+        out1 = in3 ^ in6 ^ in7;
+        out3 = tmp0 ^ in0;
+        out0 = tmp0 ^ out1 ^ in5;
+        tmp1 = out0 ^ in4;
+        out2 = tmp1 ^ in7;
+        out5 = tmp1 ^ in3;
+        out4 = out5 ^ in1;
+        out7 = tmp0 ^ out4;
+        out6 = tmp1 ^ out3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_49(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in0 ^ in2;
+        tmp0 = in2 ^ in5;
+        out2 = in4 ^ in5 ^ in6;
+        tmp1 = tmp0 ^ out2 ^ in3;
+        out7 = out2 ^ in1;
+        out5 = tmp1 ^ in7;
+        out4 = out5 ^ out7 ^ in6;
+        out1 = tmp0 ^ out4;
+        out6 = out1 ^ out7 ^ in0;
+        out0 = tmp1 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_4A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in6;
+        tmp1 = in3 ^ in7;
+        out0 = tmp0 ^ in5;
+        out3 = tmp1 ^ in0;
+        out5 = tmp1 ^ out0;
+        out4 = out0 ^ in1 ^ in4;
+        out1 = out3 ^ in6;
+        out2 = out4 ^ in7;
+        out6 = out1 ^ in4;
+        out7 = tmp0 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_4B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in0 ^ in7;
+        tmp0 = in1 ^ in5;
+        tmp1 = in2 ^ in6;
+        tmp2 = out3 ^ in3;
+        out7 = tmp0 ^ in4;
+        out4 = tmp0 ^ tmp1;
+        tmp3 = tmp1 ^ in0;
+        out6 = tmp2 ^ in4;
+        out5 = tmp2 ^ tmp3;
+        out1 = tmp2 ^ in1 ^ in6;
+        out2 = out7 ^ in6 ^ in7;
+        out0 = tmp3 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_4C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in3 ^ in6;
+        tmp0 = in2 ^ in5;
+        tmp1 = out1 ^ in5 ^ in7;
+        out0 = tmp0 ^ in7;
+        tmp2 = tmp0 ^ in4;
+        out6 = tmp1 ^ in0;
+        out2 = tmp2 ^ in0;
+        out5 = tmp2 ^ in6;
+        out3 = tmp0 ^ out6 ^ in1;
+        out7 = out0 ^ out5 ^ in1;
+        out4 = tmp1 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_4D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in5;
+        tmp1 = in1 ^ in6;
+        out4 = in1 ^ in3 ^ in5;
+        tmp2 = tmp0 ^ in7;
+        out2 = tmp0 ^ in4;
+        out1 = tmp1 ^ in3;
+        out7 = tmp1 ^ in4;
+        out0 = tmp2 ^ in2;
+        out6 = tmp2 ^ in3;
+        out5 = out7 ^ in1 ^ in2;
+        out3 = tmp1 ^ out0 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_4E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in2 ^ in5;
+        out7 = in1 ^ in4 ^ in7;
+        out1 = in0 ^ in3 ^ in6;
+        out5 = out0 ^ in6;
+        out4 = out7 ^ in5;
+        out3 = out1 ^ in1;
+        out6 = out1 ^ in7;
+        out2 = out4 ^ in0 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_4F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out5 = in2 ^ in6;
+        out7 = in1 ^ in4;
+        out3 = in0 ^ in1 ^ in6;
+        out4 = in1 ^ in5 ^ in7;
+        out0 = in0 ^ in2 ^ in5;
+        out6 = in0 ^ in3 ^ in7;
+        out1 = out3 ^ in3;
+        out2 = out4 ^ in0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_50(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in2 ^ in7;
+        tmp0 = in3 ^ in5;
+        out0 = out2 ^ in4 ^ in6;
+        out1 = tmp0 ^ in7;
+        tmp1 = tmp0 ^ in6;
+        out3 = out0 ^ in3;
+        out7 = tmp1 ^ in1;
+        tmp2 = tmp1 ^ in0;
+        out5 = out3 ^ in1 ^ in2;
+        out4 = tmp2 ^ in2;
+        out6 = tmp2 ^ out3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_51(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in7;
+        out3 = in2 ^ in4 ^ in6 ^ in7;
+        out0 = out3 ^ in0;
+        out6 = out0 ^ in5;
+        out4 = out6 ^ in3 ^ in7;
+        out1 = out0 ^ out4 ^ in1;
+        out7 = out1 ^ in6;
+        out5 = out7 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_52(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in1 ^ in2;
+        tmp0 = in2 ^ in4;
+        tmp1 = in3 ^ in5;
+        tmp2 = in3 ^ in6;
+        tmp3 = in0 ^ in7;
+        out0 = tmp0 ^ in6;
+        out6 = tmp0 ^ tmp3;
+        out7 = tmp1 ^ in1;
+        out1 = tmp1 ^ tmp3;
+        out3 = tmp2 ^ in4;
+        out5 = tmp2 ^ in1 ^ in7;
+        out4 = tmp2 ^ out1 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_53(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in1;
+        out3 = in4 ^ in6;
+        out0 = out3 ^ in0 ^ in2;
+        out6 = out0 ^ in7;
+        out4 = out6 ^ in5;
+        out7 = out0 ^ out4 ^ in1 ^ in3;
+        out1 = out7 ^ in0;
+        out5 = out7 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_54(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in3 ^ in5;
+        tmp0 = in1 ^ in3;
+        tmp1 = in2 ^ in4;
+        tmp2 = in0 ^ in7;
+        out5 = in1 ^ in4 ^ in6;
+        out4 = tmp2 ^ out1;
+        out7 = tmp0 ^ in6;
+        out3 = tmp0 ^ tmp1;
+        out0 = tmp1 ^ in7;
+        tmp3 = tmp2 ^ in2;
+        out2 = tmp3 ^ in6;
+        out6 = tmp3 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_55(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in3;
+        tmp1 = in1 ^ in4;
+        tmp2 = in6 ^ in7;
+        out7 = tmp0 ^ tmp2;
+        out1 = tmp0 ^ in5;
+        out3 = tmp1 ^ in2;
+        out5 = tmp1 ^ in5 ^ in6;
+        out2 = tmp2 ^ in0;
+        out4 = out5 ^ out7 ^ in0;
+        out6 = out2 ^ in2 ^ in5;
+        out0 = out5 ^ out6 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_56(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in2 ^ in4;
+        tmp0 = in0 ^ in2;
+        out4 = in0 ^ in5;
+        out7 = in1 ^ in3;
+        out5 = in1 ^ in6;
+        out6 = tmp0 ^ in7;
+        out2 = tmp0 ^ out5;
+        out1 = out4 ^ in3;
+        out3 = out7 ^ in4 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_57(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in5;
+        tmp1 = in1 ^ in7;
+        out0 = in0 ^ in2 ^ in4;
+        out5 = in1 ^ in5 ^ in6;
+        out4 = tmp0 ^ in4;
+        out1 = tmp0 ^ in1 ^ in3;
+        out2 = tmp0 ^ out5;
+        out3 = tmp1 ^ in4;
+        out7 = tmp1 ^ in3;
+        out6 = tmp1 ^ out2 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_58(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in2 ^ in5;
+        tmp0 = in2 ^ in3 ^ in4;
+        out5 = tmp0 ^ in1;
+        out6 = tmp0 ^ in0 ^ in5;
+        out3 = out6 ^ in7;
+        tmp1 = out2 ^ out5;
+        out7 = tmp1 ^ in6;
+        out4 = tmp1 ^ out3 ^ in3;
+        out0 = out4 ^ out7 ^ in0;
+        out1 = tmp0 ^ out0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_59(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in5;
+        tmp0 = in0 ^ in5 ^ in7;
+        out3 = tmp0 ^ in2 ^ in4;
+        out0 = out3 ^ in6;
+        tmp1 = out0 ^ in7;
+        out6 = tmp1 ^ in3;
+        out5 = out6 ^ in0 ^ in1 ^ in6;
+        out4 = tmp0 ^ out5;
+        out1 = tmp1 ^ out4;
+        out7 = out1 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_5A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in2;
+        tmp1 = in2 ^ in5;
+        out5 = tmp0 ^ in3;
+        out4 = tmp0 ^ in0;
+        tmp2 = tmp1 ^ in4;
+        out2 = tmp1 ^ in1 ^ in7;
+        out7 = tmp2 ^ out5;
+        out6 = out4 ^ out7 ^ in5;
+        out0 = tmp2 ^ in6;
+        out1 = out0 ^ out6 ^ in7;
+        out3 = tmp1 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_5B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in3;
+        tmp1 = in0 ^ in4;
+        tmp2 = in1 ^ in5;
+        out5 = tmp0 ^ tmp2;
+        tmp3 = tmp1 ^ in6;
+        out3 = tmp1 ^ in5;
+        out2 = tmp2 ^ in7;
+        tmp4 = out3 ^ in2;
+        out7 = out2 ^ in3 ^ in4;
+        out0 = tmp4 ^ in6;
+        out6 = tmp0 ^ tmp3;
+        out4 = tmp2 ^ tmp4;
+        out1 = tmp3 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_5C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in6;
+        tmp1 = in0 ^ in2 ^ in5;
+        out1 = tmp0 ^ in5;
+        tmp2 = tmp0 ^ in1;
+        out2 = tmp1 ^ in6;
+        out6 = tmp1 ^ in3;
+        out4 = tmp2 ^ in0;
+        out7 = tmp2 ^ in4;
+        out3 = tmp1 ^ out7;
+        out0 = out3 ^ out4 ^ in7;
+        out5 = out0 ^ in1 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_5D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        tmp1 = in0 ^ in6;
+        out2 = tmp1 ^ in5;
+        tmp2 = out2 ^ in3;
+        out6 = tmp2 ^ in2;
+        out1 = tmp0 ^ tmp2;
+        tmp3 = out1 ^ in4 ^ in5;
+        out4 = tmp3 ^ in0;
+        out7 = tmp3 ^ in7;
+        tmp4 = out4 ^ out6;
+        out5 = tmp4 ^ in7;
+        out0 = tmp0 ^ out5;
+        out3 = tmp1 ^ tmp4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_5E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in5;
+        tmp1 = in3 ^ in5;
+        tmp2 = in1 ^ in7;
+        out7 = in1 ^ in3 ^ in4;
+        out0 = tmp0 ^ in4;
+        tmp3 = tmp1 ^ in0;
+        out5 = tmp2 ^ in2;
+        out1 = tmp3 ^ in6;
+        out6 = tmp0 ^ tmp3;
+        tmp4 = tmp2 ^ out1;
+        out3 = tmp4 ^ in4;
+        out4 = tmp1 ^ tmp4;
+        out2 = tmp0 ^ out4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_5F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in5;
+        tmp1 = in0 ^ in6;
+        tmp2 = tmp0 ^ in7;
+        tmp3 = tmp1 ^ in3;
+        out2 = tmp1 ^ tmp2;
+        out5 = tmp2 ^ in2;
+        out6 = tmp3 ^ in2;
+        out3 = out2 ^ in4;
+        out4 = out3 ^ in5;
+        out1 = tmp0 ^ tmp3;
+        out7 = tmp3 ^ out4;
+        out0 = out4 ^ out5 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_60(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in2 ^ in5;
+        tmp0 = in3 ^ in6;
+        out1 = in3 ^ in4 ^ in7;
+        out7 = out4 ^ in1;
+        tmp1 = out4 ^ in4;
+        out0 = tmp0 ^ in2;
+        out5 = tmp0 ^ in0;
+        out2 = tmp0 ^ tmp1;
+        out3 = tmp1 ^ in7;
+        out6 = out3 ^ out7 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_61(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in5;
+        out4 = tmp0 ^ in4;
+        tmp1 = out4 ^ in3;
+        out3 = tmp1 ^ in7;
+        out2 = tmp1 ^ in2 ^ in6;
+        out1 = tmp0 ^ out3 ^ in1;
+        out0 = out2 ^ out4 ^ in0;
+        out7 = tmp1 ^ out1;
+        out6 = out0 ^ out1 ^ in2;
+        out5 = tmp0 ^ out0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_62(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in4 ^ in5;
+        tmp0 = in0 ^ in3 ^ in4;
+        out1 = tmp0 ^ in7;
+        out5 = tmp0 ^ in6;
+        tmp1 = out1 ^ in0;
+        tmp2 = tmp1 ^ out3;
+        out4 = tmp2 ^ in2;
+        tmp3 = tmp2 ^ in1;
+        out0 = out4 ^ in5 ^ in6;
+        out7 = tmp3 ^ out0;
+        out6 = tmp0 ^ tmp3;
+        out2 = tmp1 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_63(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in4;
+        tmp1 = in1 ^ in7;
+        out3 = tmp0 ^ in5;
+        tmp2 = out3 ^ in6;
+        out4 = out3 ^ in2 ^ in7;
+        out5 = tmp2 ^ in0;
+        tmp3 = out5 ^ in3;
+        out0 = tmp3 ^ out4;
+        out2 = tmp1 ^ tmp2;
+        out6 = tmp1 ^ tmp3;
+        tmp4 = tmp0 ^ out2;
+        out1 = tmp4 ^ out5;
+        out7 = tmp4 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_64(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in2 ^ in3;
+        out1 = in3 ^ in4;
+        out7 = in1 ^ in2;
+        tmp0 = in4 ^ in5;
+        tmp1 = in0 ^ in7;
+        out4 = in5 ^ in6 ^ in7;
+        out2 = tmp0 ^ out0 ^ in0;
+        out3 = tmp0 ^ out7 ^ in6;
+        out5 = tmp1 ^ in6;
+        out6 = tmp1 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_65(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3;
+        tmp1 = in4 ^ in5;
+        tmp2 = in6 ^ in7;
+        out7 = in1 ^ in2 ^ in7;
+        out1 = in1 ^ in3 ^ in4;
+        out0 = tmp0 ^ in2;
+        out2 = tmp0 ^ tmp1;
+        out4 = tmp1 ^ tmp2;
+        tmp3 = tmp2 ^ in0;
+        out3 = out4 ^ out7 ^ in3;
+        out5 = tmp3 ^ in5;
+        out6 = tmp3 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_66(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in2;
+        tmp1 = in2 ^ in3;
+        tmp2 = in0 ^ in4;
+        out7 = tmp0 ^ in6;
+        out0 = tmp1 ^ in7;
+        out1 = tmp2 ^ in3;
+        tmp3 = tmp2 ^ in6;
+        tmp4 = out1 ^ in5;
+        out5 = tmp3 ^ in7;
+        out4 = tmp3 ^ tmp4;
+        out2 = tmp0 ^ tmp4 ^ in7;
+        out6 = tmp1 ^ out2 ^ in4;
+        out3 = tmp3 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_67(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3;
+        tmp1 = tmp0 ^ in1;
+        tmp2 = tmp0 ^ in7;
+        out1 = tmp1 ^ in4;
+        out0 = tmp2 ^ in2;
+        tmp3 = out1 ^ in7;
+        out2 = tmp3 ^ in5;
+        out3 = out2 ^ in0 ^ in6;
+        out7 = tmp1 ^ out0 ^ in6;
+        out5 = tmp1 ^ out3;
+        out4 = tmp2 ^ out5;
+        out6 = tmp3 ^ out4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_68(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in4;
+        tmp1 = in2 ^ in3 ^ in5;
+        tmp2 = tmp0 ^ in1;
+        tmp3 = tmp0 ^ in6;
+        out0 = tmp1 ^ in6;
+        out6 = tmp2 ^ in0;
+        out7 = tmp1 ^ tmp2;
+        out1 = tmp3 ^ in7;
+        out2 = out1 ^ in2;
+        out4 = tmp2 ^ out2;
+        out3 = out4 ^ out6 ^ in3;
+        out5 = tmp3 ^ out3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_69(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in6 ^ in7;
+        out2 = tmp0 ^ in3 ^ in4;
+        out1 = out2 ^ in1;
+        out3 = out2 ^ in0 ^ in2;
+        out4 = out1 ^ in2 ^ in3;
+        out6 = out1 ^ in0 ^ in7;
+        out7 = out4 ^ in5 ^ in6;
+        out5 = out4 ^ out6 ^ in5;
+        out0 = tmp0 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_6A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in6;
+        out3 = in0 ^ in4 ^ in6;
+        tmp1 = tmp0 ^ in3;
+        out4 = tmp1 ^ in1;
+        tmp2 = tmp1 ^ in7;
+        out2 = out4 ^ in4;
+        out0 = tmp2 ^ in5;
+        out5 = tmp2 ^ out3;
+        out7 = out2 ^ in3 ^ in5;
+        out1 = tmp0 ^ out5;
+        out6 = tmp1 ^ out7 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_6B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in4 ^ in6;
+        out2 = tmp0 ^ in1 ^ in3;
+        out4 = out2 ^ in2;
+        tmp1 = out2 ^ in0;
+        out7 = out4 ^ in3 ^ in5 ^ in7;
+        out1 = tmp1 ^ in7;
+        out3 = tmp1 ^ in1;
+        out6 = tmp1 ^ in5;
+        out0 = tmp1 ^ out7 ^ in6;
+        out5 = tmp0 ^ out0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_6C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in1;
+        tmp0 = in2 ^ in3;
+        out5 = in0 ^ in2;
+        out1 = in3 ^ in4 ^ in6;
+        tmp1 = out5 ^ in1;
+        out0 = tmp0 ^ in5;
+        out6 = tmp0 ^ tmp1;
+        out3 = tmp1 ^ in4;
+        out7 = out3 ^ in0;
+        out2 = out6 ^ out7 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_6D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in1 ^ in4;
+        tmp0 = in0 ^ in2;
+        tmp1 = out4 ^ in3;
+        out7 = out4 ^ in2 ^ in7;
+        out5 = tmp0 ^ in5;
+        out3 = tmp0 ^ tmp1;
+        out1 = tmp1 ^ in6;
+        out0 = out5 ^ in3;
+        out2 = out3 ^ out7 ^ in4;
+        out6 = out1 ^ in0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_6E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in3;
+        tmp1 = in0 ^ in4;
+        out4 = tmp0 ^ in7;
+        out6 = tmp0 ^ in0 ^ in5;
+        out5 = tmp1 ^ in2;
+        tmp2 = tmp1 ^ in3;
+        out3 = tmp2 ^ out4;
+        out1 = tmp2 ^ in6;
+        out2 = tmp0 ^ out5;
+        out0 = out2 ^ out3 ^ in5;
+        out7 = out1 ^ out2 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_6F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in7;
+        tmp1 = tmp0 ^ in4;
+        tmp2 = tmp0 ^ in0 ^ in2;
+        out4 = tmp1 ^ in1;
+        out0 = tmp2 ^ in5;
+        out3 = out4 ^ in0;
+        out2 = out3 ^ in7;
+        out1 = out2 ^ in6;
+        out6 = out1 ^ in4 ^ in5;
+        out7 = tmp2 ^ out1;
+        out5 = tmp1 ^ out0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_70(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in2;
+        tmp0 = in2 ^ in4;
+        out2 = in2 ^ in3 ^ in5;
+        tmp1 = tmp0 ^ in6;
+        tmp2 = out2 ^ in7;
+        out0 = tmp1 ^ in3;
+        out4 = tmp1 ^ in0;
+        out7 = tmp2 ^ in1;
+        out6 = out4 ^ in1;
+        out5 = out7 ^ in0 ^ in2;
+        out1 = tmp0 ^ tmp2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_71(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in3 ^ in5;
+        out3 = in2 ^ in3;
+        tmp0 = in0 ^ in2;
+        tmp1 = out2 ^ in1;
+        out4 = tmp0 ^ in6;
+        tmp2 = tmp0 ^ in1;
+        out7 = tmp1 ^ in2;
+        out1 = tmp1 ^ in4 ^ in7;
+        out0 = out4 ^ in3 ^ in4;
+        out6 = tmp2 ^ in4;
+        out5 = tmp2 ^ out3 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_72(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in7;
+        tmp0 = in0 ^ in4;
+        tmp1 = tmp0 ^ in3 ^ in7;
+        out1 = tmp1 ^ in5;
+        out5 = out1 ^ in1;
+        tmp2 = tmp0 ^ out5;
+        out2 = tmp2 ^ in2;
+        out7 = out2 ^ in6;
+        out6 = tmp1 ^ out7;
+        out4 = tmp2 ^ out6;
+        out0 = out4 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_73(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in3 ^ in7;
+        out2 = out3 ^ in1 ^ in5;
+        out1 = out2 ^ in0 ^ in4;
+        out5 = out1 ^ in5;
+        out6 = out1 ^ out3 ^ in2;
+        out0 = out2 ^ out6 ^ in6;
+        out7 = out0 ^ out1 ^ in3;
+        out4 = out0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_74(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in4;
+        tmp1 = in1 ^ in2 ^ in6;
+        out4 = in0 ^ in4 ^ in7;
+        out5 = in0 ^ in1 ^ in5;
+        out0 = tmp0 ^ in2;
+        out1 = tmp0 ^ in5;
+        out3 = tmp1 ^ in7;
+        out6 = tmp1 ^ in0;
+        out2 = tmp1 ^ out5 ^ in3;
+        out7 = out3 ^ in3 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_75(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in0 ^ in7;
+        tmp0 = in1 ^ in3;
+        out5 = in0 ^ in1;
+        out7 = tmp0 ^ in2;
+        tmp1 = tmp0 ^ in4;
+        out6 = out5 ^ in2;
+        tmp2 = out7 ^ in6;
+        out1 = tmp1 ^ in5;
+        out0 = tmp1 ^ out6;
+        out3 = tmp2 ^ in7;
+        out2 = tmp2 ^ out6 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_76(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in1 ^ in6;
+        tmp0 = in0 ^ in5;
+        tmp1 = in3 ^ in7;
+        tmp2 = tmp0 ^ in4;
+        tmp3 = tmp1 ^ in2;
+        out5 = tmp2 ^ in1;
+        out1 = tmp2 ^ in3;
+        out0 = tmp3 ^ in4;
+        out4 = out1 ^ in5;
+        out7 = tmp3 ^ out3;
+        out2 = tmp0 ^ out7;
+        out6 = tmp1 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_77(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in0 ^ in3;
+        tmp0 = in1 ^ in4;
+        tmp1 = in1 ^ in6;
+        tmp2 = out4 ^ in5;
+        out5 = tmp0 ^ in0;
+        out1 = tmp0 ^ tmp2;
+        out3 = tmp1 ^ in3;
+        out2 = tmp1 ^ tmp2 ^ in7;
+        out7 = out3 ^ in2;
+        tmp3 = out7 ^ in6;
+        out6 = tmp2 ^ tmp3;
+        out0 = tmp3 ^ out5 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_78(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3;
+        tmp1 = in2 ^ in7;
+        tmp2 = in0 ^ in5 ^ in6;
+        out2 = tmp1 ^ in3;
+        out3 = tmp2 ^ in2;
+        out5 = out3 ^ in1 ^ in3;
+        out0 = tmp0 ^ out3 ^ in4;
+        out1 = tmp1 ^ out0;
+        out4 = out1 ^ out5 ^ in5;
+        out7 = tmp0 ^ out4;
+        out6 = tmp2 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_79(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in3 ^ in7;
+        tmp0 = in3 ^ in4;
+        tmp1 = in1 ^ in5;
+        tmp2 = tmp1 ^ in2;
+        out4 = tmp2 ^ in0 ^ in7;
+        tmp3 = out4 ^ in5;
+        out5 = tmp3 ^ out2 ^ in6;
+        out7 = tmp0 ^ tmp2;
+        out6 = tmp0 ^ tmp3;
+        out3 = tmp1 ^ out5;
+        out0 = out3 ^ in4;
+        out1 = tmp3 ^ out0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_7A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in2;
+        out2 = tmp0 ^ in3;
+        tmp1 = out2 ^ in4;
+        out4 = tmp1 ^ in0 ^ in5;
+        out5 = out4 ^ in6;
+        out6 = out5 ^ in7;
+        out7 = out6 ^ in0;
+        out0 = out7 ^ in1;
+        out1 = tmp0 ^ out6;
+        out3 = tmp1 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_7B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in1 ^ in3;
+        tmp0 = in0 ^ in5;
+        out4 = tmp0 ^ out2 ^ in2;
+        tmp1 = out4 ^ in4;
+        out6 = tmp1 ^ in7;
+        out5 = tmp1 ^ in5 ^ in6;
+        out0 = out6 ^ in1 ^ in6;
+        tmp2 = out0 ^ in2;
+        out1 = tmp2 ^ in1;
+        out3 = tmp2 ^ in4;
+        out7 = tmp0 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_7C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in5;
+        tmp1 = tmp0 ^ in4;
+        out0 = tmp1 ^ in2;
+        out1 = tmp1 ^ in6;
+        out7 = out0 ^ in1 ^ in5 ^ in7;
+        out5 = out1 ^ out7 ^ in0;
+        out3 = out5 ^ in6;
+        out6 = tmp0 ^ out5;
+        out2 = out6 ^ in1;
+        out4 = out2 ^ out7 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_7D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in2;
+        tmp1 = tmp0 ^ in3;
+        tmp2 = tmp0 ^ in6;
+        out7 = tmp1 ^ in4;
+        tmp3 = tmp2 ^ in0;
+        out5 = tmp3 ^ in7;
+        out4 = tmp3 ^ in2 ^ in5;
+        out2 = tmp1 ^ out5;
+        out6 = tmp2 ^ out2;
+        out0 = out4 ^ out7 ^ in6;
+        out1 = tmp3 ^ out0;
+        out3 = out6 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_7E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in4;
+        tmp1 = in0 ^ in5;
+        out1 = tmp0 ^ tmp1 ^ in6;
+        out3 = tmp1 ^ in1;
+        out4 = out1 ^ in1 ^ in7;
+        tmp2 = out4 ^ in3;
+        out5 = tmp2 ^ in2;
+        out6 = tmp0 ^ out5;
+        out7 = tmp1 ^ out4 ^ in2;
+        out2 = out6 ^ in5 ^ in7;
+        out0 = tmp2 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_7F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in7;
+        tmp1 = tmp0 ^ in3 ^ in5;
+        tmp2 = tmp1 ^ in0;
+        out0 = tmp2 ^ in4;
+        out6 = tmp2 ^ in1;
+        out3 = tmp0 ^ out6;
+        tmp3 = out3 ^ in6;
+        out1 = tmp3 ^ in4;
+        out2 = tmp3 ^ in5;
+        out4 = tmp3 ^ in7;
+        out5 = tmp1 ^ out1;
+        out7 = out0 ^ out4 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_80(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in3;
+        tmp1 = in4 ^ in5;
+        out1 = in2 ^ in6 ^ in7;
+        out5 = tmp0 ^ in4;
+        tmp2 = tmp0 ^ in1;
+        out6 = tmp1 ^ in3;
+        out7 = tmp1 ^ in0 ^ in6;
+        out4 = tmp2 ^ in7;
+        out3 = tmp2 ^ out6;
+        out2 = out3 ^ out5 ^ in6;
+        out0 = out2 ^ in3 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_81(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in4 ^ in6;
+        tmp1 = tmp0 ^ in3;
+        out6 = tmp1 ^ in5;
+        out5 = out6 ^ in2 ^ in6;
+        out3 = out5 ^ in1;
+        out2 = tmp0 ^ out3;
+        out1 = out3 ^ out6 ^ in7;
+        out4 = tmp1 ^ out1;
+        out7 = out2 ^ out4 ^ in0;
+        out0 = out7 ^ in1 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_82(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in1 ^ in2;
+        tmp0 = in6 ^ in7;
+        out5 = in2 ^ in3;
+        out6 = in3 ^ in4;
+        out7 = in0 ^ in4 ^ in5;
+        out0 = in1 ^ in5 ^ in6;
+        out1 = tmp0 ^ in0 ^ in2;
+        out2 = tmp0 ^ in3 ^ in5;
+        out3 = tmp0 ^ out0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_83(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        tmp1 = in2 ^ in5;
+        tmp2 = in3 ^ in6;
+        out4 = in1 ^ in2 ^ in4;
+        out0 = tmp0 ^ in5 ^ in6;
+        out5 = tmp1 ^ in3;
+        tmp3 = tmp1 ^ in7;
+        out6 = tmp2 ^ in4;
+        out2 = tmp2 ^ tmp3;
+        tmp4 = tmp3 ^ out4;
+        out1 = tmp3 ^ out0;
+        out3 = tmp4 ^ in3;
+        out7 = tmp0 ^ tmp4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_84(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in2 ^ in6;
+        out6 = in3 ^ in5;
+        out0 = in1 ^ in5 ^ in7;
+        out7 = in0 ^ in4 ^ in6;
+        out4 = in1 ^ in3 ^ in6;
+        out5 = in2 ^ in4 ^ in7;
+        out2 = out6 ^ in0 ^ in1;
+        out3 = out5 ^ in5 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_85(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in6;
+        tmp1 = in3 ^ in6;
+        tmp2 = tmp0 ^ in4;
+        out1 = tmp0 ^ in2;
+        out6 = tmp1 ^ in5;
+        out4 = tmp2 ^ in3;
+        tmp3 = out1 ^ out6;
+        out2 = tmp3 ^ in0;
+        out3 = tmp2 ^ tmp3 ^ in7;
+        out7 = out2 ^ out3 ^ in1;
+        out5 = tmp1 ^ out3;
+        out0 = tmp2 ^ out7 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_86(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out6 = in3;
+        out7 = in0 ^ in4;
+        out0 = in1 ^ in5;
+        out5 = in2 ^ in7;
+        out3 = in4 ^ in5 ^ in6;
+        out1 = in0 ^ in2 ^ in6;
+        out4 = in1 ^ in6 ^ in7;
+        out2 = in0 ^ in3 ^ in5 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_87(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out6 = in3 ^ in6;
+        tmp0 = in0 ^ in1;
+        out7 = in0 ^ in4 ^ in7;
+        out5 = in2 ^ in5 ^ in7;
+        out3 = out6 ^ in4 ^ in5;
+        out0 = tmp0 ^ in5;
+        tmp1 = tmp0 ^ in6;
+        out2 = out5 ^ in0 ^ in3;
+        out1 = tmp1 ^ in2;
+        out4 = tmp1 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_88(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in2 ^ in7;
+        tmp0 = in5 ^ in6;
+        out0 = in1 ^ in6 ^ in7;
+        out6 = in4 ^ in5 ^ in7;
+        out3 = out0 ^ out1 ^ in0 ^ in4;
+        out7 = tmp0 ^ in0;
+        tmp1 = tmp0 ^ in3;
+        out2 = out0 ^ in3;
+        out4 = tmp1 ^ in2;
+        out5 = tmp1 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_89(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in7;
+        tmp1 = in2 ^ in7;
+        tmp2 = tmp0 ^ in6;
+        out1 = tmp1 ^ in1;
+        out7 = tmp2 ^ in5;
+        out0 = tmp2 ^ in1;
+        out2 = out1 ^ in3 ^ in6;
+        out6 = out7 ^ in0 ^ in4;
+        out5 = out6 ^ in3;
+        out3 = tmp0 ^ out2 ^ in4;
+        out4 = tmp1 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_8A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in1 ^ in6;
+        out7 = in0 ^ in5;
+        out2 = in3 ^ in6;
+        out6 = in4 ^ in7;
+        out1 = in0 ^ in2 ^ in7;
+        out3 = out0 ^ out6 ^ in0;
+        out4 = out1 ^ out7 ^ in6;
+        out5 = out2 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_8B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        tmp1 = in3 ^ in6;
+        tmp2 = in5 ^ in7;
+        tmp3 = tmp0 ^ in7;
+        out0 = tmp0 ^ in6;
+        out2 = tmp1 ^ in2;
+        out5 = tmp1 ^ tmp2;
+        out7 = tmp2 ^ in0;
+        tmp4 = tmp3 ^ in4;
+        out1 = tmp3 ^ in2;
+        out6 = tmp4 ^ out0;
+        out4 = out6 ^ in2 ^ in5;
+        out3 = tmp1 ^ tmp4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_8C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in2;
+        out0 = in1 ^ in7;
+        out7 = in0 ^ in6;
+        out5 = in4 ^ in6;
+        out6 = in5 ^ in7;
+        out2 = out0 ^ in0 ^ in3;
+        out3 = out5 ^ out7 ^ in2 ^ in7;
+        out4 = out6 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_8D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in1 ^ in2;
+        tmp0 = in6 ^ in7;
+        out0 = in0 ^ in1 ^ in7;
+        out5 = in4 ^ in5 ^ in6;
+        out6 = tmp0 ^ in5;
+        out7 = tmp0 ^ in0;
+        out4 = tmp0 ^ out5 ^ in3;
+        out2 = out0 ^ in2 ^ in3;
+        out3 = out2 ^ in1 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_8E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in1;
+        out4 = in5;
+        out7 = in0;
+        out5 = in6;
+        out6 = in7;
+        out3 = in0 ^ in4;
+        out1 = in0 ^ in2;
+        out2 = in0 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_8F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in0 ^ in1;
+        tmp0 = in0 ^ in3;
+        out4 = in4 ^ in5;
+        out7 = in0 ^ in7;
+        out5 = in5 ^ in6;
+        out6 = in6 ^ in7;
+        out1 = out0 ^ in2;
+        out2 = tmp0 ^ in2;
+        out3 = tmp0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_90(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in2;
+        tmp1 = in2 ^ in6 ^ in7;
+        out3 = tmp0 ^ in7;
+        out1 = tmp1 ^ in5;
+        tmp2 = out1 ^ in4;
+        out6 = tmp2 ^ in3;
+        out5 = out6 ^ in1;
+        out4 = out5 ^ in0;
+        out0 = tmp0 ^ tmp2;
+        out7 = tmp0 ^ out4;
+        out2 = tmp1 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_91(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in4;
+        tmp1 = tmp0 ^ in3 ^ in5;
+        out2 = tmp1 ^ in1;
+        out6 = tmp1 ^ in7;
+        tmp2 = out2 ^ in5 ^ in7;
+        out3 = tmp2 ^ in4;
+        out5 = tmp2 ^ in6;
+        out1 = tmp1 ^ out5 ^ in2;
+        tmp3 = out1 ^ in0;
+        out4 = tmp3 ^ in3;
+        out0 = tmp0 ^ tmp3;
+        out7 = tmp2 ^ tmp3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_92(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in1;
+        tmp0 = in4 ^ in5;
+        tmp1 = tmp0 ^ in1;
+        out2 = tmp0 ^ in3 ^ in7;
+        out0 = tmp1 ^ in6;
+        out7 = out2 ^ in0;
+        out4 = out0 ^ in0 ^ in2;
+        out5 = out4 ^ out7 ^ in5;
+        out6 = tmp1 ^ out5;
+        out1 = out6 ^ out7 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_93(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in1 ^ in3;
+        tmp0 = in2 ^ in7;
+        tmp1 = out3 ^ in6;
+        tmp2 = tmp0 ^ in4;
+        out5 = tmp0 ^ tmp1;
+        out6 = tmp2 ^ in3;
+        out2 = out6 ^ in5;
+        out0 = out2 ^ out5 ^ in0;
+        out7 = tmp1 ^ out0;
+        out1 = tmp2 ^ out0;
+        out4 = out1 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_94(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in2 ^ in6;
+        tmp0 = in1 ^ in4 ^ in5;
+        out1 = out3 ^ in5;
+        out5 = tmp0 ^ out3;
+        out0 = tmp0 ^ in7;
+        out4 = tmp0 ^ in0 ^ in3;
+        out6 = out1 ^ in3 ^ in7;
+        out2 = out4 ^ in6;
+        out7 = out0 ^ out2 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_95(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in3;
+        out3 = tmp0 ^ in6;
+        tmp1 = tmp0 ^ in7;
+        tmp2 = out3 ^ in0;
+        out6 = tmp1 ^ in5;
+        tmp3 = tmp2 ^ in4;
+        out7 = tmp3 ^ in2;
+        tmp4 = tmp3 ^ in5;
+        out2 = tmp4 ^ in1;
+        tmp5 = out2 ^ in6;
+        out0 = tmp1 ^ tmp5;
+        out1 = tmp5 ^ out7;
+        out4 = tmp2 ^ out1;
+        out5 = tmp4 ^ out4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_96(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in6 ^ in7;
+        tmp0 = in1 ^ in5;
+        tmp1 = in5 ^ in6;
+        out6 = out3 ^ in2 ^ in3;
+        out0 = tmp0 ^ in4;
+        tmp2 = tmp1 ^ in2;
+        out4 = out0 ^ in0 ^ in7;
+        out1 = tmp2 ^ in0;
+        out5 = tmp2 ^ in1;
+        out7 = tmp0 ^ out4 ^ in3;
+        out2 = tmp1 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_97(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in4;
+        tmp1 = in2 ^ in6;
+        out3 = in3 ^ in6 ^ in7;
+        out7 = tmp0 ^ in3;
+        tmp2 = tmp0 ^ in5;
+        out5 = tmp1 ^ in1;
+        out6 = tmp1 ^ out3;
+        out0 = tmp2 ^ in1;
+        out2 = tmp2 ^ out3 ^ in2;
+        tmp3 = out0 ^ in4;
+        out4 = tmp3 ^ in7;
+        out1 = tmp1 ^ tmp3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_98(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in5 ^ in7;
+        tmp1 = in1 ^ in4 ^ in7;
+        out1 = tmp0 ^ in2;
+        out0 = tmp1 ^ in6;
+        out2 = tmp1 ^ in3;
+        out6 = out0 ^ out1 ^ in1;
+        out5 = tmp0 ^ out2;
+        out3 = tmp1 ^ out6 ^ in0;
+        out7 = out0 ^ out5 ^ in0;
+        out4 = out6 ^ out7 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_99(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3;
+        out5 = in1 ^ in3 ^ in4;
+        out6 = in2 ^ in4 ^ in5;
+        out4 = tmp0 ^ in2;
+        tmp1 = tmp0 ^ in6;
+        tmp2 = out5 ^ in7;
+        out7 = tmp1 ^ in5;
+        out0 = tmp1 ^ tmp2;
+        out2 = tmp2 ^ in2;
+        out3 = out0 ^ out6 ^ in3;
+        out1 = tmp1 ^ out3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_9A(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in3 ^ in4;
+        tmp0 = in0 ^ in5;
+        tmp1 = in1 ^ in6;
+        out5 = in1 ^ in3 ^ in5;
+        tmp2 = tmp0 ^ in7;
+        out3 = tmp0 ^ tmp1;
+        out0 = tmp1 ^ in4;
+        out7 = tmp2 ^ in3;
+        out1 = tmp2 ^ in2;
+        out6 = out0 ^ in1 ^ in2;
+        out4 = out1 ^ in4 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_9B(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out5 = in1 ^ in3;
+        tmp0 = in3 ^ in5;
+        out6 = in2 ^ in4;
+        out4 = in0 ^ in2 ^ in7;
+        out7 = tmp0 ^ in0;
+        out2 = out6 ^ in3;
+        out1 = out4 ^ in1 ^ in5;
+        out3 = out7 ^ in1 ^ in6;
+        out0 = tmp0 ^ out3 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_9C(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out1 = in2 ^ in5;
+        tmp0 = in0 ^ in3 ^ in6;
+        out3 = out1 ^ in0;
+        out6 = out1 ^ in6;
+        out7 = tmp0 ^ in7;
+        out4 = out7 ^ in4;
+        out2 = out4 ^ in1;
+        out0 = tmp0 ^ out2;
+        out5 = out0 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_9D(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out6 = in2 ^ in5;
+        tmp0 = in0 ^ in3;
+        out5 = in1 ^ in4 ^ in7;
+        out1 = out6 ^ in1;
+        out3 = tmp0 ^ out6;
+        out7 = tmp0 ^ in6;
+        out0 = out5 ^ in0;
+        out4 = out7 ^ in7;
+        out2 = out5 ^ out7 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_9E(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in1 ^ in4;
+        tmp0 = in0 ^ in5;
+        out6 = in2 ^ in6;
+        out7 = in0 ^ in3 ^ in7;
+        out4 = in0 ^ in4 ^ in6;
+        out5 = in1 ^ in5 ^ in7;
+        out1 = tmp0 ^ in2;
+        out3 = tmp0 ^ in7;
+        out2 = out4 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_9F(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out6 = in2;
+        out7 = in0 ^ in3;
+        tmp0 = in0 ^ in1;
+        out4 = in0 ^ in6;
+        out5 = in1 ^ in7;
+        out1 = tmp0 ^ in2 ^ in5;
+        out2 = out7 ^ in2 ^ in4 ^ in6;
+        out3 = out7 ^ in5 ^ in7;
+        out0 = tmp0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A0(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in6;
+        out2 = tmp0 ^ in7;
+        tmp1 = tmp0 ^ in5;
+        out6 = out2 ^ in3 ^ in4;
+        out0 = tmp1 ^ in3;
+        tmp2 = out0 ^ in2;
+        out3 = tmp2 ^ in7;
+        tmp3 = tmp2 ^ in1;
+        out5 = tmp3 ^ in0;
+        out4 = tmp3 ^ out6;
+        out7 = out5 ^ out6 ^ in1;
+        out1 = tmp1 ^ out4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A1(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in5;
+        tmp1 = tmp0 ^ in1;
+        tmp2 = tmp0 ^ in4;
+        out4 = tmp1 ^ in7;
+        out7 = tmp2 ^ in0;
+        out6 = tmp2 ^ out4 ^ in3;
+        out3 = out4 ^ in6;
+        out2 = out3 ^ in5;
+        out1 = out2 ^ in4;
+        out5 = out1 ^ out6 ^ in0;
+        out0 = tmp1 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A2(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in6;
+        tmp0 = in1 ^ in3 ^ in5;
+        out3 = tmp0 ^ in6;
+        out4 = tmp0 ^ in2 ^ in4;
+        out0 = out3 ^ in7;
+        out6 = out0 ^ in4;
+        out1 = out0 ^ out4 ^ in0;
+        out7 = out1 ^ in5;
+        out5 = out7 ^ in3 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A3(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in2 ^ in6;
+        out3 = in1 ^ in5 ^ in6;
+        tmp0 = out2 ^ in0;
+        out4 = out2 ^ out3 ^ in3;
+        tmp1 = tmp0 ^ in4;
+        out0 = tmp0 ^ out4 ^ in7;
+        out5 = tmp1 ^ in3;
+        out7 = tmp1 ^ in5;
+        out1 = tmp1 ^ in1 ^ in7;
+        out6 = tmp1 ^ out0 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A4(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in3;
+        tmp1 = in2 ^ in4;
+        tmp2 = in2 ^ in5;
+        tmp3 = in0 ^ in7;
+        out0 = tmp0 ^ in5;
+        out6 = tmp0 ^ in6 ^ in7;
+        out1 = tmp1 ^ in6;
+        out7 = tmp1 ^ tmp3;
+        out3 = tmp2 ^ in3;
+        tmp4 = tmp2 ^ out1;
+        out2 = tmp3 ^ in1;
+        out5 = tmp4 ^ out7;
+        out4 = tmp4 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A5(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in2 ^ in5;
+        tmp0 = in1 ^ in6;
+        tmp1 = in0 ^ in1;
+        tmp2 = in2 ^ in4;
+        out6 = in1 ^ in3 ^ in7;
+        out4 = tmp0 ^ in5;
+        out1 = tmp0 ^ tmp2;
+        out0 = tmp1 ^ in3 ^ in5;
+        out2 = tmp1 ^ in2 ^ in7;
+        out7 = tmp2 ^ in0;
+        out5 = tmp0 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A6(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in0;
+        out3 = in3 ^ in5 ^ in7;
+        out1 = in0 ^ in2 ^ in4 ^ in6;
+        out0 = out3 ^ in1;
+        out7 = out1 ^ in7;
+        out6 = out0 ^ in6;
+        out5 = out7 ^ in5;
+        out4 = out6 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A7(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in0 ^ in2;
+        out3 = in5 ^ in7;
+        out7 = out2 ^ in4 ^ in6;
+        out6 = out3 ^ in1 ^ in3;
+        out1 = out7 ^ in1;
+        out5 = out7 ^ in7;
+        out0 = out6 ^ in0;
+        out4 = out6 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A8(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in4;
+        tmp1 = in1 ^ in6;
+        tmp2 = in0 ^ in2 ^ in7;
+        out1 = tmp0 ^ in7;
+        out4 = tmp0 ^ in6;
+        out0 = tmp1 ^ in3;
+        out2 = tmp1 ^ in5;
+        out6 = tmp1 ^ in4;
+        out7 = tmp2 ^ in5;
+        out3 = tmp2 ^ out0 ^ in6;
+        out5 = out7 ^ in2 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_A9(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in2 ^ in6;
+        out6 = in1 ^ in4;
+        out7 = in0 ^ in2 ^ in5;
+        out5 = in0 ^ in3 ^ in7;
+        out2 = out4 ^ in1 ^ in5;
+        out1 = out6 ^ in2 ^ in7;
+        out0 = out2 ^ out7 ^ in3;
+        out3 = out1 ^ in0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_AA(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in2;
+        tmp1 = in1 ^ in3;
+        tmp2 = in6 ^ in7;
+        out1 = tmp0 ^ in4 ^ in7;
+        out3 = tmp1 ^ in0;
+        out0 = tmp1 ^ tmp2;
+        out2 = tmp2 ^ in5;
+        out7 = tmp0 ^ out2;
+        out6 = out1 ^ out7 ^ in1;
+        out5 = out0 ^ out6 ^ in0;
+        out4 = out5 ^ out7 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_AB(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in0 ^ in1;
+        tmp0 = in1 ^ in4;
+        tmp1 = in0 ^ in7;
+        out6 = tmp0 ^ in5;
+        out1 = tmp0 ^ tmp1 ^ in2;
+        out5 = tmp1 ^ in3 ^ in4;
+        out0 = tmp0 ^ out5 ^ in6;
+        out4 = out0 ^ out3 ^ in2;
+        out2 = out4 ^ in3 ^ in5;
+        out7 = tmp1 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_AC(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in1 ^ in3;
+        out1 = in2 ^ in4;
+        tmp0 = in0 ^ in2;
+        out4 = in4 ^ in7;
+        out5 = in0 ^ in5;
+        out6 = in1 ^ in6;
+        out7 = tmp0 ^ in7;
+        out3 = tmp0 ^ in3 ^ in6;
+        out2 = out5 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_AD(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in7;
+        out5 = in0;
+        out6 = in1;
+        out7 = in0 ^ in2;
+        out0 = in0 ^ in1 ^ in3;
+        out2 = out7 ^ in1 ^ in5;
+        out1 = in1 ^ in2 ^ in4;
+        out3 = out7 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_AE(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in3 ^ in4;
+        tmp0 = in0 ^ in4;
+        tmp1 = in0 ^ in7;
+        out0 = in1 ^ in3 ^ in7;
+        out1 = tmp0 ^ in2;
+        out5 = tmp0 ^ in5;
+        tmp2 = tmp1 ^ in6;
+        out2 = tmp1 ^ in5;
+        out3 = tmp2 ^ in3;
+        out7 = tmp2 ^ in2;
+        out6 = tmp2 ^ out2 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_AF(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in3;
+        tmp0 = in0 ^ in7;
+        out5 = in0 ^ in4;
+        out6 = in1 ^ in5;
+        out7 = in0 ^ in2 ^ in6;
+        out0 = tmp0 ^ in1 ^ in3;
+        out3 = tmp0 ^ in6;
+        out2 = tmp0 ^ in2 ^ in5;
+        out1 = out5 ^ in1 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B0(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in4;
+        tmp1 = in3 ^ in6;
+        out2 = tmp0 ^ in7;
+        tmp2 = tmp0 ^ tmp1;
+        out0 = tmp2 ^ in5;
+        out3 = tmp2 ^ in2;
+        out6 = out3 ^ in6;
+        tmp3 = out6 ^ in0 ^ in1;
+        out7 = tmp3 ^ in5;
+        out5 = tmp3 ^ out2;
+        out1 = out0 ^ out5 ^ in0;
+        out4 = tmp1 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B1(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in4;
+        out2 = tmp0 ^ in2 ^ in7;
+        tmp1 = out2 ^ in6;
+        out1 = tmp1 ^ in5;
+        out3 = tmp1 ^ in7;
+        out4 = tmp1 ^ in0;
+        out6 = out3 ^ in3;
+        out0 = out6 ^ in0 ^ in2 ^ in5;
+        out5 = tmp1 ^ out0 ^ in1;
+        out7 = tmp0 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B2(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in4;
+        tmp0 = in4 ^ in7;
+        tmp1 = in1 ^ in3 ^ in6;
+        out3 = tmp0 ^ tmp1;
+        tmp2 = tmp1 ^ in0;
+        out0 = out3 ^ in5;
+        out4 = tmp2 ^ in2;
+        tmp3 = out4 ^ in6;
+        out5 = tmp0 ^ tmp3;
+        out1 = tmp3 ^ out0;
+        tmp4 = out1 ^ in7;
+        out7 = tmp4 ^ in3;
+        out6 = tmp2 ^ tmp4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B3(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in2 ^ in4;
+        tmp0 = in0 ^ in5;
+        tmp1 = in1 ^ in6;
+        out3 = tmp1 ^ in4 ^ in7;
+        tmp2 = tmp0 ^ out3;
+        out0 = tmp2 ^ in3;
+        out1 = tmp2 ^ in2;
+        out5 = out0 ^ in2 ^ in6;
+        out7 = tmp1 ^ out5;
+        out4 = out7 ^ in1 ^ in5 ^ in7;
+        out6 = tmp0 ^ out4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B4(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in0 ^ in1;
+        out5 = out4 ^ in2;
+        tmp0 = out4 ^ in4;
+        out6 = out5 ^ in0 ^ in3;
+        out7 = tmp0 ^ out6;
+        out2 = tmp0 ^ in6 ^ in7;
+        out3 = out7 ^ in0 ^ in7;
+        out0 = out5 ^ out7 ^ in5;
+        out1 = out0 ^ out6 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B5(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        tmp1 = in2 ^ in4;
+        out4 = tmp0 ^ in4;
+        out3 = tmp1 ^ in7;
+        tmp2 = out4 ^ in5;
+        out7 = out3 ^ in0 ^ in3;
+        out0 = tmp2 ^ in3;
+        out2 = tmp0 ^ out3 ^ in6;
+        out5 = tmp1 ^ tmp2;
+        out6 = out2 ^ out7 ^ in2;
+        out1 = tmp0 ^ out0 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B6(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in3 ^ in4;
+        tmp0 = in1 ^ in2;
+        tmp1 = in0 ^ in4;
+        tmp2 = in3 ^ in5;
+        tmp3 = out3 ^ in1 ^ in7;
+        out5 = tmp0 ^ tmp1;
+        out6 = tmp0 ^ tmp2;
+        out2 = tmp1 ^ in6;
+        out4 = tmp1 ^ tmp3;
+        out0 = tmp3 ^ in5;
+        out1 = out2 ^ in2 ^ in5;
+        out7 = tmp2 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B7(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in4;
+        tmp0 = in0 ^ in4;
+        out2 = tmp0 ^ in2 ^ in6;
+        tmp1 = out2 ^ in7;
+        out1 = out2 ^ in1 ^ in5;
+        out7 = tmp1 ^ in3;
+        out5 = out1 ^ in6;
+        out6 = tmp0 ^ out1 ^ in3;
+        out0 = tmp1 ^ out6;
+        out4 = out0 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B8(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in4;
+        tmp1 = in2 ^ in5;
+        out2 = tmp0 ^ in5;
+        out4 = tmp1 ^ in0;
+        tmp2 = tmp1 ^ in7;
+        out6 = tmp2 ^ out2;
+        out7 = out4 ^ in3;
+        out1 = tmp2 ^ in4;
+        out3 = tmp0 ^ out7;
+        out0 = out3 ^ out4 ^ in6;
+        out5 = out0 ^ in0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_B9(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in2;
+        tmp1 = in4 ^ in5;
+        out4 = tmp0 ^ tmp1;
+        tmp2 = tmp0 ^ in3 ^ in7;
+        out3 = out4 ^ in1;
+        out7 = tmp2 ^ in5;
+        out2 = out3 ^ in0;
+        out1 = out2 ^ in7;
+        out6 = out1 ^ in5 ^ in6;
+        out0 = tmp2 ^ out6;
+        out5 = tmp1 ^ out0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_BA(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in5 ^ in7;
+        out2 = tmp0 ^ in4;
+        tmp1 = out2 ^ in2;
+        out1 = tmp1 ^ in0;
+        out6 = tmp1 ^ in1;
+        out4 = out1 ^ in3 ^ in4;
+        tmp2 = out4 ^ out6;
+        out7 = out4 ^ in6 ^ in7;
+        out5 = tmp2 ^ in6;
+        out3 = tmp0 ^ tmp2;
+        out0 = out6 ^ out7 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_BB(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in2 ^ in4 ^ in5 ^ in7;
+        tmp0 = out2 ^ in1;
+        out4 = out2 ^ in0 ^ in3;
+        out1 = tmp0 ^ in0;
+        out6 = tmp0 ^ in6;
+        out3 = out1 ^ in2;
+        tmp1 = out4 ^ out6 ^ in4;
+        out0 = tmp1 ^ in7;
+        out5 = tmp1 ^ in5;
+        out7 = tmp0 ^ tmp1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_BC(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in2;
+        tmp1 = in2 ^ in4;
+        out0 = in1 ^ in3 ^ in4;
+        out6 = in1 ^ in2 ^ in7;
+        out7 = tmp0 ^ in3;
+        out5 = tmp0 ^ out6 ^ in6;
+        out1 = tmp1 ^ in5;
+        tmp2 = out1 ^ out5 ^ in1;
+        out3 = tmp2 ^ in3;
+        out4 = tmp1 ^ tmp2;
+        out2 = tmp2 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_BD(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3;
+        tmp1 = in1 ^ in4;
+        out0 = tmp0 ^ tmp1;
+        out7 = tmp0 ^ in2 ^ in7;
+        out1 = tmp1 ^ in2 ^ in5;
+        tmp2 = out1 ^ in0;
+        out2 = tmp2 ^ in6;
+        out3 = out2 ^ in1 ^ in7;
+        out4 = out3 ^ in2;
+        out5 = tmp1 ^ out4;
+        out6 = tmp2 ^ out4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_BE(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3 ^ in6;
+        out4 = tmp0 ^ in5;
+        out7 = tmp0 ^ in2;
+        out3 = out4 ^ in4;
+        out1 = out3 ^ out7 ^ in0;
+        out2 = out3 ^ in3 ^ in7;
+        out0 = out2 ^ out4 ^ in1;
+        out5 = tmp0 ^ out0;
+        out6 = out1 ^ out5 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_BF(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in4;
+        out3 = tmp0 ^ in5 ^ in6;
+        out4 = out3 ^ in3;
+        tmp1 = out3 ^ in7;
+        out2 = tmp1 ^ in2;
+        out5 = tmp1 ^ in1;
+        tmp2 = out2 ^ in5;
+        out7 = tmp2 ^ in3 ^ in4;
+        tmp3 = tmp0 ^ out5;
+        out0 = tmp3 ^ out4;
+        out1 = tmp2 ^ tmp3;
+        out6 = tmp3 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C0(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out5 = in2 ^ in5;
+        tmp0 = in1 ^ in4;
+        tmp1 = in3 ^ in6;
+        out0 = out5 ^ in1;
+        out4 = tmp0 ^ in7;
+        out3 = tmp0 ^ tmp1;
+        out1 = tmp1 ^ in2;
+        out6 = tmp1 ^ in0;
+        out7 = out4 ^ in0;
+        out2 = out4 ^ out5 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C1(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out5 = in2;
+        tmp0 = in0 ^ in1;
+        out4 = in1 ^ in7;
+        out6 = in0 ^ in3;
+        out3 = in1 ^ in4 ^ in6;
+        tmp1 = tmp0 ^ in2;
+        out7 = tmp0 ^ in4;
+        out0 = tmp1 ^ in5;
+        out1 = tmp1 ^ out6 ^ in6;
+        out2 = out6 ^ out7 ^ in5 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C2(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in1 ^ in3 ^ in4;
+        tmp0 = in0 ^ in3 ^ in6;
+        out5 = in2 ^ in4 ^ in5;
+        tmp1 = out4 ^ in7;
+        out1 = tmp0 ^ in2;
+        out6 = tmp0 ^ in5;
+        out2 = out5 ^ in3;
+        out7 = tmp0 ^ tmp1;
+        out3 = tmp1 ^ in2 ^ in6;
+        out0 = tmp1 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C3(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in1 ^ in3;
+        tmp0 = in0 ^ in2;
+        tmp1 = in3 ^ in5;
+        out5 = in2 ^ in4;
+        tmp2 = tmp0 ^ out4;
+        out2 = tmp1 ^ in4;
+        out6 = tmp1 ^ in0;
+        out0 = tmp1 ^ tmp2 ^ in7;
+        out1 = tmp2 ^ in6;
+        out7 = out1 ^ out5 ^ in3;
+        out3 = tmp0 ^ out7 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C4(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in7;
+        out3 = tmp0 ^ in4;
+        tmp1 = tmp0 ^ in2;
+        out1 = tmp1 ^ in6;
+        out5 = tmp1 ^ in5;
+        out4 = out1 ^ out3 ^ in1;
+        out0 = out4 ^ in4 ^ in5;
+        out2 = out0 ^ out3 ^ in0;
+        out7 = out1 ^ out2 ^ in7;
+        out6 = tmp1 ^ out0 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C5(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in4 ^ in7;
+        tmp0 = in3 ^ in7;
+        out4 = in1 ^ in2 ^ in6;
+        out6 = in0 ^ in3 ^ in4;
+        out5 = tmp0 ^ in2;
+        out1 = tmp0 ^ out4;
+        out0 = out4 ^ in0 ^ in5;
+        out2 = out0 ^ out5 ^ in4;
+        out7 = tmp0 ^ out2 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C6(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in5 ^ in6;
+        tmp1 = in1 ^ in7;
+        tmp2 = tmp0 ^ in0;
+        tmp3 = tmp0 ^ tmp1;
+        tmp4 = tmp2 ^ in4;
+        out0 = tmp3 ^ in2;
+        out6 = tmp4 ^ in3;
+        out2 = out6 ^ in2;
+        out7 = tmp1 ^ tmp4;
+        out3 = tmp2 ^ out2;
+        tmp5 = out3 ^ in5;
+        out5 = tmp5 ^ in7;
+        out4 = tmp3 ^ tmp5;
+        out1 = tmp4 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C7(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in2 ^ in4;
+        tmp0 = in3 ^ in5;
+        tmp1 = out3 ^ in7;
+        out6 = tmp0 ^ in0 ^ in4;
+        out5 = tmp1 ^ in3;
+        out2 = out6 ^ in6;
+        out7 = out2 ^ in1 ^ in3;
+        out0 = tmp1 ^ out7;
+        out1 = tmp0 ^ out0;
+        out4 = out1 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C8(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out0 = in1 ^ in2;
+        out1 = in2 ^ in3;
+        tmp0 = in5 ^ in6;
+        tmp1 = in0 ^ in7;
+        out2 = out1 ^ in1 ^ in4;
+        out4 = tmp0 ^ in4;
+        out5 = tmp0 ^ in7;
+        out6 = tmp1 ^ in6;
+        out7 = tmp1 ^ in1;
+        out3 = out2 ^ in0 ^ in2 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_C9(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in5 ^ in6;
+        out7 = in0 ^ in1;
+        tmp0 = in1 ^ in3;
+        out5 = in6 ^ in7;
+        out6 = in0 ^ in7;
+        out0 = out7 ^ in2;
+        out3 = out7 ^ in4 ^ in5;
+        out1 = tmp0 ^ in2;
+        out2 = tmp0 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_CA(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in7;
+        tmp1 = in2 ^ in7;
+        tmp2 = tmp0 ^ in6;
+        out0 = tmp1 ^ in1;
+        tmp3 = tmp1 ^ in3;
+        out6 = tmp2 ^ in5;
+        out7 = tmp2 ^ in1;
+        out2 = tmp3 ^ in4;
+        out5 = out6 ^ in0 ^ in4;
+        out4 = out5 ^ in3;
+        out1 = tmp0 ^ tmp3;
+        out3 = tmp3 ^ out5 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_CB(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in4 ^ in7;
+        tmp1 = in5 ^ in7;
+        out7 = in0 ^ in1 ^ in6;
+        out5 = tmp0 ^ in6;
+        out2 = tmp0 ^ in3;
+        out6 = tmp1 ^ in0;
+        out4 = tmp1 ^ in3 ^ in6;
+        tmp2 = out5 ^ out7 ^ in2;
+        out1 = tmp2 ^ out2;
+        out0 = tmp2 ^ in4;
+        out3 = tmp2 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_CC(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in5;
+        tmp1 = in1 ^ in6;
+        out1 = in2 ^ in3 ^ in7;
+        out5 = tmp0 ^ in6;
+        out0 = tmp1 ^ in2;
+        tmp2 = out5 ^ in0 ^ in7;
+        out3 = tmp2 ^ in4;
+        out6 = tmp0 ^ out3;
+        out7 = tmp1 ^ tmp2 ^ in3;
+        tmp3 = out1 ^ out6;
+        out4 = tmp2 ^ tmp3;
+        out2 = tmp3 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_CD(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out5 = in3 ^ in6;
+        tmp0 = in0 ^ in1;
+        tmp1 = in2 ^ in7;
+        out6 = in0 ^ in4 ^ in7;
+        out2 = tmp0 ^ out5 ^ in4;
+        out7 = tmp0 ^ in5;
+        out0 = tmp0 ^ in2 ^ in6;
+        out4 = tmp1 ^ in5;
+        out1 = tmp1 ^ in1 ^ in3;
+        out3 = out6 ^ in5 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_CE(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in5;
+        tmp1 = tmp0 ^ in3;
+        out4 = tmp1 ^ in4;
+        tmp2 = out4 ^ in6;
+        out3 = tmp2 ^ in0;
+        out5 = tmp2 ^ in2;
+        out2 = out3 ^ in5 ^ in7;
+        out6 = tmp1 ^ out2;
+        out7 = out2 ^ out4 ^ in1;
+        out1 = tmp2 ^ out6;
+        out0 = tmp0 ^ out7 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_CF(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in6;
+        tmp1 = in0 ^ in1 ^ in5;
+        out4 = in2 ^ in3 ^ in5;
+        out5 = tmp0 ^ in4;
+        out7 = tmp1 ^ in6;
+        out1 = tmp1 ^ out4 ^ in7;
+        tmp2 = out5 ^ in0;
+        out2 = tmp2 ^ in7;
+        out3 = tmp2 ^ out4;
+        out6 = tmp0 ^ out2 ^ in5;
+        out0 = tmp0 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D0(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3;
+        tmp1 = in1 ^ in4;
+        tmp2 = in2 ^ in5;
+        out7 = tmp0 ^ tmp1;
+        out0 = tmp1 ^ tmp2;
+        tmp3 = tmp2 ^ in3;
+        out1 = tmp3 ^ in6;
+        tmp4 = out1 ^ in1;
+        out2 = tmp4 ^ in7;
+        out3 = out2 ^ in2;
+        out4 = tmp0 ^ out3;
+        out5 = tmp3 ^ out3;
+        out6 = tmp4 ^ out4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D1(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in5 ^ in6;
+        tmp1 = tmp0 ^ in1;
+        out1 = tmp1 ^ in2;
+        out2 = tmp1 ^ in7;
+        out3 = out2 ^ in3;
+        out5 = out3 ^ in2;
+        tmp2 = out3 ^ in0;
+        out4 = tmp2 ^ in4;
+        out7 = tmp0 ^ out4;
+        out6 = tmp2 ^ out1 ^ in6;
+        out0 = out2 ^ out6 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D2(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in5 ^ in6;
+        out2 = tmp0 ^ in2 ^ in3;
+        out1 = out2 ^ in0;
+        out3 = out2 ^ in1;
+        out4 = out1 ^ in1 ^ in2;
+        out6 = out1 ^ in6 ^ in7;
+        out7 = out4 ^ in4 ^ in5;
+        out5 = out4 ^ out6 ^ in4;
+        out0 = tmp0 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D3(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in3 ^ in5 ^ in6;
+        tmp0 = out2 ^ in2;
+        tmp1 = tmp0 ^ in1;
+        out1 = tmp1 ^ in0;
+        out3 = tmp1 ^ in3;
+        out4 = out1 ^ in2 ^ in4;
+        tmp2 = out4 ^ in5;
+        out7 = tmp2 ^ in7;
+        out0 = tmp0 ^ out7;
+        tmp3 = out0 ^ in0;
+        out5 = tmp3 ^ in6;
+        out6 = tmp2 ^ tmp3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D4(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in3 ^ in5;
+        tmp0 = in1 ^ in5;
+        tmp1 = tmp0 ^ in2;
+        out4 = tmp1 ^ in0;
+        tmp2 = tmp1 ^ in6;
+        out2 = out4 ^ in3 ^ in7;
+        out0 = tmp2 ^ in4;
+        out5 = tmp2 ^ out3;
+        out1 = tmp0 ^ out5 ^ in7;
+        out6 = tmp0 ^ out2 ^ in4;
+        out7 = tmp1 ^ out6 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D5(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in5;
+        tmp0 = in0 ^ in4;
+        tmp1 = tmp0 ^ in1 ^ in5;
+        out4 = tmp1 ^ in2;
+        out0 = out4 ^ in6;
+        tmp2 = tmp0 ^ out0;
+        out5 = tmp2 ^ in3;
+        out1 = out5 ^ in7;
+        out6 = tmp1 ^ out1;
+        out7 = tmp2 ^ out6;
+        out2 = out7 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D6(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in2 ^ in4 ^ in6;
+        out5 = tmp0 ^ in3;
+        out0 = tmp0 ^ in5 ^ in7;
+        out3 = out0 ^ out5 ^ in2;
+        tmp1 = out3 ^ in0;
+        out1 = tmp1 ^ in6;
+        out2 = tmp1 ^ in7;
+        out4 = tmp1 ^ in1;
+        out6 = tmp1 ^ in4;
+        out7 = tmp0 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D7(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in3;
+        out3 = in2 ^ in5 ^ in7;
+        out2 = tmp0 ^ in5;
+        tmp1 = tmp0 ^ out3 ^ in1;
+        out1 = tmp1 ^ in6;
+        out4 = tmp1 ^ in4;
+        tmp2 = out1 ^ in4;
+        out6 = tmp2 ^ in1;
+        out7 = tmp2 ^ in2;
+        out0 = tmp2 ^ in3;
+        out5 = tmp2 ^ in0 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D8(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in0;
+        out5 = in1;
+        tmp0 = in1 ^ in2;
+        out6 = in0 ^ in2;
+        out0 = tmp0 ^ in4;
+        tmp1 = tmp0 ^ in3;
+        out7 = tmp1 ^ out6;
+        out2 = tmp1 ^ in6;
+        out3 = out7 ^ in7;
+        out1 = tmp1 ^ in1 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_D9(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in0 ^ in4;
+        out5 = in1 ^ in5;
+        out2 = in1 ^ in3 ^ in6;
+        out3 = in0 ^ in1 ^ in7;
+        out6 = in0 ^ in2 ^ in6;
+        out0 = out4 ^ in1 ^ in2;
+        out1 = out5 ^ in2 ^ in3;
+        out7 = out3 ^ in3;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_DA(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out5 = in1 ^ in4;
+        tmp0 = in2 ^ in7;
+        tmp1 = in0 ^ in2 ^ in3;
+        out0 = tmp0 ^ out5;
+        out4 = tmp0 ^ tmp1;
+        out2 = tmp0 ^ in3 ^ in6;
+        out1 = tmp1 ^ in5;
+        out3 = tmp1 ^ in1;
+        out6 = out1 ^ in3;
+        out7 = out3 ^ in2 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_DB(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        tmp1 = in1 ^ in5;
+        tmp2 = in3 ^ in7;
+        out3 = tmp0 ^ in2;
+        out5 = tmp1 ^ in4;
+        out6 = tmp1 ^ out3 ^ in6;
+        out2 = tmp2 ^ in6;
+        tmp3 = tmp2 ^ in4;
+        tmp4 = out3 ^ in3;
+        out4 = tmp3 ^ in0;
+        out1 = tmp4 ^ in5;
+        out0 = tmp3 ^ tmp4;
+        out7 = tmp0 ^ out2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_DC(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in2;
+        tmp1 = in0 ^ in3;
+        out6 = tmp0 ^ in4;
+        tmp2 = tmp0 ^ in7;
+        out3 = tmp1 ^ in6;
+        tmp3 = tmp1 ^ in1;
+        out1 = tmp1 ^ tmp2 ^ in5;
+        out4 = tmp2 ^ in6;
+        out2 = tmp3 ^ in2;
+        out7 = tmp3 ^ in5;
+        out5 = tmp2 ^ out2;
+        out0 = out2 ^ out3 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_DD(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in0 ^ in6;
+        out2 = in0 ^ in1 ^ in3;
+        out6 = out3 ^ in2 ^ in4;
+        out7 = out2 ^ in5 ^ in7;
+        out0 = out6 ^ in1;
+        out4 = out6 ^ in7;
+        out5 = out7 ^ in0;
+        out1 = out5 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_DE(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in3 ^ in6;
+        tmp1 = in3 ^ in4 ^ in7;
+        out4 = tmp0 ^ in0;
+        out5 = tmp1 ^ in1;
+        out3 = out4 ^ in7;
+        out2 = out3 ^ in6;
+        out1 = out2 ^ in5;
+        out6 = tmp1 ^ out1;
+        out0 = tmp0 ^ out5;
+        out7 = out0 ^ out1 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_DF(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in0 ^ in3 ^ in7;
+        tmp0 = out2 ^ in1 ^ in5;
+        out1 = tmp0 ^ in2;
+        out7 = tmp0 ^ in6;
+        out5 = tmp0 ^ in0 ^ in4;
+        tmp1 = out1 ^ out5 ^ in6;
+        out4 = tmp1 ^ in3;
+        out6 = tmp1 ^ in5;
+        tmp2 = tmp1 ^ in7;
+        out0 = tmp2 ^ in1;
+        out3 = tmp2 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E0(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in1 ^ in7;
+        tmp0 = in2 ^ in4;
+        out4 = out3 ^ in3 ^ in5;
+        out2 = tmp0 ^ in1;
+        tmp1 = tmp0 ^ in6;
+        out0 = out4 ^ in2;
+        out6 = out4 ^ in0;
+        out1 = tmp1 ^ in3;
+        out5 = tmp1 ^ in0;
+        out7 = out5 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E1(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in1 ^ in4;
+        tmp0 = in1 ^ in7;
+        out3 = tmp0 ^ in3;
+        tmp1 = out3 ^ in5;
+        out4 = tmp1 ^ in4;
+        tmp2 = tmp1 ^ in0;
+        out0 = tmp2 ^ in2;
+        out6 = tmp2 ^ in6;
+        tmp3 = out0 ^ out4 ^ in6;
+        out5 = tmp3 ^ in5;
+        out7 = tmp0 ^ tmp3;
+        out1 = tmp2 ^ out5 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E2(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in1 ^ in2;
+        out4 = in1 ^ in5;
+        out2 = in2 ^ in4 ^ in7;
+        out5 = in0 ^ in2 ^ in6;
+        out0 = out3 ^ in3 ^ in5;
+        out7 = out3 ^ in0 ^ in4;
+        out6 = out2 ^ out7 ^ in3;
+        out1 = out5 ^ in3 ^ in4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E3(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in4 ^ in7;
+        tmp0 = in1 ^ in3;
+        out3 = tmp0 ^ in2;
+        tmp1 = out3 ^ in0;
+        out0 = tmp1 ^ in5;
+        tmp2 = tmp1 ^ in4;
+        out1 = tmp2 ^ in6;
+        tmp3 = tmp2 ^ in3;
+        out7 = tmp3 ^ in7;
+        out6 = out1 ^ out2 ^ in2;
+        tmp4 = tmp0 ^ out0;
+        out5 = tmp4 ^ in6;
+        out4 = tmp3 ^ tmp4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E4(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in6;
+        tmp0 = in0 ^ in4;
+        tmp1 = tmp0 ^ in2 ^ in6;
+        out2 = tmp1 ^ in1;
+        out7 = out2 ^ in5;
+        tmp2 = tmp0 ^ out7;
+        out4 = tmp2 ^ in3;
+        out0 = out4 ^ in7;
+        out6 = tmp1 ^ out0;
+        out5 = tmp2 ^ out6;
+        out1 = out5 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E5(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in3 ^ in6;
+        tmp0 = in0 ^ in1;
+        tmp1 = in5 ^ in7;
+        out2 = tmp0 ^ in4 ^ in6;
+        tmp2 = tmp1 ^ out2;
+        out6 = tmp2 ^ in3;
+        out7 = tmp2 ^ in2;
+        out0 = out6 ^ in2 ^ in4;
+        out5 = out6 ^ in1 ^ in2;
+        out1 = tmp0 ^ out5 ^ in5;
+        out4 = tmp1 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E6(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in2 ^ in6 ^ in7;
+        out2 = out3 ^ in0 ^ in4;
+        out4 = out3 ^ in1 ^ in5;
+        out1 = out2 ^ in3;
+        out7 = out2 ^ out4 ^ in2;
+        out0 = out4 ^ in3 ^ in7;
+        out5 = out1 ^ in4;
+        out6 = out0 ^ out2 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E7(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in3;
+        out3 = tmp0 ^ in6 ^ in7;
+        tmp1 = out3 ^ in0;
+        out5 = tmp1 ^ in5;
+        tmp2 = tmp1 ^ in4;
+        tmp3 = out5 ^ in7;
+        out1 = tmp2 ^ in1;
+        out0 = tmp3 ^ in1;
+        out6 = out1 ^ in2;
+        out2 = tmp0 ^ tmp2;
+        tmp4 = tmp3 ^ out6;
+        out4 = tmp4 ^ in6;
+        out7 = tmp4 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E8(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in3 ^ in6;
+        tmp0 = in4 ^ in7;
+        out1 = in2 ^ in3 ^ in4;
+        out5 = tmp0 ^ in0;
+        tmp1 = tmp0 ^ in1;
+        tmp2 = tmp1 ^ in5;
+        out0 = tmp1 ^ out1;
+        out2 = tmp2 ^ in2;
+        out6 = tmp2 ^ out5;
+        tmp3 = out6 ^ in6;
+        out3 = tmp3 ^ in7;
+        out7 = tmp3 ^ in2 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_E9(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        tmp1 = in3 ^ in6;
+        tmp2 = tmp0 ^ in6;
+        out4 = tmp1 ^ in4;
+        out6 = tmp2 ^ in5;
+        out7 = tmp2 ^ in2 ^ in7;
+        out3 = out6 ^ in3 ^ in7;
+        out0 = tmp1 ^ out7;
+        out2 = out3 ^ out4 ^ in0;
+        out5 = tmp0 ^ out2;
+        out1 = out0 ^ out5 ^ in5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_EA(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in6 ^ in7;
+        out5 = in0 ^ in7;
+        out6 = in0 ^ in1;
+        out0 = in1 ^ in2 ^ in3;
+        out2 = in2 ^ in4 ^ in5;
+        out7 = out6 ^ in2;
+        out1 = out0 ^ out6 ^ in4;
+        out3 = out7 ^ in5 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_EB(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in4 ^ in5;
+        tmp0 = in0 ^ in1;
+        out4 = in4 ^ in6 ^ in7;
+        out5 = in0 ^ in5 ^ in7;
+        out6 = tmp0 ^ in6;
+        tmp1 = tmp0 ^ in2;
+        out0 = tmp1 ^ in3;
+        out7 = tmp1 ^ in7;
+        out1 = out0 ^ in4;
+        out3 = out0 ^ in5 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_EC(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out3 = in0 ^ in5;
+        out4 = in2 ^ in3 ^ in7;
+        out5 = in0 ^ in3 ^ in4;
+        out6 = out3 ^ in1 ^ in4;
+        out1 = out4 ^ in4;
+        out0 = out4 ^ in1 ^ in6;
+        out2 = out0 ^ out5 ^ in5;
+        out7 = out2 ^ in4 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_ED(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in2 ^ in4;
+        tmp1 = in3 ^ in5;
+        out4 = tmp0 ^ in3 ^ in7;
+        out3 = tmp1 ^ in0;
+        out1 = out4 ^ in1;
+        out5 = out3 ^ in4;
+        out7 = out1 ^ out5 ^ in6;
+        out2 = tmp0 ^ out7;
+        out0 = tmp1 ^ out7;
+        out6 = out2 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_EE(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in2;
+        tmp0 = in0 ^ in1;
+        out5 = in0 ^ in3;
+        tmp1 = tmp0 ^ in2;
+        out6 = tmp0 ^ in4;
+        tmp2 = tmp1 ^ out5;
+        out7 = tmp1 ^ in5;
+        out1 = tmp2 ^ out6 ^ in7;
+        out0 = tmp2 ^ in6;
+        tmp3 = out7 ^ in1;
+        out3 = tmp3 ^ in7;
+        out2 = tmp3 ^ in4 ^ in6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_EF(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out4 = in2 ^ in4;
+        tmp0 = in0 ^ in5;
+        tmp1 = in4 ^ in6;
+        out5 = tmp0 ^ in3;
+        out2 = tmp0 ^ tmp1;
+        out6 = tmp1 ^ in0 ^ in1;
+        out3 = out5 ^ in2 ^ in7;
+        out7 = out3 ^ in1 ^ in3;
+        out0 = out4 ^ out6 ^ in3;
+        out1 = tmp1 ^ out0 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F0(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in2;
+        tmp1 = in4 ^ in5;
+        out2 = tmp0 ^ in6;
+        out3 = tmp1 ^ in1;
+        tmp2 = tmp1 ^ in7;
+        out1 = out2 ^ out3 ^ in3;
+        tmp3 = tmp0 ^ tmp2;
+        out0 = tmp3 ^ in3;
+        out5 = tmp3 ^ in0;
+        out4 = out1 ^ out5 ^ in4;
+        out7 = out4 ^ in2;
+        out6 = tmp2 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F1(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in1 ^ in6;
+        tmp0 = in3 ^ in5;
+        out3 = tmp0 ^ in1 ^ in4;
+        tmp1 = out3 ^ in2;
+        out1 = tmp1 ^ in6;
+        tmp2 = tmp1 ^ in0;
+        tmp3 = out1 ^ in5;
+        out0 = tmp2 ^ in7;
+        out6 = tmp2 ^ in4;
+        out7 = tmp3 ^ in0;
+        out5 = tmp0 ^ out0;
+        out4 = tmp3 ^ out5 ^ in1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F2(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in4 ^ in5;
+        out2 = in2 ^ in6 ^ in7;
+        tmp1 = tmp0 ^ in1;
+        tmp2 = tmp1 ^ in2;
+        out0 = tmp2 ^ in3;
+        out3 = tmp2 ^ in7;
+        out5 = out3 ^ in0 ^ in4;
+        tmp3 = tmp0 ^ out5;
+        out7 = tmp3 ^ in3;
+        out4 = tmp3 ^ out2;
+        out1 = out0 ^ out4 ^ in4;
+        out6 = tmp1 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F3(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in6 ^ in7;
+        tmp0 = in0 ^ in1;
+        out4 = tmp0 ^ in6;
+        tmp1 = tmp0 ^ in2;
+        out5 = tmp1 ^ in7;
+        out6 = tmp1 ^ in3;
+        out7 = out6 ^ in4;
+        out0 = out7 ^ in5;
+        out1 = out0 ^ in6;
+        out3 = out0 ^ in0 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F4(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in0 ^ in1 ^ in2;
+        tmp0 = out2 ^ in3;
+        out4 = tmp0 ^ in4;
+        out5 = out4 ^ in5;
+        out6 = out5 ^ in6;
+        out7 = out6 ^ in7;
+        out0 = out7 ^ in0;
+        out1 = out0 ^ in1;
+        out3 = tmp0 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F5(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in0 ^ in1;
+        tmp0 = out2 ^ in2;
+        out4 = tmp0 ^ in3;
+        out5 = out4 ^ in4;
+        out6 = out5 ^ in5;
+        out7 = out6 ^ in6;
+        out0 = out7 ^ in7;
+        out1 = out0 ^ in0;
+        out3 = tmp0 ^ out0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F6(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in7;
+        out2 = tmp0 ^ in2;
+        out4 = out2 ^ in1 ^ in4;
+        out7 = out4 ^ in3 ^ in5;
+        out5 = out7 ^ in4 ^ in7;
+        out0 = tmp0 ^ out7 ^ in6;
+        tmp1 = out0 ^ in1;
+        out6 = out0 ^ in0 ^ in5;
+        out3 = tmp1 ^ in3;
+        out1 = tmp0 ^ tmp1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F7(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in0 ^ in7;
+        tmp0 = out2 ^ in1;
+        out4 = tmp0 ^ in2;
+        out5 = out4 ^ in3 ^ in7;
+        out6 = out5 ^ in4;
+        out7 = out6 ^ in5;
+        out0 = out7 ^ in6;
+        out1 = out0 ^ in7;
+        out3 = tmp0 ^ out1;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F8(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in4;
+        tmp1 = in3 ^ in5;
+        tmp2 = tmp0 ^ in6;
+        out4 = tmp0 ^ tmp1;
+        out1 = tmp1 ^ in2 ^ in4;
+        out3 = tmp2 ^ in1;
+        out5 = out3 ^ in5;
+        out7 = out1 ^ out5 ^ in7;
+        out6 = tmp1 ^ out7;
+        out0 = tmp2 ^ out7;
+        out2 = out6 ^ in0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_F9(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in3 ^ in5;
+        tmp1 = in0 ^ in6;
+        out4 = tmp0 ^ in0;
+        tmp2 = tmp1 ^ in4;
+        tmp3 = tmp1 ^ in2;
+        out5 = tmp2 ^ in1;
+        out3 = out5 ^ in3;
+        tmp4 = tmp3 ^ out3;
+        out1 = tmp4 ^ in5;
+        out0 = tmp4 ^ in0 ^ in7;
+        out6 = tmp0 ^ out0 ^ in4;
+        out7 = tmp2 ^ tmp4;
+        out2 = tmp3 ^ out6;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_FA(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in1;
+        tmp1 = tmp0 ^ in2;
+        tmp2 = tmp0 ^ in5;
+        tmp3 = tmp1 ^ in7;
+        out5 = tmp2 ^ in6;
+        out6 = tmp3 ^ in6;
+        out7 = tmp3 ^ in3;
+        out3 = out6 ^ in4;
+        out2 = tmp1 ^ out5;
+        out4 = out2 ^ out3 ^ in1;
+        out0 = out4 ^ out7 ^ in5;
+        out1 = tmp2 ^ out0;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_FB(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in5 ^ in6;
+        tmp0 = in0 ^ in1;
+        out4 = in0 ^ in5 ^ in7;
+        out5 = tmp0 ^ in6;
+        tmp1 = tmp0 ^ in2;
+        out6 = tmp1 ^ in7;
+        out7 = tmp1 ^ in3;
+        out0 = out7 ^ in4;
+        out1 = out0 ^ in5;
+        out3 = out0 ^ in6 ^ in7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_FC(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in1 ^ in2;
+        tmp1 = in0 ^ in7;
+        out2 = tmp0 ^ tmp1 ^ in5;
+        out3 = tmp1 ^ in4;
+        tmp2 = out2 ^ in6;
+        out6 = tmp2 ^ in4;
+        out7 = tmp2 ^ in3;
+        out4 = out6 ^ in1 ^ in3;
+        tmp3 = out4 ^ in0;
+        out1 = tmp3 ^ in6;
+        out0 = tmp3 ^ in1 ^ in5;
+        out5 = tmp0 ^ out4;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_FD(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in5;
+        tmp1 = in1 ^ in7;
+        out2 = tmp0 ^ tmp1;
+        out6 = out2 ^ in2 ^ in4;
+        tmp2 = out6 ^ in0;
+        out1 = tmp2 ^ in3;
+        out0 = tmp0 ^ out1 ^ in6;
+        out5 = out0 ^ in2;
+        tmp3 = out5 ^ in1;
+        out3 = tmp3 ^ in6;
+        out7 = tmp2 ^ tmp3;
+        out4 = tmp1 ^ out7;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_FE(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        tmp0 = in0 ^ in2;
+        out2 = tmp0 ^ in5;
+        out3 = tmp0 ^ in4;
+        tmp1 = out3 ^ in6;
+        out4 = tmp1 ^ in5;
+        tmp2 = tmp1 ^ in1;
+        out6 = tmp2 ^ in7;
+        tmp3 = tmp2 ^ in0;
+        out0 = tmp3 ^ in3;
+        tmp4 = out0 ^ out4 ^ in7;
+        out5 = tmp4 ^ in6;
+        out7 = tmp4 ^ in2;
+        out1 = tmp3 ^ out5;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void gf8_muladd_FF(void *out, void *in)
+{
+    unsigned int i;
+    uint64_t *in_ptr = (uint64_t *)in;
+    uint64_t *out_ptr = (uint64_t *)out;
+
+    for (i = 0; i < WIDTH; i++) {
+        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
+        uint64_t tmp0, tmp1, tmp2, tmp3;
+
+        uint64_t in0 = out_ptr[0];
+        uint64_t in1 = out_ptr[WIDTH];
+        uint64_t in2 = out_ptr[WIDTH * 2];
+        uint64_t in3 = out_ptr[WIDTH * 3];
+        uint64_t in4 = out_ptr[WIDTH * 4];
+        uint64_t in5 = out_ptr[WIDTH * 5];
+        uint64_t in6 = out_ptr[WIDTH * 6];
+        uint64_t in7 = out_ptr[WIDTH * 7];
+
+        out2 = in0 ^ in5;
+        tmp0 = in4 ^ in7;
+        tmp1 = out2 ^ in2;
+        out4 = tmp1 ^ in6;
+        out7 = tmp1 ^ in1 ^ in3;
+        out1 = tmp0 ^ out7;
+        tmp2 = out1 ^ in5;
+        out6 = tmp2 ^ in3;
+        tmp3 = tmp2 ^ in7;
+        out0 = tmp3 ^ in6;
+        out3 = tmp3 ^ in1;
+        out5 = tmp0 ^ out0 ^ in2;
+
+        out_ptr[0] = out0 ^ in_ptr[0];
+        out_ptr[WIDTH] = out1 ^ in_ptr[WIDTH];
+        out_ptr[WIDTH * 2] = out2 ^ in_ptr[WIDTH * 2];
+        out_ptr[WIDTH * 3] = out3 ^ in_ptr[WIDTH * 3];
+        out_ptr[WIDTH * 4] = out4 ^ in_ptr[WIDTH * 4];
+        out_ptr[WIDTH * 5] = out5 ^ in_ptr[WIDTH * 5];
+        out_ptr[WIDTH * 6] = out6 ^ in_ptr[WIDTH * 6];
+        out_ptr[WIDTH * 7] = out7 ^ in_ptr[WIDTH * 7];
+
+        in_ptr++;
+        out_ptr++;
+    }
+}
+
+static void (*gf8_muladd[])(void *out, void *in) = {
+    gf8_muladd_00, gf8_muladd_01, gf8_muladd_02, gf8_muladd_03,
+    gf8_muladd_04, gf8_muladd_05, gf8_muladd_06, gf8_muladd_07,
+    gf8_muladd_08, gf8_muladd_09, gf8_muladd_0A, gf8_muladd_0B,
+    gf8_muladd_0C, gf8_muladd_0D, gf8_muladd_0E, gf8_muladd_0F,
+    gf8_muladd_10, gf8_muladd_11, gf8_muladd_12, gf8_muladd_13,
+    gf8_muladd_14, gf8_muladd_15, gf8_muladd_16, gf8_muladd_17,
+    gf8_muladd_18, gf8_muladd_19, gf8_muladd_1A, gf8_muladd_1B,
+    gf8_muladd_1C, gf8_muladd_1D, gf8_muladd_1E, gf8_muladd_1F,
+    gf8_muladd_20, gf8_muladd_21, gf8_muladd_22, gf8_muladd_23,
+    gf8_muladd_24, gf8_muladd_25, gf8_muladd_26, gf8_muladd_27,
+    gf8_muladd_28, gf8_muladd_29, gf8_muladd_2A, gf8_muladd_2B,
+    gf8_muladd_2C, gf8_muladd_2D, gf8_muladd_2E, gf8_muladd_2F,
+    gf8_muladd_30, gf8_muladd_31, gf8_muladd_32, gf8_muladd_33,
+    gf8_muladd_34, gf8_muladd_35, gf8_muladd_36, gf8_muladd_37,
+    gf8_muladd_38, gf8_muladd_39, gf8_muladd_3A, gf8_muladd_3B,
+    gf8_muladd_3C, gf8_muladd_3D, gf8_muladd_3E, gf8_muladd_3F,
+    gf8_muladd_40, gf8_muladd_41, gf8_muladd_42, gf8_muladd_43,
+    gf8_muladd_44, gf8_muladd_45, gf8_muladd_46, gf8_muladd_47,
+    gf8_muladd_48, gf8_muladd_49, gf8_muladd_4A, gf8_muladd_4B,
+    gf8_muladd_4C, gf8_muladd_4D, gf8_muladd_4E, gf8_muladd_4F,
+    gf8_muladd_50, gf8_muladd_51, gf8_muladd_52, gf8_muladd_53,
+    gf8_muladd_54, gf8_muladd_55, gf8_muladd_56, gf8_muladd_57,
+    gf8_muladd_58, gf8_muladd_59, gf8_muladd_5A, gf8_muladd_5B,
+    gf8_muladd_5C, gf8_muladd_5D, gf8_muladd_5E, gf8_muladd_5F,
+    gf8_muladd_60, gf8_muladd_61, gf8_muladd_62, gf8_muladd_63,
+    gf8_muladd_64, gf8_muladd_65, gf8_muladd_66, gf8_muladd_67,
+    gf8_muladd_68, gf8_muladd_69, gf8_muladd_6A, gf8_muladd_6B,
+    gf8_muladd_6C, gf8_muladd_6D, gf8_muladd_6E, gf8_muladd_6F,
+    gf8_muladd_70, gf8_muladd_71, gf8_muladd_72, gf8_muladd_73,
+    gf8_muladd_74, gf8_muladd_75, gf8_muladd_76, gf8_muladd_77,
+    gf8_muladd_78, gf8_muladd_79, gf8_muladd_7A, gf8_muladd_7B,
+    gf8_muladd_7C, gf8_muladd_7D, gf8_muladd_7E, gf8_muladd_7F,
+    gf8_muladd_80, gf8_muladd_81, gf8_muladd_82, gf8_muladd_83,
+    gf8_muladd_84, gf8_muladd_85, gf8_muladd_86, gf8_muladd_87,
+    gf8_muladd_88, gf8_muladd_89, gf8_muladd_8A, gf8_muladd_8B,
+    gf8_muladd_8C, gf8_muladd_8D, gf8_muladd_8E, gf8_muladd_8F,
+    gf8_muladd_90, gf8_muladd_91, gf8_muladd_92, gf8_muladd_93,
+    gf8_muladd_94, gf8_muladd_95, gf8_muladd_96, gf8_muladd_97,
+    gf8_muladd_98, gf8_muladd_99, gf8_muladd_9A, gf8_muladd_9B,
+    gf8_muladd_9C, gf8_muladd_9D, gf8_muladd_9E, gf8_muladd_9F,
+    gf8_muladd_A0, gf8_muladd_A1, gf8_muladd_A2, gf8_muladd_A3,
+    gf8_muladd_A4, gf8_muladd_A5, gf8_muladd_A6, gf8_muladd_A7,
+    gf8_muladd_A8, gf8_muladd_A9, gf8_muladd_AA, gf8_muladd_AB,
+    gf8_muladd_AC, gf8_muladd_AD, gf8_muladd_AE, gf8_muladd_AF,
+    gf8_muladd_B0, gf8_muladd_B1, gf8_muladd_B2, gf8_muladd_B3,
+    gf8_muladd_B4, gf8_muladd_B5, gf8_muladd_B6, gf8_muladd_B7,
+    gf8_muladd_B8, gf8_muladd_B9, gf8_muladd_BA, gf8_muladd_BB,
+    gf8_muladd_BC, gf8_muladd_BD, gf8_muladd_BE, gf8_muladd_BF,
+    gf8_muladd_C0, gf8_muladd_C1, gf8_muladd_C2, gf8_muladd_C3,
+    gf8_muladd_C4, gf8_muladd_C5, gf8_muladd_C6, gf8_muladd_C7,
+    gf8_muladd_C8, gf8_muladd_C9, gf8_muladd_CA, gf8_muladd_CB,
+    gf8_muladd_CC, gf8_muladd_CD, gf8_muladd_CE, gf8_muladd_CF,
+    gf8_muladd_D0, gf8_muladd_D1, gf8_muladd_D2, gf8_muladd_D3,
+    gf8_muladd_D4, gf8_muladd_D5, gf8_muladd_D6, gf8_muladd_D7,
+    gf8_muladd_D8, gf8_muladd_D9, gf8_muladd_DA, gf8_muladd_DB,
+    gf8_muladd_DC, gf8_muladd_DD, gf8_muladd_DE, gf8_muladd_DF,
+    gf8_muladd_E0, gf8_muladd_E1, gf8_muladd_E2, gf8_muladd_E3,
+    gf8_muladd_E4, gf8_muladd_E5, gf8_muladd_E6, gf8_muladd_E7,
+    gf8_muladd_E8, gf8_muladd_E9, gf8_muladd_EA, gf8_muladd_EB,
+    gf8_muladd_EC, gf8_muladd_ED, gf8_muladd_EE, gf8_muladd_EF,
+    gf8_muladd_F0, gf8_muladd_F1, gf8_muladd_F2, gf8_muladd_F3,
+    gf8_muladd_F4, gf8_muladd_F5, gf8_muladd_F6, gf8_muladd_F7,
+    gf8_muladd_F8, gf8_muladd_F9, gf8_muladd_FA, gf8_muladd_FB,
+    gf8_muladd_FC, gf8_muladd_FD, gf8_muladd_FE, gf8_muladd_FF
+};
+
+static uint64_t zero[EC_METHOD_WORD_SIZE * 8] = {0, };
+
+void ec_code_c_prepare(ec_gf_t *gf, uint32_t *values, uint32_t count)
+{
+    uint32_t i, last, tmp;
+
+    last = 1;
+    for (i = count; i > 0; i--) {
+        if (values[i - 1] != 0) {
+            tmp = values[i - 1];
+            values[i - 1] = ec_gf_div(gf, tmp, last);
+            last = tmp;
+        }
+    }
+}
+
+void ec_code_c_linear(void *dst, void *src, uint64_t offset, uint32_t *values,
+                      uint32_t count)
+{
+    src += offset;
+    gf8_muladd_00(dst, src);
+    while (--count > 0) {
+        src += EC_METHOD_CHUNK_SIZE;
+        gf8_muladd[*values](dst, src);
+        values++;
+    }
+}
+
+void ec_code_c_interleaved(void *dst, void **src, uint64_t offset,
+                           uint32_t *values, uint32_t count)
+{
+    uint32_t i, last, tmp;
+
+    i = 0;
+    while ((last = *values++) == 0) {
+        i++;
+    }
+    gf8_muladd_00(dst, src[i++] + offset);
+    while (i < count) {
+        tmp = *values++;
+        if (tmp != 0) {
+            gf8_muladd[last](dst, src[i] + offset);
+            last = tmp;
+        }
+        i++;
+    }
+    gf8_muladd[last](dst, zero);
+}
diff --git a/xlators/cluster/ec/src/ec-code-c.h b/xlators/cluster/ec/src/ec-code-c.h
new file mode 100644
index 00000000000..92e8070e514
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-c.h
@@ -0,0 +1,24 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#ifndef __EC_CODE_C_H__
+#define __EC_CODE_C_H__
+
+#include "ec-types.h"
+
+void ec_code_c_prepare(ec_gf_t *gf, uint32_t *values, uint32_t count);
+
+void ec_code_c_linear(void *dst, void *src, uint64_t offset, uint32_t *values,
+                      uint32_t count);
+
+void ec_code_c_interleaved(void *dst, void **src, uint64_t offset,
+                           uint32_t *values, uint32_t count);
+
+#endif /* __EC_CODE_C_H__ */
diff --git a/xlators/cluster/ec/src/ec-code-intel.c b/xlators/cluster/ec/src/ec-code-intel.c
new file mode 100644
index 00000000000..b9fdcad4421
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-intel.c
@@ -0,0 +1,600 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#include <inttypes.h>
+#include <string.h>
+#include <errno.h>
+
+#include "ec-code-intel.h"
+
+static void
+ec_code_intel_init(ec_code_intel_t *intel)
+{
+    memset(intel, 0, sizeof(ec_code_intel_t));
+}
+
+static void
+ec_code_intel_prefix(ec_code_intel_t *intel, uint8_t prefix)
+{
+    intel->prefix.data[intel->prefix.bytes++] = prefix;
+}
+
+static void
+ec_code_intel_rex(ec_code_intel_t *intel, gf_boolean_t w)
+{
+    gf_boolean_t present = _gf_false;
+
+    if (w) {
+        intel->rex.w = 1;
+        present = _gf_true;
+    }
+    if (intel->modrm.present) {
+        if (intel->modrm.reg > 7) {
+            intel->modrm.reg &= 7;
+            intel->rex.r = 1;
+            present = _gf_true;
+        }
+        if (intel->sib.present) {
+            if (intel->sib.index > 7) {
+                intel->sib.index &= 7;
+                intel->rex.x = 1;
+                present = _gf_true;
+            }
+            if (intel->sib.base > 7) {
+                intel->sib.base &= 7;
+                intel->rex.b = 1;
+                present = _gf_true;
+            }
+        } else if (intel->modrm.rm > 7) {
+            intel->modrm.rm &= 7;
+            intel->rex.b = 1;
+            present = _gf_true;
+        }
+    } else if (intel->reg > 7) {
+        intel->reg &= 7;
+        intel->rex.b = 1;
+        present = _gf_true;
+    }
+    intel->rex.present = present;
+}
+
+static void
+ec_code_intel_vex(ec_code_intel_t *intel, gf_boolean_t w, gf_boolean_t l,
+                  ec_code_vex_opcode_t opcode, ec_code_vex_prefix_t prefix,
+                  uint32_t reg)
+{
+    ec_code_intel_rex(intel, w);
+    if (((intel->rex.w == 1) ||
+         (intel->rex.x == 0) ||
+         (intel->rex.b == 0)) ||
+        ((opcode != VEX_OPCODE_NONE) && (opcode != VEX_OPCODE_0F))) {
+        intel->rex.present = _gf_false;
+
+        intel->vex.bytes = 3;
+        intel->vex.data[0] = 0xC4;
+        intel->vex.data[1] = ((intel->rex.r << 7) | (intel->rex.x << 6) |
+                              (intel->rex.b << 5) | opcode) ^ 0xE0;
+        intel->vex.data[2] = (intel->rex.w << 7) | ((~reg & 0x0F) << 3) |
+                             (l ? 0x04 : 0x00) | prefix;
+    } else {
+        intel->vex.bytes = 2;
+        intel->vex.data[0] = 0xC5;
+        intel->vex.data[1] = (intel->rex.r << 7) | ((~reg & 0x0F) << 3) |
+                             (l ? 0x04 : 0x00) | prefix;
+    }
+}
+
+static void
+ec_code_intel_modrm_reg(ec_code_intel_t *intel, uint32_t rm, uint32_t reg)
+{
+    intel->modrm.present = _gf_true;
+    intel->modrm.mod = 3;
+    intel->modrm.rm = rm;
+    intel->modrm.reg = reg;
+}
+
+static void
+ec_code_intel_modrm_mem(ec_code_intel_t *intel, uint32_t reg,
+                        ec_code_intel_reg_t base, ec_code_intel_reg_t index,
+                        uint32_t scale, int32_t offset)
+{
+    if (index == REG_SP) {
+        intel->invalid = _gf_true;
+        return;
+    }
+    if ((index != REG_NULL) && (scale != 1) && (scale != 2) && (scale != 4) &&
+        (scale != 8)) {
+        intel->invalid = _gf_true;
+        return;
+    }
+    scale >>= 1;
+    if (scale == 4) {
+        scale = 3;
+    }
+
+    intel->modrm.present = _gf_true;
+    intel->modrm.reg = reg;
+
+    intel->offset.value = offset;
+    if ((offset == 0) && (base != REG_BP)) {
+        intel->modrm.mod = 0;
+        intel->offset.bytes = 0;
+    } else if ((offset >= -128) && (offset <= 127)) {
+        intel->modrm.mod = 1;
+        intel->offset.bytes = 1;
+    } else {
+        intel->modrm.mod = 2;
+        intel->offset.bytes = 4;
+    }
+
+    intel->modrm.rm = base;
+    if ((index != REG_NULL) || (base == REG_SP)) {
+        intel->modrm.rm = 4;
+        intel->sib.present = _gf_true;
+        intel->sib.index = index;
+        if (index == REG_NULL) {
+            intel->sib.index = 4;
+        }
+        intel->sib.scale = scale;
+        intel->sib.base = base;
+        if (base == REG_NULL) {
+            intel->sib.base = 5;
+            intel->modrm.mod = 0;
+            intel->offset.bytes = 4;
+        }
+    } else if (base == REG_NULL) {
+        intel->modrm.mod = 0;
+        intel->modrm.rm = 5;
+        intel->offset.bytes = 4;
+    }
+}
+
+static void
+ec_code_intel_op_1(ec_code_intel_t *intel, uint8_t opcode, uint32_t reg)
+{
+    intel->reg = reg;
+    intel->opcode.bytes = 1;
+    intel->opcode.data[0] = opcode;
+}
+
+static void
+ec_code_intel_op_2(ec_code_intel_t *intel, uint8_t opcode1, uint8_t opcode2,
+                   uint32_t reg)
+{
+    intel->reg = reg;
+    intel->opcode.bytes = 2;
+    intel->opcode.data[0] = opcode1;
+    intel->opcode.data[1] = opcode2;
+}
+
+static void
+ec_code_intel_immediate_1(ec_code_intel_t *intel, uint32_t value)
+{
+    intel->immediate.bytes = 1;
+    intel->immediate.value = value;
+}
+
+static void
+ec_code_intel_immediate_2(ec_code_intel_t *intel, uint32_t value)
+{
+    intel->immediate.bytes = 2;
+    intel->immediate.value = value;
+}
+
+static void
+ec_code_intel_immediate_4(ec_code_intel_t *intel, uint32_t value)
+{
+    intel->immediate.bytes = 4;
+    intel->immediate.value = value;
+}
+
+static void
+ec_code_intel_emit(ec_code_builder_t *builder, ec_code_intel_t *intel)
+{
+    uint8_t insn[15];
+    uint32_t i, count;
+
+    if (intel->invalid) {
+        ec_code_error(builder, EINVAL);
+        return;
+    }
+
+    count = 0;
+    for (i = 0; i < intel->prefix.bytes; i++) {
+        insn[count++] = intel->prefix.data[i];
+    }
+    for (i = 0; i < intel->vex.bytes; i++) {
+        insn[count++] = intel->vex.data[i];
+    }
+    if (intel->rex.present) {
+        insn[count++] = 0x40 |
+                        (intel->rex.w << 3) |
+                        (intel->rex.r << 2) |
+                        (intel->rex.x << 1) |
+                        (intel->rex.b << 0);
+    }
+    for (i = 0; i < intel->opcode.bytes; i++) {
+        insn[count++] = intel->opcode.data[i];
+    }
+    if (intel->modrm.present) {
+        insn[count++] = (intel->modrm.mod << 6) |
+                        (intel->modrm.reg << 3) |
+                        (intel->modrm.rm << 0);
+        if (intel->sib.present) {
+            insn[count++] = (intel->sib.scale << 6) |
+                            (intel->sib.index << 3) |
+                            (intel->sib.base << 0);
+        }
+    }
+    for (i = 0; i < intel->offset.bytes; i++) {
+        insn[count++] = intel->offset.data[i];
+    }
+    for (i = 0; i < intel->immediate.bytes; i++) {
+        insn[count++] = intel->immediate.data[i];
+    }
+
+    ec_code_emit(builder, insn, count);
+}
+
+void
+ec_code_intel_op_push_r(ec_code_builder_t *builder, ec_code_intel_reg_t reg)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_op_1(&intel, 0x50 | (reg & 7), reg);
+    ec_code_intel_rex(&intel, _gf_false);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_pop_r(ec_code_builder_t *builder, ec_code_intel_reg_t reg)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_op_1(&intel, 0x58 | (reg & 7), reg);
+    ec_code_intel_rex(&intel, _gf_false);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_ret(ec_code_builder_t *builder, uint32_t size)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    if (size == 0) {
+        ec_code_intel_op_1(&intel, 0xC3, 0);
+    } else {
+        ec_code_intel_immediate_2(&intel, size);
+        ec_code_intel_op_1(&intel, 0xC2, 0);
+    }
+    ec_code_intel_rex(&intel, _gf_false);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_mov_r2r(ec_code_builder_t *builder, ec_code_intel_reg_t src,
+                         ec_code_intel_reg_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_reg(&intel, dst, src);
+    ec_code_intel_op_1(&intel, 0x89, 0);
+    ec_code_intel_rex(&intel, _gf_true);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_mov_r2m(ec_code_builder_t *builder, ec_code_intel_reg_t src,
+                         ec_code_intel_reg_t base, ec_code_intel_reg_t index,
+                         uint32_t scale, int32_t offset)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_mem(&intel, src, base, index, scale, offset);
+    ec_code_intel_op_1(&intel, 0x89, 0);
+    ec_code_intel_rex(&intel, _gf_true);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_mov_m2r(ec_code_builder_t *builder, ec_code_intel_reg_t base,
+                         ec_code_intel_reg_t index, uint32_t scale,
+                         int32_t offset, ec_code_intel_reg_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_mem(&intel, dst, base, index, scale, offset);
+    ec_code_intel_op_1(&intel, 0x8B, 0);
+    ec_code_intel_rex(&intel, _gf_true);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_xor_r2r(ec_code_builder_t *builder, ec_code_intel_reg_t src,
+                         ec_code_intel_reg_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_reg(&intel, dst, src);
+    ec_code_intel_op_1(&intel, 0x31, 0);
+    ec_code_intel_rex(&intel, _gf_true);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_xor_m2r(ec_code_builder_t *builder, ec_code_intel_reg_t base,
+                         ec_code_intel_reg_t index, uint32_t scale,
+                         int32_t offset, ec_code_intel_reg_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_mem(&intel, dst, base, index, scale, offset);
+    ec_code_intel_op_1(&intel, 0x33, 0);
+    ec_code_intel_rex(&intel, _gf_true);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_add_i2r(ec_code_builder_t *builder, int32_t value,
+                         ec_code_intel_reg_t reg)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    if ((value >= -128) && (value < 128)) {
+        ec_code_intel_modrm_reg(&intel, reg, 0);
+        ec_code_intel_op_1(&intel, 0x83, 0);
+        ec_code_intel_immediate_1(&intel, value);
+    } else {
+        if (reg == REG_AX) {
+            ec_code_intel_op_1(&intel, 0x05, reg);
+        } else {
+            ec_code_intel_modrm_reg(&intel, reg, 0);
+            ec_code_intel_op_1(&intel, 0x81, 0);
+        }
+        ec_code_intel_immediate_4(&intel, value);
+    }
+    ec_code_intel_rex(&intel, _gf_true);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_test_i2r(ec_code_builder_t *builder, uint32_t value,
+                          ec_code_intel_reg_t reg)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    if (reg == REG_AX) {
+        ec_code_intel_op_1(&intel, 0xA9, reg);
+    } else {
+        ec_code_intel_modrm_reg(&intel, reg, 0);
+        ec_code_intel_op_1(&intel, 0xF7, 0);
+    }
+    ec_code_intel_immediate_4(&intel, value);
+    ec_code_intel_rex(&intel, _gf_true);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_jne(ec_code_builder_t *builder, uint32_t address)
+{
+    ec_code_intel_t intel;
+    int32_t rel;
+
+    ec_code_intel_init(&intel);
+
+    rel = address - builder->address - 2;
+    if ((rel >= -128) && (rel < 128)) {
+        ec_code_intel_op_1(&intel, 0x75, 0);
+        ec_code_intel_immediate_1(&intel, rel);
+    } else {
+        rel -= 4;
+        ec_code_intel_op_2(&intel, 0x0F, 0x85, 0);
+        ec_code_intel_immediate_4(&intel, rel);
+    }
+    ec_code_intel_rex(&intel, _gf_false);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_mov_sse2sse(ec_code_builder_t *builder, uint32_t src,
+                             uint32_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_prefix(&intel, 0x66);
+    ec_code_intel_modrm_reg(&intel, src, dst);
+    ec_code_intel_op_2(&intel, 0x0F, 0x6F, 0);
+    ec_code_intel_rex(&intel, _gf_false);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_mov_sse2m(ec_code_builder_t *builder, uint32_t src,
+                           ec_code_intel_reg_t base, ec_code_intel_reg_t index,
+                           uint32_t scale, int32_t offset)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_prefix(&intel, 0x66);
+    ec_code_intel_modrm_mem(&intel, src, base, index, scale, offset);
+    ec_code_intel_op_2(&intel, 0x0F, 0x7F, 0);
+    ec_code_intel_rex(&intel, _gf_false);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_mov_m2sse(ec_code_builder_t *builder,
+                           ec_code_intel_reg_t base, ec_code_intel_reg_t index,
+                           uint32_t scale, int32_t offset, uint32_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_prefix(&intel, 0x66);
+    ec_code_intel_modrm_mem(&intel, dst, base, index, scale, offset);
+    ec_code_intel_op_2(&intel, 0x0F, 0x6F, 0);
+    ec_code_intel_rex(&intel, _gf_false);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_xor_sse2sse(ec_code_builder_t *builder, uint32_t src,
+                             uint32_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_prefix(&intel, 0x66);
+    ec_code_intel_modrm_reg(&intel, src, dst);
+    ec_code_intel_op_2(&intel, 0x0F, 0xEF, 0);
+    ec_code_intel_rex(&intel, _gf_false);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_xor_m2sse(ec_code_builder_t *builder,
+                           ec_code_intel_reg_t base, ec_code_intel_reg_t index,
+                           uint32_t scale, int32_t offset, uint32_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_prefix(&intel, 0x66);
+    ec_code_intel_modrm_mem(&intel, dst, base, index, scale, offset);
+    ec_code_intel_op_2(&intel, 0x0F, 0xEF, 0);
+    ec_code_intel_rex(&intel, _gf_false);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_mov_avx2avx(ec_code_builder_t *builder, uint32_t src,
+                             uint32_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_reg(&intel, src, dst);
+    ec_code_intel_op_1(&intel, 0x6F, 0);
+    ec_code_intel_vex(&intel, _gf_false, _gf_true, VEX_OPCODE_0F,
+                      VEX_PREFIX_66, VEX_REG_NONE);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_mov_avx2m(ec_code_builder_t *builder, uint32_t src,
+                           ec_code_intel_reg_t base, ec_code_intel_reg_t index,
+                           uint32_t scale, int32_t offset)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_mem(&intel, src, base, index, scale, offset);
+    ec_code_intel_op_1(&intel, 0x7F, 0);
+    ec_code_intel_vex(&intel, _gf_false, _gf_true, VEX_OPCODE_0F,
+                      VEX_PREFIX_66, VEX_REG_NONE);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_mov_m2avx(ec_code_builder_t *builder,
+                           ec_code_intel_reg_t base, ec_code_intel_reg_t index,
+                           uint32_t scale, int32_t offset, uint32_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_mem(&intel, dst, base, index, scale, offset);
+    ec_code_intel_op_1(&intel, 0x6F, 0);
+    ec_code_intel_vex(&intel, _gf_false, _gf_true, VEX_OPCODE_0F,
+                      VEX_PREFIX_66, VEX_REG_NONE);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_xor_avx2avx(ec_code_builder_t *builder, uint32_t src,
+                             uint32_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_reg(&intel, src, dst);
+    ec_code_intel_op_1(&intel, 0xEF, 0);
+    ec_code_intel_vex(&intel, _gf_false, _gf_true, VEX_OPCODE_0F,
+                      VEX_PREFIX_66, dst);
+
+    ec_code_intel_emit(builder, &intel);
+}
+
+void
+ec_code_intel_op_xor_m2avx(ec_code_builder_t *builder,
+                           ec_code_intel_reg_t base, ec_code_intel_reg_t index,
+                           uint32_t scale, int32_t offset, uint32_t dst)
+{
+    ec_code_intel_t intel;
+
+    ec_code_intel_init(&intel);
+
+    ec_code_intel_modrm_mem(&intel, dst, base, index, scale, offset);
+    ec_code_intel_op_1(&intel, 0xEF, 0);
+    ec_code_intel_vex(&intel, _gf_false, _gf_true, VEX_OPCODE_0F,
+                      VEX_PREFIX_66, dst);
+
+    ec_code_intel_emit(builder, &intel);
+}
diff --git a/xlators/cluster/ec/src/ec-code-intel.h b/xlators/cluster/ec/src/ec-code-intel.h
new file mode 100644
index 00000000000..903d023f962
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-intel.h
@@ -0,0 +1,184 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#ifndef __EC_CODE_INTEL_H__
+#define __EC_CODE_INTEL_H__
+
+#include "ec-code.h"
+
+#define VEX_REG_NONE 0
+
+enum _ec_code_intel_reg;
+typedef enum _ec_code_intel_reg ec_code_intel_reg_t;
+
+enum _ec_code_vex_prefix;
+typedef enum _ec_code_vex_prefix ec_code_vex_prefix_t;
+
+enum _ec_code_vex_opcode;
+typedef enum _ec_code_vex_opcode ec_code_vex_opcode_t;
+
+struct _ec_code_intel_buffer;
+typedef struct _ec_code_intel_buffer ec_code_intel_buffer_t;
+
+struct _ec_code_intel_sib;
+typedef struct _ec_code_intel_sib ec_code_intel_sib_t;
+
+struct _ec_code_intel_modrm;
+typedef struct _ec_code_intel_modrm ec_code_intel_modrm_t;
+
+struct _ec_code_intel_rex;
+typedef struct _ec_code_intel_rex ec_code_intel_rex_t;
+
+struct _ec_code_intel;
+typedef struct _ec_code_intel ec_code_intel_t;
+
+enum _ec_code_intel_reg {
+    REG_NULL = -1,
+    REG_AX,
+    REG_CX,
+    REG_DX,
+    REG_BX,
+    REG_SP,
+    REG_BP,
+    REG_SI,
+    REG_DI,
+    REG_8,
+    REG_9,
+    REG_10,
+    REG_11,
+    REG_12,
+    REG_13,
+    REG_14,
+    REG_15
+};
+
+enum _ec_code_vex_prefix {
+    VEX_PREFIX_NONE = 0,
+    VEX_PREFIX_66,
+    VEX_PREFIX_F3,
+    VEX_PREFIX_F2
+};
+
+enum _ec_code_vex_opcode {
+    VEX_OPCODE_NONE = 0,
+    VEX_OPCODE_0F,
+    VEX_OPCODE_0F_38,
+    VEX_OPCODE_0F_3A
+};
+
+struct _ec_code_intel_buffer {
+    uint32_t bytes;
+    union {
+        uint8_t  data[4];
+        uint32_t value;
+    };
+};
+
+struct _ec_code_intel_sib {
+    gf_boolean_t present;
+    uint32_t     base;
+    uint32_t     index;
+    uint32_t     scale;
+};
+
+struct _ec_code_intel_modrm {
+    gf_boolean_t present;
+    uint32_t     mod;
+    uint32_t     rm;
+    uint32_t     reg;
+};
+
+struct _ec_code_intel_rex {
+    gf_boolean_t present;
+    uint32_t     w;
+    uint32_t     r;
+    uint32_t     x;
+    uint32_t     b;
+};
+
+struct _ec_code_intel {
+    gf_boolean_t           invalid;
+    ec_code_intel_buffer_t prefix;
+    ec_code_intel_buffer_t opcode;
+    ec_code_intel_buffer_t offset;
+    ec_code_intel_buffer_t immediate;
+    ec_code_intel_buffer_t vex;
+    ec_code_intel_rex_t    rex;
+    ec_code_intel_modrm_t  modrm;
+    ec_code_intel_sib_t    sib;
+    uint32_t               reg;
+};
+
+void ec_code_intel_op_push_r(ec_code_builder_t *builder,
+                             ec_code_intel_reg_t reg);
+void ec_code_intel_op_pop_r(ec_code_builder_t *builder,
+                            ec_code_intel_reg_t reg);
+void ec_code_intel_op_ret(ec_code_builder_t *builder, uint32_t size);
+
+void ec_code_intel_op_mov_r2r(ec_code_builder_t *builder,
+                              ec_code_intel_reg_t src,
+                              ec_code_intel_reg_t dst);
+void ec_code_intel_op_mov_r2m(ec_code_builder_t *builder,
+                              ec_code_intel_reg_t src,
+                              ec_code_intel_reg_t base,
+                              ec_code_intel_reg_t index, uint32_t scale,
+                              int32_t offset);
+void ec_code_intel_op_mov_m2r(ec_code_builder_t *builder,
+                              ec_code_intel_reg_t base,
+                              ec_code_intel_reg_t index, uint32_t scale,
+                              int32_t offset, ec_code_intel_reg_t dst);
+void ec_code_intel_op_xor_r2r(ec_code_builder_t *builder,
+                              ec_code_intel_reg_t src,
+                              ec_code_intel_reg_t dst);
+void ec_code_intel_op_xor_m2r(ec_code_builder_t *builder,
+                              ec_code_intel_reg_t base,
+                              ec_code_intel_reg_t index, uint32_t scale,
+                              int32_t offset, ec_code_intel_reg_t dst);
+void ec_code_intel_op_add_i2r(ec_code_builder_t *builder, int32_t value,
+                              ec_code_intel_reg_t reg);
+void ec_code_intel_op_test_i2r(ec_code_builder_t *builder, uint32_t value,
+                               ec_code_intel_reg_t reg);
+void ec_code_intel_op_jne(ec_code_builder_t *builder, uint32_t address);
+
+void ec_code_intel_op_mov_sse2sse(ec_code_builder_t *builder, uint32_t src,
+                                  uint32_t dst);
+void ec_code_intel_op_mov_sse2m(ec_code_builder_t *builder, uint32_t src,
+                                ec_code_intel_reg_t base,
+                                ec_code_intel_reg_t index, uint32_t scale,
+                                int32_t offset);
+void ec_code_intel_op_mov_m2sse(ec_code_builder_t *builder,
+                                ec_code_intel_reg_t base,
+                                ec_code_intel_reg_t index, uint32_t scale,
+                                int32_t offset, uint32_t dst);
+void ec_code_intel_op_xor_sse2sse(ec_code_builder_t *builder, uint32_t src,
+                                  uint32_t dst);
+void ec_code_intel_op_xor_m2sse(ec_code_builder_t *builder,
+                                ec_code_intel_reg_t base,
+                                ec_code_intel_reg_t index, uint32_t scale,
+                                int32_t offset, uint32_t dst);
+
+void ec_code_intel_op_mov_avx2avx(ec_code_builder_t *builder, uint32_t src,
+                                  uint32_t dst);
+void ec_code_intel_op_mov_avx2m(ec_code_builder_t *builder, uint32_t src,
+                                ec_code_intel_reg_t base,
+                                ec_code_intel_reg_t index, uint32_t scale,
+                                int32_t offset);
+void ec_code_intel_op_mov_m2avx(ec_code_builder_t *builder,
+                                ec_code_intel_reg_t base,
+                                ec_code_intel_reg_t index, uint32_t scale,
+                                int32_t offset, uint32_t dst);
+void ec_code_intel_op_xor_avx2avx(ec_code_builder_t *builder, uint32_t src,
+                                  uint32_t dst);
+void ec_code_intel_op_xor_m2avx(ec_code_builder_t *builder,
+                                ec_code_intel_reg_t base,
+                                ec_code_intel_reg_t index, uint32_t scale,
+                                int32_t offset, uint32_t dst);
+
+#endif /* __EC_CODE_INTEL_H__ */
diff --git a/xlators/cluster/ec/src/ec-code-sse.c b/xlators/cluster/ec/src/ec-code-sse.c
new file mode 100644
index 00000000000..6f2c6fa593f
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-sse.c
@@ -0,0 +1,108 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#include <errno.h>
+
+#include "ec-code-intel.h"
+
+static void
+ec_code_sse_prolog(ec_code_builder_t *builder)
+{
+    builder->loop = builder->address;
+}
+
+static void
+ec_code_sse_epilog(ec_code_builder_t *builder)
+{
+    ec_code_intel_op_add_i2r(builder, 16, REG_DX);
+    ec_code_intel_op_add_i2r(builder, 16, REG_DI);
+    ec_code_intel_op_test_i2r(builder, builder->width - 1, REG_DX);
+    ec_code_intel_op_jne(builder, builder->loop);
+
+    ec_code_intel_op_ret(builder, 0);
+}
+
+static void
+ec_code_sse_load(ec_code_builder_t *builder, uint32_t dst, uint32_t idx,
+                 uint32_t bit)
+{
+    if (builder->linear) {
+        ec_code_intel_op_mov_m2sse(builder, REG_SI, REG_DX, 1,
+                                   idx * builder->width * builder->bits +
+                                   bit * builder->width,
+                                   dst);
+    } else {
+        if (builder->base != idx) {
+            ec_code_intel_op_mov_m2r(builder, REG_SI, REG_NULL, 0, idx * 8,
+                                     REG_AX);
+            builder->base = idx;
+        }
+        ec_code_intel_op_mov_m2sse(builder, REG_AX, REG_DX, 1,
+                                   bit * builder->width, dst);
+    }
+}
+
+static void
+ec_code_sse_store(ec_code_builder_t *builder, uint32_t src, uint32_t bit)
+{
+    ec_code_intel_op_mov_sse2m(builder, src, REG_DI, REG_NULL, 0,
+                               bit * builder->width);
+}
+
+static void
+ec_code_sse_copy(ec_code_builder_t *builder, uint32_t dst, uint32_t src)
+{
+    ec_code_intel_op_mov_sse2sse(builder, src, dst);
+}
+
+static void
+ec_code_sse_xor2(ec_code_builder_t *builder, uint32_t dst, uint32_t src)
+{
+    ec_code_intel_op_xor_sse2sse(builder, src, dst);
+}
+
+static void
+ec_code_sse_xorm(ec_code_builder_t *builder, uint32_t dst, uint32_t idx,
+                 uint32_t bit)
+{
+    if (builder->linear) {
+       ec_code_intel_op_xor_m2sse(builder, REG_SI, REG_DX, 1,
+                                  idx * builder->width * builder->bits +
+                                  bit * builder->width,
+                                  dst);
+    } else {
+        if (builder->base != idx) {
+            ec_code_intel_op_mov_m2r(builder, REG_SI, REG_NULL, 0, idx * 8,
+                                     REG_AX);
+            builder->base = idx;
+        }
+        ec_code_intel_op_xor_m2sse(builder, REG_AX, REG_DX, 1,
+                                   bit * builder->width, dst);
+    }
+}
+
+static char *ec_code_sse_needed_flags[] = {
+    "sse2",
+    NULL
+};
+
+ec_code_gen_t ec_code_gen_sse = {
+    .name   = "sse",
+    .flags  = ec_code_sse_needed_flags,
+    .width  = 16,
+    .prolog = ec_code_sse_prolog,
+    .epilog = ec_code_sse_epilog,
+    .load   = ec_code_sse_load,
+    .store  = ec_code_sse_store,
+    .copy   = ec_code_sse_copy,
+    .xor2   = ec_code_sse_xor2,
+    .xor3   = NULL,
+    .xorm   = ec_code_sse_xorm
+};
diff --git a/xlators/cluster/ec/src/ec-code-sse.h b/xlators/cluster/ec/src/ec-code-sse.h
new file mode 100644
index 00000000000..f1acbcf894b
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-sse.h
@@ -0,0 +1,18 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#ifndef __EC_CODE_SSE_H__
+#define __EC_CODE_SSE_H__
+
+#include "ec-code.h"
+
+extern ec_code_gen_t ec_code_gen_sse;
+
+#endif /* __EC_CODE_SSE_H__ */
diff --git a/xlators/cluster/ec/src/ec-code-x64.c b/xlators/cluster/ec/src/ec-code-x64.c
new file mode 100644
index 00000000000..e94ddd4b155
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-x64.c
@@ -0,0 +1,150 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#include <errno.h>
+
+#include "ec-code-intel.h"
+
+static ec_code_intel_reg_t ec_code_x64_regmap[] = {
+    REG_AX, REG_CX, REG_BP, REG_8,  REG_9, REG_10,
+    REG_11, REG_12, REG_13, REG_14, REG_15
+};
+
+static void
+ec_code_x64_prolog(ec_code_builder_t *builder)
+{
+    uint32_t i;
+
+    ec_code_intel_op_push_r(builder, REG_BP);
+    if (!builder->linear) {
+        ec_code_intel_op_push_r(builder, REG_BX);
+    }
+    if (builder->regs > 11) {
+        ec_code_error(builder, EINVAL);
+        return;
+    }
+    for (i = 7; i < builder->regs; i++) {
+        ec_code_intel_op_push_r(builder, ec_code_x64_regmap[i]);
+    }
+
+    builder->loop = builder->address;
+}
+
+static void
+ec_code_x64_epilog(ec_code_builder_t *builder)
+{
+    uint32_t i;
+
+    ec_code_intel_op_add_i2r(builder, 8, REG_DX);
+    ec_code_intel_op_add_i2r(builder, 8, REG_DI);
+    ec_code_intel_op_test_i2r(builder, builder->width - 1, REG_DX);
+    ec_code_intel_op_jne(builder, builder->loop);
+
+    if (builder->regs > 11) {
+        ec_code_error(builder, EINVAL);
+    }
+    for (i = builder->regs; i > 7; i--) {
+        ec_code_intel_op_pop_r(builder, ec_code_x64_regmap[i - 1]);
+    }
+    if (!builder->linear) {
+        ec_code_intel_op_pop_r(builder, REG_BX);
+    }
+    ec_code_intel_op_pop_r(builder, REG_BP);
+    ec_code_intel_op_ret(builder, 0);
+}
+
+static void
+ec_code_x64_load(ec_code_builder_t *builder, uint32_t dst, uint32_t idx,
+                 uint32_t bit)
+{
+    dst = ec_code_x64_regmap[dst];
+
+    if (builder->linear) {
+        ec_code_intel_op_mov_m2r(builder, REG_SI, REG_DX, 1,
+                                 idx * builder->width * builder->bits +
+                                 bit * builder->width,
+                                 dst);
+    } else {
+        if (builder->base != idx) {
+            ec_code_intel_op_mov_m2r(builder, REG_SI, REG_NULL, 0, idx * 8,
+                                     REG_BX);
+            builder->base = idx;
+        }
+        ec_code_intel_op_mov_m2r(builder, REG_BX, REG_DX, 1,
+                                 bit * builder->width, dst);
+    }
+}
+
+static void
+ec_code_x64_store(ec_code_builder_t *builder, uint32_t src, uint32_t bit)
+{
+    src = ec_code_x64_regmap[src];
+
+    ec_code_intel_op_mov_r2m(builder, src, REG_DI, REG_NULL, 0,
+                             bit * builder->width);
+}
+
+static void
+ec_code_x64_copy(ec_code_builder_t *builder, uint32_t dst, uint32_t src)
+{
+    dst = ec_code_x64_regmap[dst];
+    src = ec_code_x64_regmap[src];
+
+    ec_code_intel_op_mov_r2r(builder, src, dst);
+}
+
+static void
+ec_code_x64_xor2(ec_code_builder_t *builder, uint32_t dst, uint32_t src)
+{
+    dst = ec_code_x64_regmap[dst];
+    src = ec_code_x64_regmap[src];
+
+    ec_code_intel_op_xor_r2r(builder, src, dst);
+}
+
+static void
+ec_code_x64_xorm(ec_code_builder_t *builder, uint32_t dst, uint32_t idx,
+                 uint32_t bit)
+{
+    dst = ec_code_x64_regmap[dst];
+
+    if (builder->linear) {
+        ec_code_intel_op_xor_m2r(builder, REG_SI, REG_DX, 1,
+                                 idx * builder->width * builder->bits +
+                                 bit * builder->width,
+                                 dst);
+    } else {
+        if (builder->base != idx) {
+            ec_code_intel_op_mov_m2r(builder, REG_SI, REG_NULL, 0, idx * 8,
+                                     REG_BX);
+            builder->base = idx;
+        }
+        ec_code_intel_op_xor_m2r(builder, REG_BX, REG_DX, 1,
+                                 bit * builder->width, dst);
+    }
+}
+
+static char *ec_code_x64_needed_flags[] = {
+    NULL
+};
+
+ec_code_gen_t ec_code_gen_x64 = {
+    .name   = "x64",
+    .flags  = ec_code_x64_needed_flags,
+    .width  = sizeof(uint64_t),
+    .prolog = ec_code_x64_prolog,
+    .epilog = ec_code_x64_epilog,
+    .load   = ec_code_x64_load,
+    .store  = ec_code_x64_store,
+    .copy   = ec_code_x64_copy,
+    .xor2   = ec_code_x64_xor2,
+    .xor3   = NULL,
+    .xorm   = ec_code_x64_xorm
+};
diff --git a/xlators/cluster/ec/src/ec-code-x64.h b/xlators/cluster/ec/src/ec-code-x64.h
new file mode 100644
index 00000000000..bd8174e4bf5
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code-x64.h
@@ -0,0 +1,18 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#ifndef __EC_CODE_X64_H__
+#define __EC_CODE_X64_H__
+
+#include "ec-code.h"
+
+extern ec_code_gen_t ec_code_gen_x64;
+
+#endif /* __EC_CODE_X64_H__ */
diff --git a/xlators/cluster/ec/src/ec-code.c b/xlators/cluster/ec/src/ec-code.c
new file mode 100644
index 00000000000..a1f652779f3
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code.c
@@ -0,0 +1,904 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#include <string.h>
+#include <sys/mman.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <ctype.h>
+
+#include "syscall.h"
+
+#include "ec-mem-types.h"
+#include "ec-code.h"
+#include "ec-messages.h"
+#include "ec-code-c.h"
+
+#ifdef USE_EC_DYNAMIC_X64
+#include "ec-code-x64.h"
+#endif
+
+#ifdef USE_EC_DYNAMIC_SSE
+#include "ec-code-sse.h"
+#endif
+
+#ifdef USE_EC_DYNAMIC_AVX
+#include "ec-code-avx.h"
+#endif
+
+#define EC_PROC_BUFFER_SIZE 4096
+
+#define PROC_CPUINFO "/proc/cpuinfo"
+
+struct _ec_code_proc;
+typedef struct _ec_code_proc ec_code_proc_t;
+
+struct _ec_code_proc {
+    int32_t      fd;
+    gf_boolean_t eof;
+    gf_boolean_t error;
+    gf_boolean_t skip;
+    ssize_t      size;
+    ssize_t      pos;
+    char         buffer[EC_PROC_BUFFER_SIZE];
+};
+
+static ec_code_gen_t *ec_code_gen_table[] = {
+#ifdef USE_EC_DYNAMIC_AVX
+    &ec_code_gen_avx,
+#endif
+#ifdef USE_EC_DYNAMIC_SSE
+    &ec_code_gen_sse,
+#endif
+#ifdef USE_EC_DYNAMIC_X64
+    &ec_code_gen_x64,
+#endif
+    NULL
+};
+
+static void
+ec_code_arg_set(ec_code_arg_t *arg, uint32_t value)
+{
+    arg->value = value;
+}
+
+static void
+ec_code_arg_assign(ec_code_builder_t *builder, ec_code_op_t *op,
+                   ec_code_arg_t *arg, uint32_t reg)
+{
+    arg->value = reg;
+
+    if (builder->regs <= reg) {
+        builder->regs = reg + 1;
+    }
+
+}
+
+static void
+ec_code_arg_use(ec_code_builder_t *builder, ec_code_op_t *op,
+                ec_code_arg_t *arg, uint32_t reg)
+{
+    arg->value = reg;
+}
+
+static void
+ec_code_arg_update(ec_code_builder_t *builder, ec_code_op_t *op,
+                   ec_code_arg_t *arg, uint32_t reg)
+{
+    arg->value = reg;
+}
+
+static ec_code_op_t *
+ec_code_op_next(ec_code_builder_t *builder)
+{
+    ec_code_op_t *op;
+
+    op = &builder->ops[builder->count++];
+    memset(op, 0, sizeof(ec_code_op_t));
+
+    return op;
+}
+
+static void
+ec_code_load(ec_code_builder_t *builder, uint32_t bit, uint32_t offset)
+{
+    ec_code_op_t *op;
+
+    op = ec_code_op_next(builder);
+
+    op->op = EC_GF_OP_LOAD;
+    ec_code_arg_assign(builder, op, &op->arg1, builder->map[bit]);
+    ec_code_arg_set(&op->arg2, offset);
+    ec_code_arg_set(&op->arg3, bit);
+}
+
+static void
+ec_code_store(ec_code_builder_t *builder, uint32_t reg, uint32_t bit)
+{
+    ec_code_op_t *op;
+
+    op = ec_code_op_next(builder);
+
+    op->op = EC_GF_OP_STORE;
+    ec_code_arg_use(builder, op, &op->arg1, builder->map[reg]);
+    ec_code_arg_set(&op->arg2, 0);
+    ec_code_arg_set(&op->arg3, bit);
+}
+
+static void
+ec_code_copy(ec_code_builder_t *builder, uint32_t dst, uint32_t src)
+{
+    ec_code_op_t *op;
+
+    op = ec_code_op_next(builder);
+
+    op->op = EC_GF_OP_COPY;
+    ec_code_arg_assign(builder, op, &op->arg1, builder->map[dst]);
+    ec_code_arg_use(builder, op, &op->arg2, builder->map[src]);
+    ec_code_arg_set(&op->arg3, 0);
+}
+
+static void
+ec_code_xor2(ec_code_builder_t *builder, uint32_t dst, uint32_t src)
+{
+    ec_code_op_t *op;
+
+    op = ec_code_op_next(builder);
+
+    op->op = EC_GF_OP_XOR2;
+    ec_code_arg_update(builder, op, &op->arg1, builder->map[dst]);
+    ec_code_arg_use(builder, op, &op->arg2, builder->map[src]);
+    ec_code_arg_set(&op->arg3, 0);
+}
+
+static void
+ec_code_xor3(ec_code_builder_t *builder, uint32_t dst, uint32_t src1,
+             uint32_t src2)
+{
+    ec_code_op_t *op;
+
+    if (builder->code->gen->xor3 == NULL) {
+        ec_code_copy(builder, dst, src1);
+        ec_code_xor2(builder, dst, src2);
+
+        return;
+    }
+
+    op = ec_code_op_next(builder);
+
+    op->op = EC_GF_OP_XOR3;
+    ec_code_arg_assign(builder, op, &op->arg1, builder->map[dst]);
+    ec_code_arg_use(builder, op, &op->arg2, builder->map[src1]);
+    ec_code_arg_use(builder, op, &op->arg3, builder->map[src2]);
+}
+
+static void
+ec_code_xorm(ec_code_builder_t *builder, uint32_t bit, uint32_t offset)
+{
+    ec_code_op_t *op;
+
+    op = ec_code_op_next(builder);
+
+    op->op = EC_GF_OP_XORM;
+    ec_code_arg_update(builder, op, &op->arg1, builder->map[bit]);
+    ec_code_arg_set(&op->arg2, offset);
+    ec_code_arg_set(&op->arg3, bit);
+}
+
+static void
+ec_code_dup(ec_code_builder_t *builder, ec_gf_op_t *op)
+{
+    switch (op->op) {
+    case EC_GF_OP_COPY:
+        ec_code_copy(builder, op->arg1, op->arg2);
+        break;
+    case EC_GF_OP_XOR2:
+        ec_code_xor2(builder, op->arg1, op->arg2);
+        break;
+    case EC_GF_OP_XOR3:
+        ec_code_xor3(builder, op->arg1, op->arg2, op->arg3);
+        break;
+    default:
+        break;
+    }
+}
+
+static void
+ec_code_gf_load(ec_code_builder_t *builder, uint32_t offset)
+{
+    uint32_t i;
+
+    for (i = 0; i < builder->code->gf->bits; i++) {
+        ec_code_load(builder, i, offset);
+    }
+}
+
+static void
+ec_code_gf_load_xor(ec_code_builder_t *builder, uint32_t offset)
+{
+    uint32_t i;
+
+    for (i = 0; i < builder->code->gf->bits; i++) {
+        ec_code_xorm(builder, i, offset);
+    }
+}
+
+static void
+ec_code_gf_store(ec_code_builder_t *builder)
+{
+    uint32_t i;
+
+    for (i = 0; i < builder->code->gf->bits; i++) {
+        ec_code_store(builder, i, i);
+    }
+}
+
+static void
+ec_code_gf_clear(ec_code_builder_t *builder)
+{
+    uint32_t i;
+
+    ec_code_xor2(builder, 0, 0);
+    for (i = 0; i < builder->code->gf->bits; i++) {
+        ec_code_store(builder, 0, i);
+    }
+}
+
+static void
+ec_code_gf_mul(ec_code_builder_t *builder, uint32_t value)
+{
+    ec_gf_mul_t *mul;
+    ec_gf_op_t *op;
+    uint32_t map[EC_GF_MAX_REGS];
+    int32_t i;
+
+    mul = builder->code->gf->table[value];
+    for (op = mul->ops; op->op != EC_GF_OP_END; op++) {
+        ec_code_dup(builder, op);
+    }
+
+    for (i = 0; i < mul->regs; i++) {
+        map[i] = builder->map[mul->map[i]];
+    }
+    memcpy(builder->map, map, sizeof(uint32_t) * mul->regs);
+}
+
+static ec_code_builder_t *
+ec_code_prepare(ec_code_t *code, uint32_t count, uint32_t width,
+                gf_boolean_t linear)
+{
+    ec_code_builder_t *builder;
+    uint32_t i;
+
+    count *= code->gf->bits + code->gf->max_ops;
+    count += code->gf->bits;
+    builder = GF_MALLOC(sizeof(ec_code_builder_t) +
+                        sizeof(ec_code_op_t) * count, ec_mt_ec_code_builder_t);
+    if (builder == NULL) {
+        return NULL;
+    }
+
+    builder->address = 0;
+    builder->code = code;
+    builder->size = 0;
+    builder->count = 0;
+    builder->regs = 0;
+    builder->error = 0;
+    builder->bits = code->gf->bits;
+    builder->width = width;
+    builder->data = NULL;
+    builder->linear = linear;
+    builder->base = -1;
+
+    for (i = 0; i < EC_GF_MAX_REGS; i++) {
+        builder->map[i] = i;
+    }
+
+    return builder;
+}
+
+static size_t
+ec_code_space_size(void)
+{
+    return (sizeof(ec_code_space_t) + 15) & ~15;
+}
+
+static size_t
+ec_code_chunk_size(void)
+{
+    return (sizeof(ec_code_chunk_t) + 15) & ~15;
+}
+
+static ec_code_chunk_t *
+ec_code_chunk_from_space(ec_code_space_t *space)
+{
+    return (ec_code_chunk_t *)((uintptr_t)space + ec_code_space_size());
+}
+
+static void *
+ec_code_func_from_chunk(ec_code_chunk_t *chunk)
+{
+    return (void *)((uintptr_t)chunk + ec_code_chunk_size());
+}
+
+static ec_code_chunk_t *
+ec_code_chunk_from_func(ec_code_func_linear_t func)
+{
+    return (ec_code_chunk_t *)((uintptr_t)func - ec_code_chunk_size());
+}
+
+static ec_code_chunk_t *
+ec_code_chunk_split(ec_code_chunk_t *chunk, size_t size)
+{
+    ec_code_chunk_t *extra;
+    ssize_t avail;
+
+    avail = chunk->size - size - ec_code_chunk_size();
+    if (avail > 0) {
+        extra = (ec_code_chunk_t *)((uintptr_t)chunk + chunk->size - avail);
+        extra->size = avail;
+        list_add(&extra->list, &chunk->list);
+        chunk->size = size;
+    }
+    list_del_init(&chunk->list);
+
+    return chunk;
+}
+
+static gf_boolean_t
+ec_code_chunk_touch(ec_code_chunk_t *prev, ec_code_chunk_t *next)
+{
+    uintptr_t end;
+
+    end = (uintptr_t)prev + ec_code_chunk_size() + prev->size;
+    return (end == (uintptr_t)next);
+}
+
+static void
+ec_code_chunk_merge(ec_code_chunk_t *chunk)
+{
+    ec_code_chunk_t *item;
+
+    list_for_each_entry(item, &chunk->space->chunks, list) {
+        if (ec_code_chunk_touch(item, chunk)) {
+            item->size += chunk->size + ec_code_chunk_size();
+            chunk = item;
+
+            goto check;
+        }
+        if ((uintptr_t)item > (uintptr_t)chunk) {
+            list_add_tail(&chunk->list, &item->list);
+            if (ec_code_chunk_touch(chunk, item)) {
+                chunk->size += item->size + ec_code_chunk_size();
+                list_del_init(&item->list);
+            }
+
+            goto check;
+        }
+    }
+    list_add_tail(&chunk->list, &chunk->space->chunks);
+
+check:
+    if (chunk->size == EC_CODE_SIZE - ec_code_space_size() -
+                                      ec_code_chunk_size()) {
+        list_del_init(&chunk->space->list);
+
+        munmap(chunk->space, chunk->space->size);
+    }
+}
+
+static ec_code_chunk_t *
+ec_code_space_alloc(ec_code_t *code, size_t size)
+{
+    ec_code_space_t *space;
+    ec_code_chunk_t *chunk;
+    size_t map_size;
+
+    size = (size + 15) & ~15;
+    list_for_each_entry(space, &code->spaces, list) {
+        list_for_each_entry(chunk, &space->chunks, list) {
+            if (chunk->size >= size) {
+                goto out;
+            }
+        }
+    }
+
+    map_size = EC_CODE_SIZE;
+    if (map_size < size) {
+        map_size = size;
+    }
+    space = mmap(NULL, map_size, PROT_EXEC | PROT_READ | PROT_WRITE,
+                 MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+    if (space == NULL) {
+        return NULL;
+    }
+    /* It's not important to check the return value of mlock(). If it fails
+     * everything will continue to work normally. */
+    mlock(space, map_size);
+
+    space->code = code;
+    space->size = map_size;
+    list_add_tail(&space->list, &code->spaces);
+    INIT_LIST_HEAD(&space->chunks);
+
+    chunk = ec_code_chunk_from_space(space);
+    chunk->size = EC_CODE_SIZE - ec_code_space_size() - ec_code_chunk_size();
+    list_add(&chunk->list, &space->chunks);
+
+out:
+    chunk->space = space;
+
+    return ec_code_chunk_split(chunk, size);
+}
+
+static ec_code_chunk_t *
+ec_code_alloc(ec_code_t *code, uint32_t size)
+{
+    ec_code_chunk_t *chunk;
+
+    LOCK(&code->lock);
+
+    chunk = ec_code_space_alloc(code, size);
+
+    UNLOCK(&code->lock);
+
+    return chunk;
+}
+
+static void
+ec_code_free(ec_code_chunk_t *chunk)
+{
+    gf_lock_t *lock;
+
+    lock = &chunk->space->code->lock;
+    LOCK(lock);
+
+    ec_code_chunk_merge(chunk);
+
+    UNLOCK(lock);
+}
+
+static gf_boolean_t
+ec_code_write(ec_code_builder_t *builder)
+{
+    ec_code_gen_t *gen;
+    ec_code_op_t *op;
+    uint32_t i;
+
+    builder->error = 0;
+    builder->size = 0;
+    builder->address = 0;
+    builder->base = -1;
+
+    gen = builder->code->gen;
+    gen->prolog(builder);
+    for (i = 0; i < builder->count; i++) {
+        op = &builder->ops[i];
+        switch (op->op) {
+        case EC_GF_OP_LOAD:
+            gen->load(builder, op->arg1.value, op->arg2.value, op->arg3.value);
+            break;
+        case EC_GF_OP_STORE:
+            gen->store(builder, op->arg1.value, op->arg3.value);
+            break;
+        case EC_GF_OP_COPY:
+            gen->copy(builder, op->arg1.value, op->arg2.value);
+            break;
+        case EC_GF_OP_XOR2:
+            gen->xor2(builder, op->arg1.value, op->arg2.value);
+            break;
+        case EC_GF_OP_XOR3:
+            gen->xor3(builder, op->arg1.value, op->arg2.value, op->arg3.value);
+            break;
+        case EC_GF_OP_XORM:
+            gen->xorm(builder, op->arg1.value, op->arg2.value, op->arg3.value);
+            break;
+        default:
+            break;
+        }
+    }
+    gen->epilog(builder);
+
+    return builder->error == 0;
+}
+
+static void *
+ec_code_compile(ec_code_builder_t *builder)
+{
+    ec_code_chunk_t *chunk;
+    void *func;
+
+    if (!ec_code_write(builder)) {
+        return NULL;
+    }
+
+    chunk = ec_code_alloc(builder->code, builder->size);
+    if (chunk == NULL) {
+        return NULL;
+    }
+    func = ec_code_func_from_chunk(chunk);
+    builder->data = (uint8_t *)func;
+
+    if (!ec_code_write(builder)) {
+        ec_code_free(chunk);
+
+        return NULL;
+    }
+
+    GF_FREE(builder);
+
+    return func;
+}
+
+ec_code_t *
+ec_code_create(ec_gf_t *gf, ec_code_gen_t *gen)
+{
+    ec_code_t *code;
+
+    code = GF_MALLOC(sizeof(ec_code_t), ec_mt_ec_code_t);
+    if (code == NULL) {
+        return NULL;
+    }
+    memset(code, 0, sizeof(ec_code_t));
+    INIT_LIST_HEAD(&code->spaces);
+    LOCK_INIT(&code->lock);
+
+    code->gf = gf;
+    code->gen = gen;
+    if (gen == NULL) {
+        code->width = sizeof(uint64_t);
+    } else {
+        code->width = gen->width;
+    }
+
+    return code;
+}
+
+void
+ec_code_destroy(ec_code_t *code)
+{
+    if (!list_empty(&code->spaces)) {
+    }
+
+    LOCK_DESTROY(&code->lock);
+
+    GF_FREE(code);
+}
+
+static uint32_t
+ec_code_value_next(uint32_t *values, uint32_t count, uint32_t *offset)
+{
+    uint32_t i, next;
+
+    next = 0;
+    for (i = *offset + 1; i < count; i++) {
+        next = values[i];
+        if (next != 0) {
+            break;
+        }
+    }
+    *offset = i;
+
+    return next;
+}
+
+void *
+ec_code_build(ec_code_t *code, uint32_t width, uint32_t *values,
+              uint32_t count, gf_boolean_t linear)
+{
+    ec_code_builder_t *builder;
+    uint32_t offset, val, next;
+
+    if (code->gen == NULL) {
+        ec_code_c_prepare(code->gf, values, count);
+        if (linear) {
+            return ec_code_c_linear;
+        } else {
+            return ec_code_c_interleaved;
+        }
+    }
+
+    builder = ec_code_prepare(code, count, width, linear);
+    if (builder == NULL) {
+        return NULL;
+    }
+
+    offset = -1;
+    next = ec_code_value_next(values, count, &offset);
+    if (next != 0) {
+        ec_code_gf_load(builder, offset);
+        do {
+            val = next;
+            next = ec_code_value_next(values, count, &offset);
+            if (next != 0) {
+                ec_code_gf_mul(builder, ec_gf_div(code->gf, val, next));
+                ec_code_gf_load_xor(builder, offset);
+            }
+        } while (next != 0);
+        ec_code_gf_mul(builder, val);
+        ec_code_gf_store(builder);
+    } else {
+        ec_code_gf_clear(builder);
+    }
+
+    return ec_code_compile(builder);
+}
+
+ec_code_func_linear_t
+ec_code_build_linear(ec_code_t *code, uint32_t width, uint32_t *values,
+                     uint32_t count)
+{
+    return (ec_code_func_linear_t)ec_code_build(code, width, values, count,
+                                                _gf_true);
+}
+
+ec_code_func_interleaved_t
+ec_code_build_interleaved(ec_code_t *code, uint32_t width, uint32_t *values,
+                          uint32_t count)
+{
+    return (ec_code_func_interleaved_t)ec_code_build(code, width, values,
+                                                     count, _gf_false);
+}
+
+void
+ec_code_release(ec_code_t *code, ec_code_func_t *func)
+{
+    if (code->gen != NULL) {
+        ec_code_free(ec_code_chunk_from_func(func->linear));
+    }
+}
+
+void
+ec_code_error(ec_code_builder_t *builder, int32_t error)
+{
+    if (builder->error == 0) {
+        builder->error = error;
+    }
+}
+
+void
+ec_code_emit(ec_code_builder_t *builder, uint8_t *bytes, uint32_t count)
+{
+    if (builder->error != 0) {
+        return;
+    }
+
+    if (builder->data != NULL) {
+        memcpy(builder->data + builder->size, bytes, count);
+    }
+
+    builder->size += count;
+    builder->address += count;
+}
+
+static char *
+ec_code_proc_trim_left(char *text, ssize_t *length)
+{
+    ssize_t len;
+
+    for (len = *length; (len > 0) && isspace(*text); len--) {
+        text++;
+    }
+    *length = len;
+
+    return text;
+}
+
+static char *
+ec_code_proc_trim_right(char *text, ssize_t *length, char sep)
+{
+    char *last;
+    ssize_t len;
+
+    len = *length;
+
+    last = text;
+    for (len = *length; (len > 0) && (*text != sep); len--) {
+        if (!isspace(*text)) {
+            last = text + 1;
+        }
+        text++;
+    }
+    *last = 0;
+    *length = len;
+
+    return text;
+}
+
+static char *
+ec_code_proc_line_parse(ec_code_proc_t *file, ssize_t *length)
+{
+    char *text, *end;
+    ssize_t len;
+
+    len = file->size - file->pos;
+    text = ec_code_proc_trim_left(file->buffer + file->pos, &len);
+    end = ec_code_proc_trim_right(text, &len, '\n');
+    if (len == 0) {
+        if (!file->eof) {
+            if (text == file->buffer) {
+                file->size = file->pos = 0;
+                file->skip = _gf_true;
+            } else {
+                file->size = file->pos = end - text;
+                memmove(file->buffer, text, file->pos + 1);
+            }
+            len = sys_read(file->fd, file->buffer + file->pos,
+                           sizeof(file->buffer) - file->pos - 1);
+            if (len > 0) {
+                file->size += len;
+            }
+            file->error = len < 0;
+            file->eof = len <= 0;
+
+            return NULL;
+        }
+        file->size = file->pos = 0;
+    } else {
+        file->pos = end - file->buffer + 1;
+    }
+
+    *length = end - text;
+
+    if (file->skip) {
+        file->skip = _gf_false;
+        text = NULL;
+    }
+
+    return text;
+}
+
+static char *
+ec_code_proc_line(ec_code_proc_t *file, ssize_t *length)
+{
+    char *text;
+
+    text = NULL;
+    while (!file->eof) {
+        text = ec_code_proc_line_parse(file, length);
+        if (text != NULL) {
+            break;
+        }
+    }
+
+    return text;
+}
+
+static char *
+ec_code_proc_split(char *text, ssize_t *length, char sep)
+{
+    text = ec_code_proc_trim_right(text, length, sep);
+    if (*length == 0) {
+        return NULL;
+    }
+    (*length)--;
+    text++;
+
+    return ec_code_proc_trim_left(text, length);
+}
+
+static uint32_t
+ec_code_cpu_check(uint32_t idx, char *list, uint32_t count)
+{
+    ec_code_gen_t *gen;
+    char **ptr;
+    char *table[count];
+    uint32_t i;
+
+    for (i = 0; i < count; i++) {
+        table[i] = list;
+        list += strlen(list) + 1;
+    }
+
+    gen = ec_code_gen_table[idx];
+    while (gen != NULL) {
+        for (ptr = gen->flags; *ptr != NULL; ptr++) {
+            for (i = 0; i < count; i++) {
+                if (strcmp(*ptr, table[i]) == 0) {
+                    break;
+                }
+            }
+            if (i >= count) {
+                gen = ec_code_gen_table[++idx];
+                break;
+            }
+        }
+        if (*ptr == NULL) {
+            break;
+        }
+    }
+
+    return idx;
+}
+
+ec_code_gen_t *
+ec_code_detect(xlator_t *xl, const char *def)
+{
+    ec_code_proc_t file;
+    ec_code_gen_t *gen = NULL;
+    char *line, *data, *list;
+    ssize_t length;
+    uint32_t count, base, select;
+
+    if (strcmp(def, "none") == 0) {
+        gf_msg(xl->name, GF_LOG_INFO, 0, EC_MSG_EXTENSION_NONE,
+               "Not using any cpu extensions");
+
+        return NULL;
+    }
+
+    file.fd = sys_openat(AT_FDCWD, PROC_CPUINFO, O_RDONLY);
+    if (file.fd < 0) {
+        goto out;
+    }
+    file.size = file.pos = 0;
+    file.eof = file.error = file.skip = _gf_false;
+
+    select = 0;
+    if (strcmp(def, "auto") != 0) {
+        while (ec_code_gen_table[select] != NULL) {
+            if (strcmp(ec_code_gen_table[select]->name, def) == 0) {
+                break;
+            }
+            select++;
+        }
+        if (ec_code_gen_table[select] == NULL) {
+            gf_msg(xl->name, GF_LOG_WARNING, EINVAL, EC_MSG_EXTENSION_UNKNOWN,
+                   "CPU extension '%s' is not known. Not using any cpu "
+                   "extensions", def);
+
+            return NULL;
+        }
+    } else {
+        def = NULL;
+    }
+
+    while ((line = ec_code_proc_line(&file, &length)) != NULL) {
+        data = ec_code_proc_split(line, &length, ':');
+        if ((data != NULL) && (strcmp(line, "flags") == 0)) {
+            list = data;
+            count = 0;
+            while ((data != NULL) && (*data != 0)) {
+                count++;
+                data = ec_code_proc_split(data, &length, ' ');
+            }
+            base = select;
+            select = ec_code_cpu_check(select, list, count);
+            if ((base != select) && (def != NULL)) {
+                gf_msg(xl->name, GF_LOG_WARNING, ENOTSUP,
+                       EC_MSG_EXTENSION_UNSUPPORTED,
+                       "CPU extension '%s' is not supported", def);
+                def = NULL;
+            }
+        }
+    }
+
+    if (file.error) {
+        gf_msg(xl->name, GF_LOG_WARNING, 0, EC_MSG_EXTENSION_FAILED,
+               "Unable to detemine supported CPU extensions. Not using any "
+               "cpu extensions");
+
+        gen = NULL;
+    } else {
+        gen = ec_code_gen_table[select];
+        if (gen == NULL) {
+            gf_msg(xl->name, GF_LOG_INFO, 0, EC_MSG_EXTENSION_NONE,
+                   "Not using any cpu extensions");
+        } else {
+            gf_msg(xl->name, GF_LOG_INFO, 0, EC_MSG_EXTENSION,
+                   "Using '%s' CPU extensions", gen->name);
+        }
+    }
+
+    sys_close(file.fd);
+
+out:
+    return gen;
+}
diff --git a/xlators/cluster/ec/src/ec-code.h b/xlators/cluster/ec/src/ec-code.h
new file mode 100644
index 00000000000..355209c3944
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-code.h
@@ -0,0 +1,44 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#ifndef __EC_CODE_H__
+#define __EC_CODE_H__
+
+#include "xlator.h"
+#include "list.h"
+
+#include "ec-types.h"
+#include "ec-galois.h"
+
+ec_code_gen_t *
+ec_code_detect(xlator_t *xl, const char *def);
+
+ec_code_t *
+ec_code_create(ec_gf_t *gf, ec_code_gen_t *gen);
+
+void
+ec_code_destroy(ec_code_t *code);
+
+ec_code_func_linear_t
+ec_code_build_linear(ec_code_t *code, uint32_t width, uint32_t *values,
+                     uint32_t count);
+ec_code_func_interleaved_t
+ec_code_build_interleaved(ec_code_t *code, uint32_t width, uint32_t *values,
+                          uint32_t count);
+void
+ec_code_release(ec_code_t *code, ec_code_func_t *func);
+
+void
+ec_code_error(ec_code_builder_t *builder, int32_t error);
+
+void
+ec_code_emit(ec_code_builder_t *builder, uint8_t *bytes, uint32_t count);
+
+#endif /* __EC_CODE_H__ */
diff --git a/xlators/cluster/ec/src/ec-combine.c b/xlators/cluster/ec/src/ec-combine.c
index 6e4b975c248..f949dbd0c9f 100644
--- a/xlators/cluster/ec/src/ec-combine.c
+++ b/xlators/cluster/ec/src/ec-combine.c
@@ -13,7 +13,7 @@
 #include "libxlator.h"
 #include "byte-order.h"
 
-#include "ec-data.h"
+#include "ec-types.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
diff --git a/xlators/cluster/ec/src/ec-common.c b/xlators/cluster/ec/src/ec-common.c
index 2e6759a2803..fd6bdf7bb9d 100644
--- a/xlators/cluster/ec/src/ec-common.c
+++ b/xlators/cluster/ec/src/ec-common.c
@@ -12,7 +12,7 @@
 #include "hashfn.h"
 
 #include "ec-mem-types.h"
-#include "ec-data.h"
+#include "ec-types.h"
 #include "ec-helpers.h"
 #include "ec-combine.h"
 #include "ec-common.h"
diff --git a/xlators/cluster/ec/src/ec-data.h b/xlators/cluster/ec/src/ec-data.h
index 4a2a11f4ccd..965bc1e25ef 100644
--- a/xlators/cluster/ec/src/ec-data.h
+++ b/xlators/cluster/ec/src/ec-data.h
@@ -11,312 +11,7 @@
 #ifndef __EC_DATA_H__
 #define __EC_DATA_H__
 
-#include "xlator.h"
-
-#include "ec.h"
-
-struct _ec_config;
-typedef struct _ec_config ec_config_t;
-
-struct _ec_fd;
-typedef struct _ec_fd ec_fd_t;
-
-struct _ec_inode;
-typedef struct _ec_inode ec_inode_t;
-
-union _ec_cbk;
-typedef union _ec_cbk ec_cbk_t;
-
-struct _ec_lock;
-typedef struct _ec_lock ec_lock_t;
-
-struct _ec_lock_link;
-typedef struct _ec_lock_link ec_lock_link_t;
-
-struct _ec_fop_data;
-typedef struct _ec_fop_data ec_fop_data_t;
-
-struct _ec_cbk_data;
-typedef struct _ec_cbk_data ec_cbk_data_t;
-
-struct _ec_heal;
-typedef struct _ec_heal ec_heal_t;
-
-typedef void (* ec_wind_f)(ec_t *, ec_fop_data_t *, int32_t);
-typedef int32_t (* ec_handler_f)(ec_fop_data_t *, int32_t);
-typedef void (* ec_resume_f)(ec_fop_data_t *, int32_t);
-
-struct _ec_config
-{
-    uint32_t version;
-    uint8_t  algorithm;
-    uint8_t  gf_word_size;
-    uint8_t  bricks;
-    uint8_t  redundancy;
-    uint32_t chunk_size;
-};
-
-struct _ec_fd
-{
-    loc_t     loc;
-    uintptr_t open;
-    int32_t   flags;
-};
-
-struct _ec_inode
-{
-    ec_lock_t        *inode_lock;
-    gf_boolean_t      have_info;
-    gf_boolean_t      have_config;
-    gf_boolean_t      have_version;
-    gf_boolean_t      have_size;
-    ec_config_t       config;
-    uint64_t          pre_version[2];
-    uint64_t          post_version[2];
-    uint64_t          pre_size;
-    uint64_t          post_size;
-    uint64_t          dirty[2];
-    struct list_head  heal;
-};
-
-typedef int32_t (* fop_heal_cbk_t)(call_frame_t *, void * cookie, xlator_t *,
-                                   int32_t, int32_t, uintptr_t, uintptr_t,
-                                   uintptr_t, dict_t *);
-typedef int32_t (* fop_fheal_cbk_t)(call_frame_t *, void * cookie, xlator_t *,
-                                    int32_t, int32_t, uintptr_t, uintptr_t,
-                                    uintptr_t, dict_t *);
-
-union _ec_cbk
-{
-    fop_access_cbk_t       access;
-    fop_create_cbk_t       create;
-    fop_discard_cbk_t      discard;
-    fop_entrylk_cbk_t      entrylk;
-    fop_fentrylk_cbk_t     fentrylk;
-    fop_fallocate_cbk_t    fallocate;
-    fop_flush_cbk_t        flush;
-    fop_fsync_cbk_t        fsync;
-    fop_fsyncdir_cbk_t     fsyncdir;
-    fop_getxattr_cbk_t     getxattr;
-    fop_fgetxattr_cbk_t    fgetxattr;
-    fop_heal_cbk_t         heal;
-    fop_fheal_cbk_t        fheal;
-    fop_inodelk_cbk_t      inodelk;
-    fop_finodelk_cbk_t     finodelk;
-    fop_link_cbk_t         link;
-    fop_lk_cbk_t           lk;
-    fop_lookup_cbk_t       lookup;
-    fop_mkdir_cbk_t        mkdir;
-    fop_mknod_cbk_t        mknod;
-    fop_open_cbk_t         open;
-    fop_opendir_cbk_t      opendir;
-    fop_readdir_cbk_t      readdir;
-    fop_readdirp_cbk_t     readdirp;
-    fop_readlink_cbk_t     readlink;
-    fop_readv_cbk_t        readv;
-    fop_removexattr_cbk_t  removexattr;
-    fop_fremovexattr_cbk_t fremovexattr;
-    fop_rename_cbk_t       rename;
-    fop_rmdir_cbk_t        rmdir;
-    fop_setattr_cbk_t      setattr;
-    fop_fsetattr_cbk_t     fsetattr;
-    fop_setxattr_cbk_t     setxattr;
-    fop_fsetxattr_cbk_t    fsetxattr;
-    fop_stat_cbk_t         stat;
-    fop_fstat_cbk_t        fstat;
-    fop_statfs_cbk_t       statfs;
-    fop_symlink_cbk_t      symlink;
-    fop_truncate_cbk_t     truncate;
-    fop_ftruncate_cbk_t    ftruncate;
-    fop_unlink_cbk_t       unlink;
-    fop_writev_cbk_t       writev;
-    fop_xattrop_cbk_t      xattrop;
-    fop_fxattrop_cbk_t     fxattrop;
-    fop_zerofill_cbk_t     zerofill;
-    fop_seek_cbk_t         seek;
-};
-
-struct _ec_lock
-{
-    ec_inode_t        *ctx;
-    gf_timer_t        *timer;
-
-    /* List of owners of this lock. All fops added to this list are running
-     * concurrently. */
-    struct list_head   owners;
-
-    /* List of fops waiting to be an owner of the lock. Fops are added to this
-     * list when the current owner has an incompatible access (shared vs
-     * exclusive) or the lock is not acquired yet. */
-    struct list_head   waiting;
-
-    /* List of fops that will wait until the next unlock/lock cycle. This
-     * happens when the currently acquired lock is decided to be released as
-     * soon as possible. In this case, all frozen fops will be continued only
-     * after the lock is reacquired. */
-    struct list_head   frozen;
-
-    int32_t            exclusive;
-    uintptr_t          mask;
-    uintptr_t          good_mask;
-    uintptr_t          healing;
-    uint32_t           refs_owners;  /* Refs for fops owning the lock */
-    uint32_t           refs_pending; /* Refs assigned to fops being prepared */
-    gf_boolean_t       acquired;
-    gf_boolean_t       getting_size;
-    gf_boolean_t       release;
-    gf_boolean_t       query;
-    fd_t              *fd;
-    loc_t              loc;
-    union
-    {
-        entrylk_type     type;
-        struct gf_flock  flock;
-    };
-};
-
-struct _ec_lock_link
-{
-    ec_lock_t        *lock;
-    ec_fop_data_t    *fop;
-    struct list_head  owner_list;
-    struct list_head  wait_list;
-    gf_boolean_t      update[2];
-    loc_t            *base;
-    uint64_t          size;
-};
-
-struct _ec_fop_data
-{
-    int32_t            id;
-    int32_t            refs;
-    int32_t            state;
-    int32_t            minimum;
-    int32_t            expected;
-    int32_t            winds;
-    int32_t            jobs;
-    int32_t            error;
-    ec_fop_data_t     *parent;
-    xlator_t          *xl;
-    call_frame_t      *req_frame;    /* frame of the calling xlator */
-    call_frame_t      *frame;        /* frame used by this fop */
-    struct list_head   cbk_list;     /* sorted list of groups of answers */
-    struct list_head   answer_list;  /* list of answers */
-    struct list_head   pending_list; /* member of ec_t.pending_fops */
-    ec_cbk_data_t     *answer;       /* accepted answer */
-    int32_t            lock_count;
-    int32_t            locked;
-    ec_lock_link_t     locks[2];
-    int32_t            first_lock;
-    gf_lock_t          lock;
-
-    uint32_t           flags;
-    uint32_t           first;
-    uintptr_t          mask;
-    uintptr_t          healing; /*Dispatch is done but call is successful only
-                                  if fop->minimum number of subvolumes succeed
-                                  which are not healing*/
-    uintptr_t          remaining;
-    uintptr_t          received; /* Mask of responses */
-    uintptr_t          good;
-
-    uid_t              uid;
-    gid_t              gid;
-
-    ec_wind_f          wind;
-    ec_handler_f       handler;
-    ec_resume_f        resume;
-    ec_cbk_t           cbks;
-    void              *data;
-    ec_heal_t         *heal;
-    struct list_head   healer;
-
-    uint64_t           user_size;
-    uint32_t           head;
-
-    int32_t            use_fd;
-
-    dict_t            *xdata;
-    dict_t            *dict;
-    int32_t            int32;
-    uint32_t           uint32;
-    uint64_t           size;
-    off_t              offset;
-    mode_t             mode[2];
-    entrylk_cmd        entrylk_cmd;
-    entrylk_type       entrylk_type;
-    gf_xattrop_flags_t xattrop_flags;
-    dev_t              dev;
-    inode_t           *inode;
-    fd_t              *fd;
-    struct iatt        iatt;
-    char              *str[2];
-    loc_t              loc[2];
-    struct gf_flock    flock;
-    struct iovec      *vector;
-    struct iobref     *buffers;
-    gf_seek_what_t     seek;
-};
-
-struct _ec_cbk_data
-{
-    struct list_head list;        // item in the sorted list of groups
-    struct list_head answer_list; // item in the list of answers
-    ec_fop_data_t *  fop;
-    ec_cbk_data_t *  next;        // next answer in the same group
-    int32_t          idx;
-    int32_t          op_ret;
-    int32_t          op_errno;
-    int32_t          count;
-    uintptr_t        mask;
-    uint64_t         dirty[2];
-
-    dict_t *         xdata;
-    dict_t *         dict;
-    int32_t          int32;
-    uintptr_t        uintptr[3];
-    uint64_t         size;
-    uint64_t         version[2];
-    inode_t *        inode;
-    fd_t *           fd;
-    struct statvfs   statvfs;
-    struct iatt      iatt[5];
-    struct gf_flock  flock;
-    struct iovec *   vector;
-    struct iobref *  buffers;
-    char            *str;
-    gf_dirent_t      entries;
-    off_t            offset;
-    gf_seek_what_t   what;
-};
-
-struct _ec_heal
-{
-    struct list_head  list;
-    gf_lock_t         lock;
-    xlator_t         *xl;
-    ec_fop_data_t    *fop;
-    void             *data;
-    ec_fop_data_t    *lookup;
-    loc_t             loc;
-    struct iatt       iatt;
-    char             *symlink;
-    fd_t             *fd;
-    int32_t           partial;
-    int32_t           done;
-    int32_t           error;
-    gf_boolean_t      nameheal;
-    uintptr_t         available;
-    uintptr_t         good;
-    uintptr_t         bad;
-    uintptr_t         open;
-    uintptr_t         fixed;
-    uint64_t          offset;
-    uint64_t          size;
-    uint64_t          total_size;
-    uint64_t          version[2];
-    uint64_t          raw_size;
-};
+#include "ec-types.h"
 
 ec_cbk_data_t * ec_cbk_data_allocate(call_frame_t * frame, xlator_t * this,
                                      ec_fop_data_t * fop, int32_t id,
@@ -332,4 +27,6 @@ void ec_fop_data_release(ec_fop_data_t * fop);
 
 void ec_fop_cleanup(ec_fop_data_t *fop);
 
+void ec_pending_fops_completed(ec_t *ec);
+
 #endif /* __EC_DATA_H__ */
diff --git a/xlators/cluster/ec/src/ec-dir-read.c b/xlators/cluster/ec/src/ec-dir-read.c
index ed53a0416e0..4fe82e3c0b6 100644
--- a/xlators/cluster/ec/src/ec-dir-read.c
+++ b/xlators/cluster/ec/src/ec-dir-read.c
@@ -11,12 +11,13 @@
 #include "xlator.h"
 #include "defaults.h"
 
+#include "ec.h"
+#include "ec-messages.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
 #include "ec-method.h"
 #include "ec-fops.h"
-#include "ec-messages.h"
 
 /* FOP: opendir */
 
diff --git a/xlators/cluster/ec/src/ec-dir-write.c b/xlators/cluster/ec/src/ec-dir-write.c
index e181170650d..1272e3dfe0d 100644
--- a/xlators/cluster/ec/src/ec-dir-write.c
+++ b/xlators/cluster/ec/src/ec-dir-write.c
@@ -11,12 +11,13 @@
 #include "xlator.h"
 #include "defaults.h"
 
+#include "ec.h"
+#include "ec-messages.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
 #include "ec-method.h"
 #include "ec-fops.h"
-#include "ec-messages.h"
 
 int
 ec_dir_write_cbk (call_frame_t *frame, xlator_t *this,
diff --git a/xlators/cluster/ec/src/ec-fops.h b/xlators/cluster/ec/src/ec-fops.h
index 8d938427a18..13f419b5a30 100644
--- a/xlators/cluster/ec/src/ec-fops.h
+++ b/xlators/cluster/ec/src/ec-fops.h
@@ -13,7 +13,7 @@
 
 #include "xlator.h"
 
-#include "ec-data.h"
+#include "ec-types.h"
 #include "ec-common.h"
 
 void ec_access(call_frame_t * frame, xlator_t * this, uintptr_t target,
diff --git a/xlators/cluster/ec/src/ec-galois.c b/xlators/cluster/ec/src/ec-galois.c
new file mode 100644
index 00000000000..7dbbac09713
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-galois.c
@@ -0,0 +1,185 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#include <string.h>
+
+#include "mem-pool.h"
+#include "list.h"
+
+#include "ec-mem-types.h"
+#include "ec-gf8.h"
+
+static ec_gf_t *
+ec_gf_alloc(uint32_t bits, uint32_t mod)
+{
+    ec_gf_t *gf;
+
+    gf = GF_MALLOC(sizeof(ec_gf_t), ec_mt_ec_gf_t);
+    if (gf == NULL) {
+        goto failed;
+    }
+
+    gf->bits = bits;
+    gf->size = 1 << bits;
+    gf->mod = mod;
+
+    gf->log = GF_MALLOC(sizeof(uint32_t) * (gf->size * 2 - 1),
+                        gf_common_mt_int);
+    if (gf->log == NULL) {
+        goto failed_gf;
+    }
+    gf->pow = GF_MALLOC(sizeof(uint32_t) * (gf->size * 2 - 1),
+                        gf_common_mt_int);
+    if (gf->pow == NULL) {
+        goto failed_log;
+    }
+
+    return gf;
+
+failed_log:
+    GF_FREE(gf->log);
+failed_gf:
+    GF_FREE(gf);
+failed:
+    return NULL;
+}
+
+static void
+ec_gf_init_tables(ec_gf_t *gf)
+{
+    uint32_t i, tmp;
+
+    memset(gf->log, -1, sizeof(uint32_t) * gf->size);
+
+    gf->pow[0] = 1;
+    gf->log[0] = gf->size;
+    gf->log[1] = 0;
+    for (i = 1; i < gf->size; i++) {
+        tmp = gf->pow[i - 1] << 1;
+        if (tmp >= gf->size) {
+            tmp ^= gf->mod;
+        }
+        gf->pow[i + gf->size - 1] = gf->pow[i] = tmp;
+        gf->log[tmp + gf->size - 1] = gf->log[tmp] = i;
+    }
+}
+
+ec_gf_t *
+ec_gf_prepare(uint32_t bits, uint32_t mod)
+{
+    ec_gf_mul_t **tbl;
+    ec_gf_t *gf;
+    uint32_t i, j;
+
+    if (bits != 8) {
+        return NULL;
+    }
+
+    tbl = ec_gf8_mul;
+    if (mod == 0) {
+        mod = 0x11d;
+    }
+
+    gf = ec_gf_alloc(bits, mod);
+    if (gf == NULL) {
+        return NULL;
+    }
+    ec_gf_init_tables(gf);
+
+    gf->table = tbl;
+    gf->min_ops = bits * bits;
+    gf->max_ops = 0;
+    gf->avg_ops = 0;
+    for (i = 1; i < gf->size; i++) {
+        for (j = 0; tbl[i]->ops[j].op != EC_GF_OP_END; j++) {
+        }
+        if (gf->max_ops < j) {
+            gf->max_ops = j;
+        }
+        if (gf->min_ops > j) {
+            gf->min_ops = j;
+        }
+        gf->avg_ops += j;
+    }
+    gf->avg_ops /= gf->size;
+
+    return gf;
+}
+
+void
+ec_gf_destroy(ec_gf_t *gf)
+{
+    GF_FREE(gf->pow);
+    GF_FREE(gf->log);
+    GF_FREE(gf);
+}
+
+uint32_t
+ec_gf_add(ec_gf_t *gf, uint32_t a, uint32_t b)
+{
+    if ((a >= gf->size) || (b >= gf->size)) {
+        return gf->size;
+    }
+
+    return a ^ b;
+}
+
+uint32_t
+ec_gf_mul(ec_gf_t *gf, uint32_t a, uint32_t b)
+{
+    if ((a >= gf->size) || (b >= gf->size)) {
+        return gf->size;
+    }
+
+    if ((a != 0) && (b != 0)) {
+        return gf->pow[gf->log[a] + gf->log[b]];
+    }
+
+    return 0;
+}
+
+uint32_t
+ec_gf_div(ec_gf_t *gf, uint32_t a, uint32_t b)
+{
+    if ((a >= gf->size) || (b >= gf->size)) {
+        return gf->size;
+    }
+
+    if (b != 0) {
+        if (a != 0) {
+            return gf->pow[gf->size - 1 + gf->log[a] - gf->log[b]];
+        }
+
+        return 0;
+    }
+
+    return gf->size;
+}
+
+uint32_t
+ec_gf_exp(ec_gf_t *gf, uint32_t a, uint32_t b)
+{
+    uint32_t r;
+
+    if ((a >= gf->size) || ((a == 0) && (b == 0))) {
+        return gf->size;
+    }
+
+    r = 1;
+    while (b != 0) {
+        if ((b & 1) != 0) {
+            r = ec_gf_mul(gf, r, a);
+        }
+        a = ec_gf_mul(gf, a, a);
+        b >>= 1;
+    }
+
+    return r;
+}
diff --git a/xlators/cluster/ec/src/ec-galois.h b/xlators/cluster/ec/src/ec-galois.h
new file mode 100644
index 00000000000..02e6b6c1bc2
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-galois.h
@@ -0,0 +1,26 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#ifndef __EC_GALOIS_H__
+#define __EC_GALOIS_H__
+
+#include <inttypes.h>
+
+#include "ec-types.h"
+
+ec_gf_t *ec_gf_prepare(uint32_t bits, uint32_t mod);
+void ec_gf_destroy(ec_gf_t *gf);
+
+uint32_t ec_gf_add(ec_gf_t *gf, uint32_t a, uint32_t b);
+uint32_t ec_gf_mul(ec_gf_t *gf, uint32_t a, uint32_t b);
+uint32_t ec_gf_div(ec_gf_t *gf, uint32_t a, uint32_t b);
+uint32_t ec_gf_exp(ec_gf_t *gf, uint32_t a, uint32_t b);
+
+#endif /* __EC_GALOIS_H__ */
diff --git a/xlators/cluster/ec/src/ec-generic.c b/xlators/cluster/ec/src/ec-generic.c
index 0ad514908aa..d67420469a8 100644
--- a/xlators/cluster/ec/src/ec-generic.c
+++ b/xlators/cluster/ec/src/ec-generic.c
@@ -10,14 +10,15 @@
 
 #include "xlator.h"
 #include "defaults.h"
+#include "byte-order.h"
 
+#include "ec.h"
+#include "ec-messages.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
 #include "ec-method.h"
 #include "ec-fops.h"
-#include "ec-messages.h"
-#include "byte-order.h"
 
 /* FOP: flush */
 
diff --git a/xlators/cluster/ec/src/ec-gf.c b/xlators/cluster/ec/src/ec-gf.c
deleted file mode 100644
index 1ae8928f20b..00000000000
--- a/xlators/cluster/ec/src/ec-gf.c
+++ /dev/null
@@ -1,11635 +0,0 @@
-/*
-  Copyright (c) 2012-2014 DataLab, s.l. <http://www.datalab.es>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-#include <inttypes.h>
-#include <string.h>
-
-#include "ec-gf.h"
-
-static void gf8_muladd_00(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    memcpy(out, in, sizeof(uint64_t) * 8 * width);
-}
-
-static void gf8_muladd_01(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        out_ptr[0] ^= in_ptr[0];
-        out_ptr[width] ^= in_ptr[width];
-        out_ptr[width * 2] ^= in_ptr[width * 2];
-        out_ptr[width * 3] ^= in_ptr[width * 3];
-        out_ptr[width * 4] ^= in_ptr[width * 4];
-        out_ptr[width * 5] ^= in_ptr[width * 5];
-        out_ptr[width * 6] ^= in_ptr[width * 6];
-        out_ptr[width * 7] ^= in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_02(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in7;
-        out1 = in0;
-        out7 = in6;
-        out5 = in4;
-        out6 = in5;
-        out3 = in2 ^ in7;
-        out4 = in3 ^ in7;
-        out2 = in1 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_03(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in0 ^ in7;
-        tmp0 = in2 ^ in7;
-        out1 = in0 ^ in1;
-        out7 = in6 ^ in7;
-        out5 = in4 ^ in5;
-        out6 = in5 ^ in6;
-        out4 = in3 ^ in4 ^ in7;
-        out2 = tmp0 ^ in1;
-        out3 = tmp0 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_04(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in6;
-        out1 = in7;
-        out7 = in5;
-        out6 = in4;
-        tmp0 = in6 ^ in7;
-        out2 = in0 ^ in6;
-        out5 = in3 ^ in7;
-        out3 = tmp0 ^ in1;
-        out4 = tmp0 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_05(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in0 ^ in6;
-        out1 = in1 ^ in7;
-        out7 = in5 ^ in7;
-        out6 = in4 ^ in6;
-        out2 = out0 ^ in2;
-        out3 = out1 ^ in3 ^ in6;
-        out5 = out7 ^ in3;
-        out4 = out6 ^ in2 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_06(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in6 ^ in7;
-        tmp0 = in1 ^ in6;
-        out1 = in0 ^ in7;
-        out7 = in5 ^ in6;
-        out6 = in4 ^ in5;
-        out4 = in2 ^ in3 ^ in6;
-        out5 = in3 ^ in4 ^ in7;
-        out3 = tmp0 ^ in2;
-        out2 = tmp0 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_07(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in6;
-        tmp1 = in5 ^ in6;
-        tmp2 = in0 ^ in7;
-        tmp3 = tmp0 ^ in3;
-        out6 = tmp1 ^ in4;
-        out7 = tmp1 ^ in7;
-        out0 = tmp2 ^ in6;
-        out1 = tmp2 ^ in1;
-        out3 = tmp3 ^ in1;
-        out4 = tmp3 ^ in4;
-        out5 = out4 ^ out7 ^ in2;
-        out2 = tmp0 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_08(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in5;
-        out1 = in6;
-        out7 = in4;
-        out6 = in3 ^ in7;
-        out3 = in0 ^ in5 ^ in6;
-        out5 = in2 ^ in6 ^ in7;
-        out2 = in5 ^ in7;
-        out4 = out2 ^ in1 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_09(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in0 ^ in5;
-        tmp0 = in3 ^ in6;
-        out1 = in1 ^ in6;
-        out7 = in4 ^ in7;
-        out2 = in2 ^ in5 ^ in7;
-        out3 = tmp0 ^ out0;
-        out6 = tmp0 ^ in7;
-        out4 = out1 ^ out7 ^ in5;
-        out5 = out2 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_0A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in5 ^ in7;
-        out1 = in0 ^ in6;
-        out7 = in4 ^ in6;
-        out2 = in1 ^ in5;
-        out6 = out0 ^ in3;
-        out3 = out0 ^ out1 ^ in2;
-        out5 = out7 ^ in2 ^ in7;
-        out4 = out2 ^ in3 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_0B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in5;
-        tmp1 = in0 ^ in6;
-        tmp2 = in4 ^ in7;
-        out0 = in0 ^ in5 ^ in7;
-        out2 = tmp0 ^ in1;
-        out1 = tmp1 ^ in1;
-        out6 = tmp1 ^ out0 ^ in3;
-        out7 = tmp2 ^ in6;
-        out4 = tmp2 ^ out6 ^ in1;
-        out3 = out6 ^ in0 ^ in2;
-        out5 = tmp0 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_0C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in5 ^ in6;
-        out1 = in6 ^ in7;
-        out7 = in4 ^ in5;
-        tmp0 = in1 ^ in5;
-        tmp1 = in0 ^ in7;
-        out5 = in2 ^ in3 ^ in6;
-        out6 = in3 ^ in4 ^ in7;
-        out2 = tmp1 ^ out0;
-        out4 = tmp0 ^ in2;
-        out3 = tmp0 ^ tmp1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_0D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in4 ^ in5;
-        tmp1 = in5 ^ in6;
-        out1 = in1 ^ in6 ^ in7;
-        out7 = tmp0 ^ in7;
-        out4 = tmp0 ^ in1 ^ in2;
-        out0 = tmp1 ^ in0;
-        tmp2 = tmp1 ^ in3;
-        out6 = tmp2 ^ out7;
-        out2 = out0 ^ in2 ^ in7;
-        out3 = out0 ^ out1 ^ in3;
-        out5 = tmp2 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_0E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        tmp1 = in2 ^ in5;
-        tmp2 = in5 ^ in6;
-        out1 = in0 ^ in6 ^ in7;
-        out3 = tmp0 ^ tmp1;
-        out2 = tmp0 ^ tmp2;
-        tmp3 = tmp1 ^ in3;
-        out7 = tmp2 ^ in4;
-        out0 = tmp2 ^ in7;
-        out4 = tmp3 ^ in1 ^ in7;
-        out5 = tmp3 ^ out7;
-        out6 = out0 ^ out5 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_0F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in6 ^ in7;
-        tmp1 = tmp0 ^ in1;
-        tmp2 = tmp0 ^ in5;
-        out1 = tmp1 ^ in0;
-        out7 = tmp2 ^ in4;
-        out0 = tmp2 ^ in0;
-        out6 = out7 ^ in3;
-        out5 = out6 ^ in2 ^ in7;
-        tmp3 = tmp1 ^ out0 ^ in2;
-        out4 = tmp1 ^ out5;
-        out2 = tmp3 ^ in6;
-        out3 = tmp3 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_10(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in4;
-        out1 = in5;
-        out7 = in3 ^ in7;
-        tmp0 = in6 ^ in7;
-        out2 = in4 ^ in6;
-        tmp1 = out2 ^ in5;
-        out6 = tmp0 ^ in2;
-        out3 = tmp0 ^ tmp1;
-        out5 = out2 ^ out3 ^ in1;
-        out4 = tmp1 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_11(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out7 = in3;
-        out0 = in0 ^ in4;
-        out1 = in1 ^ in5;
-        out6 = in2 ^ in7;
-        out4 = in0 ^ in5 ^ in6;
-        out5 = in1 ^ in6 ^ in7;
-        out2 = in2 ^ in4 ^ in6;
-        out3 = in3 ^ in4 ^ in5 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_12(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in4 ^ in7;
-        out1 = in0 ^ in5;
-        out3 = in2 ^ in4 ^ in5;
-        tmp0 = out0 ^ in6;
-        out2 = tmp0 ^ in1;
-        tmp1 = tmp0 ^ in3;
-        out6 = tmp0 ^ out3;
-        out5 = out2 ^ in5;
-        out7 = tmp1 ^ in4;
-        out4 = tmp1 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_13(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out7 = in3 ^ in6;
-        tmp0 = in0 ^ in5;
-        tmp1 = in4 ^ in7;
-        out6 = in2 ^ in5 ^ in7;
-        out4 = tmp0 ^ out7 ^ in7;
-        out1 = tmp0 ^ in1;
-        out0 = tmp1 ^ in0;
-        out5 = tmp1 ^ in1 ^ in6;
-        out3 = tmp1 ^ out6 ^ in3;
-        out2 = out5 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_14(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in4 ^ in6;
-        out1 = in5 ^ in7;
-        out2 = in0 ^ in4;
-        tmp0 = out0 ^ in5;
-        out7 = out1 ^ in3;
-        tmp1 = out1 ^ in2;
-        out3 = tmp0 ^ in1;
-        out6 = tmp0 ^ tmp1;
-        out4 = tmp1 ^ out2;
-        out5 = out3 ^ in3 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_15(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out7 = in3 ^ in5;
-        tmp0 = in0 ^ in4;
-        out1 = in1 ^ in5 ^ in7;
-        out5 = in1 ^ in3 ^ in6;
-        out0 = tmp0 ^ in6;
-        out2 = tmp0 ^ in2;
-        out3 = out5 ^ in4 ^ in5;
-        out6 = out2 ^ in0 ^ in7;
-        out4 = tmp0 ^ out6 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_16(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in5;
-        tmp1 = in4 ^ in7;
-        tmp2 = in2 ^ in3 ^ in4;
-        out1 = tmp0 ^ in7;
-        out4 = tmp0 ^ tmp2;
-        out0 = tmp1 ^ in6;
-        tmp3 = tmp1 ^ in1;
-        out6 = out0 ^ in2 ^ in5;
-        out2 = tmp3 ^ in0;
-        out3 = out6 ^ in1;
-        out7 = tmp2 ^ out6;
-        out5 = tmp3 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_17(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in5;
-        tmp1 = in3 ^ in6;
-        tmp2 = tmp0 ^ in4;
-        out4 = tmp0 ^ in0 ^ in3;
-        out7 = tmp1 ^ in5;
-        tmp3 = tmp1 ^ in1;
-        out6 = tmp2 ^ in7;
-        out5 = tmp3 ^ in4;
-        out3 = tmp3 ^ out6;
-        out0 = out3 ^ out4 ^ in1;
-        out2 = out3 ^ out7 ^ in0;
-        out1 = tmp2 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_18(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in4 ^ in5;
-        out1 = in5 ^ in6;
-        tmp0 = in4 ^ in7;
-        out5 = in1 ^ in2 ^ in5;
-        out6 = in2 ^ in3 ^ in6;
-        out2 = tmp0 ^ out1;
-        out7 = tmp0 ^ in3;
-        tmp1 = tmp0 ^ in0;
-        out3 = tmp1 ^ in6;
-        out4 = tmp1 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_19(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out5 = in1 ^ in2;
-        out7 = in3 ^ in4;
-        tmp0 = in0 ^ in7;
-        out6 = in2 ^ in3;
-        out1 = in1 ^ in5 ^ in6;
-        out0 = in0 ^ in4 ^ in5;
-        out4 = tmp0 ^ in1;
-        tmp1 = tmp0 ^ in6;
-        out2 = tmp1 ^ out0 ^ in2;
-        out3 = tmp1 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_1A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in4 ^ in5;
-        tmp1 = in5 ^ in6;
-        tmp2 = tmp0 ^ in1;
-        out0 = tmp0 ^ in7;
-        out1 = tmp1 ^ in0;
-        tmp3 = tmp1 ^ in3;
-        out5 = tmp2 ^ in2;
-        out2 = tmp2 ^ in6;
-        out7 = tmp3 ^ out0;
-        out6 = tmp3 ^ in2;
-        out4 = tmp3 ^ out2 ^ in0;
-        out3 = tmp0 ^ out1 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_1B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in4;
-        tmp1 = in2 ^ in5;
-        tmp2 = in3 ^ in6;
-        out5 = tmp0 ^ in1;
-        tmp3 = tmp0 ^ in0;
-        out6 = tmp1 ^ in3;
-        out0 = tmp1 ^ tmp3 ^ in7;
-        out7 = tmp2 ^ in4;
-        tmp4 = out5 ^ in6;
-        out3 = tmp2 ^ tmp3;
-        out2 = tmp4 ^ in5;
-        out4 = tmp4 ^ out3;
-        out1 = tmp3 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_1C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in3;
-        tmp1 = in4 ^ in6;
-        tmp2 = in5 ^ in7;
-        out6 = tmp0 ^ tmp1;
-        out0 = tmp1 ^ in5;
-        out1 = tmp2 ^ in6;
-        tmp3 = tmp2 ^ in1;
-        tmp4 = tmp2 ^ in4;
-        out2 = tmp4 ^ in0;
-        out7 = tmp4 ^ in3;
-        out5 = tmp0 ^ tmp3;
-        out3 = tmp3 ^ out2;
-        out4 = out3 ^ in2 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_1D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in3;
-        tmp1 = in0 ^ in4;
-        tmp2 = in3 ^ in4;
-        tmp3 = in2 ^ in7;
-        out3 = tmp0 ^ tmp1;
-        out5 = tmp0 ^ tmp3;
-        tmp4 = tmp1 ^ in5;
-        out6 = tmp2 ^ in2;
-        out7 = tmp2 ^ in5;
-        out2 = tmp3 ^ tmp4;
-        out4 = out3 ^ out6 ^ in6;
-        out0 = tmp4 ^ in6;
-        out1 = out2 ^ out4 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_1E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in4;
-        tmp1 = in2 ^ in7;
-        tmp2 = tmp0 ^ in1;
-        out3 = tmp1 ^ tmp2;
-        out2 = tmp2 ^ in5;
-        out4 = out3 ^ in3 ^ in6;
-        tmp3 = out4 ^ in7;
-        out6 = tmp3 ^ out2 ^ in4;
-        out7 = tmp1 ^ out6;
-        out0 = out7 ^ in3;
-        out1 = tmp0 ^ out0;
-        out5 = tmp3 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_1F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in4 ^ in6;
-        tmp1 = tmp0 ^ in5;
-        out7 = tmp1 ^ in3;
-        out0 = tmp1 ^ in0 ^ in7;
-        out6 = out7 ^ in2 ^ in6;
-        out1 = out0 ^ in1 ^ in4;
-        out4 = out0 ^ out6 ^ in1;
-        out3 = tmp0 ^ out4;
-        out2 = out4 ^ out7 ^ in7;
-        out5 = out3 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_20(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in4;
-        out0 = in3 ^ in7;
-        tmp0 = in3 ^ in4;
-        tmp1 = in6 ^ in7;
-        out2 = out0 ^ in5;
-        out4 = tmp0 ^ in5;
-        out3 = tmp0 ^ tmp1;
-        out7 = tmp1 ^ in2;
-        out6 = tmp1 ^ in1 ^ in5;
-        out5 = out2 ^ out3 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_21(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in1 ^ in4;
-        tmp0 = in4 ^ in6;
-        out4 = in3 ^ in5;
-        out7 = in2 ^ in6;
-        out0 = in0 ^ in3 ^ in7;
-        out6 = in1 ^ in5 ^ in7;
-        out3 = tmp0 ^ in7;
-        out5 = tmp0 ^ in0;
-        out2 = out4 ^ in2 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_22(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in3;
-        out1 = in0 ^ in4;
-        out7 = in2 ^ in7;
-        out4 = in4 ^ in5 ^ in7;
-        out5 = in0 ^ in5 ^ in6;
-        out6 = in1 ^ in6 ^ in7;
-        out3 = in2 ^ in3 ^ in4 ^ in6;
-        out2 = in1 ^ in3 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_23(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out7 = in2;
-        out0 = in0 ^ in3;
-        out4 = in5 ^ in7;
-        out5 = in0 ^ in6;
-        out6 = in1 ^ in7;
-        out3 = in2 ^ in4 ^ in6;
-        out1 = in0 ^ in1 ^ in4;
-        out2 = out4 ^ out6 ^ in2 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_24(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in4 ^ in7;
-        tmp0 = in3 ^ in4;
-        out0 = in3 ^ in6 ^ in7;
-        out3 = tmp0 ^ in1;
-        tmp1 = out0 ^ in5;
-        out6 = tmp1 ^ out3;
-        out2 = tmp1 ^ in0;
-        out7 = tmp1 ^ in2 ^ in3;
-        out5 = out2 ^ in4;
-        out4 = tmp0 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_25(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in1 ^ in4;
-        tmp0 = in2 ^ in5;
-        out1 = out3 ^ in7;
-        out7 = tmp0 ^ in6;
-        out6 = out1 ^ in5;
-        out4 = out7 ^ in3 ^ in7;
-        out2 = out4 ^ in0;
-        out0 = tmp0 ^ out2;
-        out5 = out0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_26(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in3 ^ in6;
-        tmp0 = in4 ^ in7;
-        out7 = in2 ^ in5 ^ in7;
-        tmp1 = out0 ^ in0 ^ in5;
-        out1 = tmp0 ^ in0;
-        tmp2 = tmp0 ^ in6;
-        out2 = tmp1 ^ in1;
-        out5 = tmp1 ^ in7;
-        out6 = tmp2 ^ in1;
-        out4 = tmp2 ^ out7;
-        out3 = out0 ^ out6 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_27(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out7 = in2 ^ in5;
-        out0 = in0 ^ in3 ^ in6;
-        out6 = in1 ^ in4 ^ in7;
-        out4 = out7 ^ in6;
-        out2 = out0 ^ out7 ^ in1;
-        out5 = out0 ^ in7;
-        out1 = out6 ^ in0;
-        out3 = out6 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_28(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in3;
-        out1 = in4 ^ in6;
-        out0 = in3 ^ in5 ^ in7;
-        tmp0 = out1 ^ in7;
-        tmp1 = out0 ^ in4;
-        out7 = tmp0 ^ in2;
-        tmp2 = tmp0 ^ in1;
-        out3 = tmp1 ^ in0;
-        out6 = tmp1 ^ tmp2;
-        out4 = tmp2 ^ in3;
-        out5 = out3 ^ in2 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_29(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in2 ^ in3;
-        tmp0 = in1 ^ in3;
-        tmp1 = in4 ^ in6;
-        tmp2 = in0 ^ in4 ^ in7;
-        out6 = tmp0 ^ in5;
-        out4 = tmp0 ^ in6 ^ in7;
-        out1 = tmp1 ^ in1;
-        out7 = tmp1 ^ in2;
-        out3 = tmp2 ^ in5;
-        out5 = tmp2 ^ in2;
-        out0 = out3 ^ in3 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_2A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in3 ^ in5;
-        tmp0 = in1 ^ in3;
-        tmp1 = in0 ^ in4;
-        out7 = in2 ^ in4 ^ in7;
-        out3 = tmp1 ^ out0 ^ in2;
-        out2 = tmp0 ^ in7;
-        out6 = tmp0 ^ in6;
-        out1 = tmp1 ^ in6;
-        out5 = tmp1 ^ out7 ^ in5;
-        out4 = out1 ^ in0 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_2B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in1 ^ in6;
-        out7 = in2 ^ in4;
-        tmp0 = in0 ^ in5;
-        tmp1 = in2 ^ in7;
-        out6 = in1 ^ in3;
-        out1 = out4 ^ in0 ^ in4;
-        out3 = tmp0 ^ out7;
-        out0 = tmp0 ^ in3;
-        out5 = tmp1 ^ in0;
-        out2 = tmp1 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_2C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in5;
-        tmp1 = in2 ^ in3 ^ in4;
-        tmp2 = tmp0 ^ in6;
-        out4 = tmp1 ^ in1;
-        out5 = tmp1 ^ in0 ^ in5;
-        tmp3 = tmp2 ^ in4;
-        out6 = tmp2 ^ out4;
-        out7 = tmp3 ^ in7;
-        out2 = tmp3 ^ out5;
-        out3 = out6 ^ in0;
-        out0 = tmp1 ^ out7;
-        out1 = tmp0 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_2D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in3;
-        out4 = tmp0 ^ in1;
-        tmp1 = tmp0 ^ in0;
-        out2 = tmp1 ^ in6;
-        out5 = tmp1 ^ in4;
-        tmp2 = out2 ^ in2;
-        tmp3 = tmp2 ^ in5;
-        out0 = tmp3 ^ in7;
-        out7 = tmp3 ^ out5;
-        out6 = out4 ^ out7 ^ in6;
-        out3 = tmp2 ^ out6;
-        out1 = out0 ^ out6 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_2E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in4 ^ in7;
-        out0 = in3 ^ in5 ^ in6;
-        tmp1 = tmp0 ^ in0;
-        tmp2 = tmp0 ^ in2;
-        out1 = tmp1 ^ in6;
-        out4 = tmp2 ^ in1;
-        out7 = tmp2 ^ in5;
-        out3 = out0 ^ out4 ^ in0;
-        out2 = out3 ^ out7 ^ in7;
-        out6 = tmp1 ^ out2;
-        out5 = tmp1 ^ out7 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_2F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3;
-        tmp1 = in2 ^ in5;
-        out4 = in1 ^ in2 ^ in7;
-        out6 = in1 ^ in3 ^ in4;
-        out5 = tmp0 ^ in2;
-        tmp2 = tmp0 ^ in6;
-        out7 = tmp1 ^ in4;
-        out0 = tmp2 ^ in5;
-        out2 = tmp2 ^ out4;
-        out1 = tmp2 ^ out6 ^ in7;
-        out3 = tmp1 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_30(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in4 ^ in5;
-        tmp0 = in3 ^ in6;
-        tmp1 = in4 ^ in7;
-        out6 = in1 ^ in2 ^ in5;
-        out3 = tmp0 ^ in5;
-        out4 = tmp0 ^ in0;
-        out7 = tmp0 ^ in2;
-        out0 = tmp1 ^ in3;
-        out2 = tmp1 ^ out3;
-        out5 = tmp1 ^ in0 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_31(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in5 ^ in6;
-        tmp0 = in4 ^ in5;
-        tmp1 = in0 ^ in3 ^ in4;
-        tmp2 = out3 ^ in2;
-        out1 = tmp0 ^ in1;
-        out0 = tmp1 ^ in7;
-        out4 = tmp1 ^ in6;
-        out6 = tmp2 ^ in1;
-        out2 = tmp2 ^ out0 ^ in0;
-        out5 = out1 ^ in0 ^ in7;
-        out7 = tmp0 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_32(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in3 ^ in4;
-        out7 = in2 ^ in3;
-        tmp0 = in5 ^ in6;
-        tmp1 = in0 ^ in7;
-        out6 = in1 ^ in2;
-        out1 = in0 ^ in4 ^ in5;
-        out2 = tmp0 ^ out0 ^ in1;
-        out3 = tmp0 ^ out7 ^ in7;
-        out4 = tmp1 ^ in6;
-        out5 = tmp1 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_33(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in3;
-        tmp1 = in0 ^ in4;
-        tmp2 = in1 ^ in5;
-        out6 = in1 ^ in2 ^ in6;
-        out7 = tmp0 ^ in7;
-        out0 = tmp1 ^ in3;
-        out1 = tmp1 ^ tmp2;
-        tmp3 = tmp2 ^ in7;
-        tmp4 = tmp2 ^ in4 ^ in6;
-        out5 = tmp3 ^ in0;
-        out3 = tmp3 ^ out6;
-        out4 = tmp4 ^ out5;
-        out2 = tmp0 ^ tmp4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_34(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in4;
-        tmp1 = in4 ^ in5;
-        tmp2 = tmp0 ^ in1;
-        tmp3 = tmp0 ^ in6;
-        out1 = tmp1 ^ in7;
-        tmp4 = tmp1 ^ in2;
-        out5 = tmp2 ^ in0;
-        out3 = tmp2 ^ out1;
-        out0 = tmp3 ^ in7;
-        out7 = tmp3 ^ tmp4;
-        out6 = tmp4 ^ in1;
-        out2 = out3 ^ out5 ^ in3;
-        out4 = tmp4 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_35(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in6;
-        tmp1 = in5 ^ in7;
-        out7 = tmp0 ^ tmp1 ^ in3;
-        out3 = tmp1 ^ in1;
-        out1 = out3 ^ in4;
-        tmp2 = out1 ^ in7;
-        out5 = tmp2 ^ in0 ^ in3;
-        out6 = tmp0 ^ tmp2;
-        out0 = out3 ^ out5 ^ in6;
-        out4 = tmp0 ^ out0;
-        out2 = out4 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_36(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in0 ^ in2;
-        tmp0 = in1 ^ in3;
-        out0 = in3 ^ in4 ^ in6;
-        out6 = in1 ^ in2 ^ in4;
-        out5 = tmp0 ^ in0;
-        tmp1 = out5 ^ in5;
-        out2 = tmp1 ^ in4;
-        out3 = tmp1 ^ out4;
-        out1 = tmp0 ^ out2 ^ in7;
-        out7 = out3 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_37(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in2;
-        tmp1 = in2 ^ in4;
-        tmp2 = tmp0 ^ in6;
-        out3 = tmp0 ^ in5;
-        out4 = tmp1 ^ in0;
-        out6 = tmp2 ^ in4;
-        out1 = out3 ^ out4 ^ in7;
-        tmp3 = out4 ^ in1 ^ in3;
-        out7 = tmp3 ^ out1;
-        out2 = tmp3 ^ in5;
-        out5 = tmp1 ^ out2;
-        out0 = tmp2 ^ tmp3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_38(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in0 ^ in3;
-        tmp0 = in3 ^ in4;
-        tmp1 = in5 ^ in7;
-        tmp2 = out3 ^ in1;
-        out2 = tmp0 ^ in6;
-        out0 = tmp0 ^ tmp1;
-        out4 = tmp1 ^ tmp2;
-        out7 = out2 ^ in2;
-        out1 = out2 ^ in3 ^ in5;
-        out6 = out4 ^ in0 ^ in2;
-        out5 = tmp2 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_39(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in0;
-        tmp0 = in1 ^ in5;
-        tmp1 = tmp0 ^ in4;
-        out1 = tmp1 ^ in6;
-        out5 = out1 ^ in0 ^ in2;
-        tmp2 = tmp0 ^ out5;
-        out2 = tmp2 ^ in0 ^ in3;
-        out7 = out2 ^ in7;
-        out6 = tmp1 ^ out7;
-        out4 = tmp2 ^ out6;
-        out0 = out4 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_3A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        tmp1 = in0 ^ in2;
-        tmp2 = in3 ^ in4;
-        tmp3 = in1 ^ in6;
-        tmp4 = in3 ^ in7;
-        out4 = tmp0 ^ in5;
-        out5 = tmp1 ^ tmp3;
-        out3 = tmp1 ^ tmp4;
-        out0 = tmp2 ^ in5;
-        out7 = tmp2 ^ in2;
-        tmp5 = tmp3 ^ in4;
-        out2 = tmp4 ^ tmp5;
-        out1 = tmp5 ^ out4;
-        out6 = tmp0 ^ out3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_3B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in6;
-        tmp1 = in2 ^ in7;
-        tmp2 = tmp0 ^ in3;
-        out3 = tmp1 ^ in0;
-        out6 = tmp1 ^ tmp2;
-        out2 = out6 ^ in4;
-        out7 = tmp0 ^ out2;
-        out0 = out3 ^ out7 ^ in5;
-        out5 = out0 ^ out2 ^ in7;
-        out1 = tmp2 ^ out0;
-        out4 = out1 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_3C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3;
-        tmp1 = in2 ^ in7;
-        tmp2 = in1 ^ in6 ^ in7;
-        out2 = tmp0 ^ in4;
-        out3 = tmp0 ^ tmp2;
-        out4 = tmp1 ^ out3 ^ in5;
-        out5 = tmp2 ^ out2 ^ in2;
-        out1 = out4 ^ out5 ^ in6;
-        out0 = out1 ^ in3;
-        out7 = tmp1 ^ out0;
-        out6 = tmp2 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_3D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in2;
-        tmp1 = tmp0 ^ in3;
-        out2 = tmp1 ^ in4;
-        tmp2 = out2 ^ in5;
-        out4 = tmp2 ^ in1 ^ in6;
-        out5 = out4 ^ in7;
-        out6 = out5 ^ in0;
-        out7 = out6 ^ in1;
-        out0 = tmp0 ^ out7;
-        out1 = tmp1 ^ out5;
-        out3 = tmp2 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_3E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in5;
-        tmp1 = tmp0 ^ in4;
-        out0 = tmp1 ^ in6;
-        out7 = tmp1 ^ in2;
-        out6 = out7 ^ in1 ^ in5 ^ in7;
-        out2 = out6 ^ in0 ^ in2;
-        out4 = out0 ^ out6 ^ in0;
-        out5 = tmp0 ^ out4;
-        out3 = out5 ^ in7;
-        out1 = out3 ^ out6 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_3F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        out3 = tmp0 ^ in2 ^ in6;
-        tmp1 = out3 ^ in5 ^ in7;
-        out4 = tmp1 ^ in4;
-        out5 = tmp1 ^ in3;
-        out1 = out4 ^ in2;
-        out7 = out1 ^ out3 ^ in3;
-        out2 = tmp0 ^ out7 ^ in5;
-        tmp2 = out2 ^ in0;
-        out6 = tmp2 ^ in6;
-        out0 = tmp1 ^ tmp2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_40(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in3 ^ in7;
-        tmp0 = in3 ^ in4;
-        tmp1 = in6 ^ in7;
-        out4 = tmp0 ^ in2;
-        out5 = tmp0 ^ in5;
-        out0 = tmp1 ^ in2;
-        out7 = tmp1 ^ in1 ^ in5;
-        out2 = out0 ^ in4;
-        out3 = out2 ^ out5 ^ in7;
-        out6 = out3 ^ out4 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_41(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in2 ^ in3;
-        tmp0 = in5 ^ in6;
-        tmp1 = in6 ^ in7;
-        out5 = in3 ^ in4;
-        out1 = in1 ^ in3 ^ in7;
-        out6 = in0 ^ in4 ^ in5;
-        out3 = tmp0 ^ in2;
-        out7 = tmp0 ^ in1;
-        out2 = tmp1 ^ in4;
-        out0 = tmp1 ^ in0 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_42(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in2 ^ in6;
-        out5 = in3 ^ in5;
-        out1 = in0 ^ in3 ^ in7;
-        out7 = in1 ^ in5 ^ in7;
-        out4 = in2 ^ in4 ^ in7;
-        out6 = in0 ^ in4 ^ in6;
-        out2 = out0 ^ in1 ^ in4;
-        out3 = out5 ^ in6 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_43(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out5 = in3;
-        out7 = in1 ^ in5;
-        out4 = in2 ^ in7;
-        out6 = in0 ^ in4;
-        out0 = in0 ^ in2 ^ in6;
-        out3 = in5 ^ in6 ^ in7;
-        out2 = in1 ^ in4 ^ in6;
-        out1 = in0 ^ in1 ^ in3 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_44(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in3;
-        out0 = in2 ^ in7;
-        tmp0 = in4 ^ in7;
-        out7 = in1 ^ in6 ^ in7;
-        out6 = in0 ^ in5 ^ in6;
-        out4 = tmp0 ^ in3 ^ in6;
-        out3 = out0 ^ in1 ^ in3 ^ in5;
-        out2 = out0 ^ in0 ^ in4;
-        out5 = tmp0 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_45(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in1 ^ in3;
-        out7 = in1 ^ in6;
-        out5 = in4 ^ in7;
-        out6 = in0 ^ in5;
-        out0 = in0 ^ in2 ^ in7;
-        out4 = in3 ^ in6 ^ in7;
-        out2 = out5 ^ in0;
-        out3 = out0 ^ out6 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_46(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in2;
-        out1 = in0 ^ in3;
-        out7 = in1 ^ in7;
-        out4 = in4 ^ in6;
-        out5 = in5 ^ in7;
-        out6 = in0 ^ in6;
-        out3 = in1 ^ in3 ^ in5;
-        out2 = out4 ^ out6 ^ in1 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_47(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in6;
-        out7 = in1;
-        out5 = in7;
-        out6 = in0;
-        tmp0 = in0 ^ in1;
-        out3 = in1 ^ in5;
-        out0 = in0 ^ in2;
-        out1 = tmp0 ^ in3;
-        out2 = tmp0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_48(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in3;
-        out1 = in3 ^ in6 ^ in7;
-        out3 = tmp0 ^ in0;
-        out0 = tmp0 ^ out1 ^ in5;
-        tmp1 = out0 ^ in4;
-        out2 = tmp1 ^ in7;
-        out5 = tmp1 ^ in3;
-        out4 = out5 ^ in1;
-        out7 = tmp0 ^ out4;
-        out6 = tmp1 ^ out3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_49(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in0 ^ in2;
-        tmp0 = in2 ^ in5;
-        out2 = in4 ^ in5 ^ in6;
-        tmp1 = tmp0 ^ out2 ^ in3;
-        out7 = out2 ^ in1;
-        out5 = tmp1 ^ in7;
-        out4 = out5 ^ out7 ^ in6;
-        out1 = tmp0 ^ out4;
-        out6 = out1 ^ out7 ^ in0;
-        out0 = tmp1 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_4A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in6;
-        tmp1 = in3 ^ in7;
-        out0 = tmp0 ^ in5;
-        out3 = tmp1 ^ in0;
-        out5 = tmp1 ^ out0;
-        out4 = out0 ^ in1 ^ in4;
-        out1 = out3 ^ in6;
-        out2 = out4 ^ in7;
-        out6 = out1 ^ in4;
-        out7 = tmp0 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_4B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in0 ^ in7;
-        tmp0 = in1 ^ in5;
-        tmp1 = in2 ^ in6;
-        tmp2 = out3 ^ in3;
-        out7 = tmp0 ^ in4;
-        out4 = tmp0 ^ tmp1;
-        tmp3 = tmp1 ^ in0;
-        out6 = tmp2 ^ in4;
-        out5 = tmp2 ^ tmp3;
-        out1 = tmp2 ^ in1 ^ in6;
-        out2 = out7 ^ in6 ^ in7;
-        out0 = tmp3 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_4C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in3 ^ in6;
-        tmp0 = in2 ^ in5;
-        tmp1 = out1 ^ in5 ^ in7;
-        out0 = tmp0 ^ in7;
-        tmp2 = tmp0 ^ in4;
-        out6 = tmp1 ^ in0;
-        out2 = tmp2 ^ in0;
-        out5 = tmp2 ^ in6;
-        out3 = tmp0 ^ out6 ^ in1;
-        out7 = out0 ^ out5 ^ in1;
-        out4 = tmp1 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_4D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in5;
-        tmp1 = in1 ^ in6;
-        out4 = in1 ^ in3 ^ in5;
-        tmp2 = tmp0 ^ in7;
-        out2 = tmp0 ^ in4;
-        out1 = tmp1 ^ in3;
-        out7 = tmp1 ^ in4;
-        out0 = tmp2 ^ in2;
-        out6 = tmp2 ^ in3;
-        out5 = out7 ^ in1 ^ in2;
-        out3 = tmp1 ^ out0 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_4E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in2 ^ in5;
-        out7 = in1 ^ in4 ^ in7;
-        out1 = in0 ^ in3 ^ in6;
-        out5 = out0 ^ in6;
-        out4 = out7 ^ in5;
-        out3 = out1 ^ in1;
-        out6 = out1 ^ in7;
-        out2 = out4 ^ in0 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_4F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out5 = in2 ^ in6;
-        out7 = in1 ^ in4;
-        out3 = in0 ^ in1 ^ in6;
-        out4 = in1 ^ in5 ^ in7;
-        out0 = in0 ^ in2 ^ in5;
-        out6 = in0 ^ in3 ^ in7;
-        out1 = out3 ^ in3;
-        out2 = out4 ^ in0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_50(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in2 ^ in7;
-        tmp0 = in3 ^ in5;
-        out0 = out2 ^ in4 ^ in6;
-        out1 = tmp0 ^ in7;
-        tmp1 = tmp0 ^ in6;
-        out3 = out0 ^ in3;
-        out7 = tmp1 ^ in1;
-        tmp2 = tmp1 ^ in0;
-        out5 = out3 ^ in1 ^ in2;
-        out4 = tmp2 ^ in2;
-        out6 = tmp2 ^ out3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_51(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in7;
-        out3 = in2 ^ in4 ^ in6 ^ in7;
-        out0 = out3 ^ in0;
-        out6 = out0 ^ in5;
-        out4 = out6 ^ in3 ^ in7;
-        out1 = out0 ^ out4 ^ in1;
-        out7 = out1 ^ in6;
-        out5 = out7 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_52(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in1 ^ in2;
-        tmp0 = in2 ^ in4;
-        tmp1 = in3 ^ in5;
-        tmp2 = in3 ^ in6;
-        tmp3 = in0 ^ in7;
-        out0 = tmp0 ^ in6;
-        out6 = tmp0 ^ tmp3;
-        out7 = tmp1 ^ in1;
-        out1 = tmp1 ^ tmp3;
-        out3 = tmp2 ^ in4;
-        out5 = tmp2 ^ in1 ^ in7;
-        out4 = tmp2 ^ out1 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_53(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in1;
-        out3 = in4 ^ in6;
-        out0 = out3 ^ in0 ^ in2;
-        out6 = out0 ^ in7;
-        out4 = out6 ^ in5;
-        out7 = out0 ^ out4 ^ in1 ^ in3;
-        out1 = out7 ^ in0;
-        out5 = out7 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_54(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in3 ^ in5;
-        tmp0 = in1 ^ in3;
-        tmp1 = in2 ^ in4;
-        tmp2 = in0 ^ in7;
-        out5 = in1 ^ in4 ^ in6;
-        out4 = tmp2 ^ out1;
-        out7 = tmp0 ^ in6;
-        out3 = tmp0 ^ tmp1;
-        out0 = tmp1 ^ in7;
-        tmp3 = tmp2 ^ in2;
-        out2 = tmp3 ^ in6;
-        out6 = tmp3 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_55(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in3;
-        tmp1 = in1 ^ in4;
-        tmp2 = in6 ^ in7;
-        out7 = tmp0 ^ tmp2;
-        out1 = tmp0 ^ in5;
-        out3 = tmp1 ^ in2;
-        out5 = tmp1 ^ in5 ^ in6;
-        out2 = tmp2 ^ in0;
-        out4 = out5 ^ out7 ^ in0;
-        out6 = out2 ^ in2 ^ in5;
-        out0 = out5 ^ out6 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_56(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in2 ^ in4;
-        tmp0 = in0 ^ in2;
-        out4 = in0 ^ in5;
-        out7 = in1 ^ in3;
-        out5 = in1 ^ in6;
-        out6 = tmp0 ^ in7;
-        out2 = tmp0 ^ out5;
-        out1 = out4 ^ in3;
-        out3 = out7 ^ in4 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_57(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in5;
-        tmp1 = in1 ^ in7;
-        out0 = in0 ^ in2 ^ in4;
-        out5 = in1 ^ in5 ^ in6;
-        out4 = tmp0 ^ in4;
-        out1 = tmp0 ^ in1 ^ in3;
-        out2 = tmp0 ^ out5;
-        out3 = tmp1 ^ in4;
-        out7 = tmp1 ^ in3;
-        out6 = tmp1 ^ out2 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_58(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in2 ^ in5;
-        tmp0 = in2 ^ in3 ^ in4;
-        out5 = tmp0 ^ in1;
-        out6 = tmp0 ^ in0 ^ in5;
-        out3 = out6 ^ in7;
-        tmp1 = out2 ^ out5;
-        out7 = tmp1 ^ in6;
-        out4 = tmp1 ^ out3 ^ in3;
-        out0 = out4 ^ out7 ^ in0;
-        out1 = tmp0 ^ out0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_59(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in5;
-        tmp0 = in0 ^ in5 ^ in7;
-        out3 = tmp0 ^ in2 ^ in4;
-        out0 = out3 ^ in6;
-        tmp1 = out0 ^ in7;
-        out6 = tmp1 ^ in3;
-        out5 = out6 ^ in0 ^ in1 ^ in6;
-        out4 = tmp0 ^ out5;
-        out1 = tmp1 ^ out4;
-        out7 = out1 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_5A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in2;
-        tmp1 = in2 ^ in5;
-        out5 = tmp0 ^ in3;
-        out4 = tmp0 ^ in0;
-        tmp2 = tmp1 ^ in4;
-        out2 = tmp1 ^ in1 ^ in7;
-        out7 = tmp2 ^ out5;
-        out6 = out4 ^ out7 ^ in5;
-        out0 = tmp2 ^ in6;
-        out1 = out0 ^ out6 ^ in7;
-        out3 = tmp1 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_5B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in3;
-        tmp1 = in0 ^ in4;
-        tmp2 = in1 ^ in5;
-        out5 = tmp0 ^ tmp2;
-        tmp3 = tmp1 ^ in6;
-        out3 = tmp1 ^ in5;
-        out2 = tmp2 ^ in7;
-        tmp4 = out3 ^ in2;
-        out7 = out2 ^ in3 ^ in4;
-        out0 = tmp4 ^ in6;
-        out6 = tmp0 ^ tmp3;
-        out4 = tmp2 ^ tmp4;
-        out1 = tmp3 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_5C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in6;
-        tmp1 = in0 ^ in2 ^ in5;
-        out1 = tmp0 ^ in5;
-        tmp2 = tmp0 ^ in1;
-        out2 = tmp1 ^ in6;
-        out6 = tmp1 ^ in3;
-        out4 = tmp2 ^ in0;
-        out7 = tmp2 ^ in4;
-        out3 = tmp1 ^ out7;
-        out0 = out3 ^ out4 ^ in7;
-        out5 = out0 ^ in1 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_5D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        tmp1 = in0 ^ in6;
-        out2 = tmp1 ^ in5;
-        tmp2 = out2 ^ in3;
-        out6 = tmp2 ^ in2;
-        out1 = tmp0 ^ tmp2;
-        tmp3 = out1 ^ in4 ^ in5;
-        out4 = tmp3 ^ in0;
-        out7 = tmp3 ^ in7;
-        tmp4 = out4 ^ out6;
-        out5 = tmp4 ^ in7;
-        out0 = tmp0 ^ out5;
-        out3 = tmp1 ^ tmp4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_5E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in5;
-        tmp1 = in3 ^ in5;
-        tmp2 = in1 ^ in7;
-        out7 = in1 ^ in3 ^ in4;
-        out0 = tmp0 ^ in4;
-        tmp3 = tmp1 ^ in0;
-        out5 = tmp2 ^ in2;
-        out1 = tmp3 ^ in6;
-        out6 = tmp0 ^ tmp3;
-        tmp4 = tmp2 ^ out1;
-        out3 = tmp4 ^ in4;
-        out4 = tmp1 ^ tmp4;
-        out2 = tmp0 ^ out4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_5F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in5;
-        tmp1 = in0 ^ in6;
-        tmp2 = tmp0 ^ in7;
-        tmp3 = tmp1 ^ in3;
-        out2 = tmp1 ^ tmp2;
-        out5 = tmp2 ^ in2;
-        out6 = tmp3 ^ in2;
-        out3 = out2 ^ in4;
-        out4 = out3 ^ in5;
-        out1 = tmp0 ^ tmp3;
-        out7 = tmp3 ^ out4;
-        out0 = out4 ^ out5 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_60(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in2 ^ in5;
-        tmp0 = in3 ^ in6;
-        out1 = in3 ^ in4 ^ in7;
-        out7 = out4 ^ in1;
-        tmp1 = out4 ^ in4;
-        out0 = tmp0 ^ in2;
-        out5 = tmp0 ^ in0;
-        out2 = tmp0 ^ tmp1;
-        out3 = tmp1 ^ in7;
-        out6 = out3 ^ out7 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_61(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in5;
-        out4 = tmp0 ^ in4;
-        tmp1 = out4 ^ in3;
-        out3 = tmp1 ^ in7;
-        out2 = tmp1 ^ in2 ^ in6;
-        out1 = tmp0 ^ out3 ^ in1;
-        out0 = out2 ^ out4 ^ in0;
-        out7 = tmp1 ^ out1;
-        out6 = out0 ^ out1 ^ in2;
-        out5 = tmp0 ^ out0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_62(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in4 ^ in5;
-        tmp0 = in0 ^ in3 ^ in4;
-        out1 = tmp0 ^ in7;
-        out5 = tmp0 ^ in6;
-        tmp1 = out1 ^ in0;
-        tmp2 = tmp1 ^ out3;
-        out4 = tmp2 ^ in2;
-        tmp3 = tmp2 ^ in1;
-        out0 = out4 ^ in5 ^ in6;
-        out7 = tmp3 ^ out0;
-        out6 = tmp0 ^ tmp3;
-        out2 = tmp1 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_63(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in4;
-        tmp1 = in1 ^ in7;
-        out3 = tmp0 ^ in5;
-        tmp2 = out3 ^ in6;
-        out4 = out3 ^ in2 ^ in7;
-        out5 = tmp2 ^ in0;
-        tmp3 = out5 ^ in3;
-        out0 = tmp3 ^ out4;
-        out2 = tmp1 ^ tmp2;
-        out6 = tmp1 ^ tmp3;
-        tmp4 = tmp0 ^ out2;
-        out1 = tmp4 ^ out5;
-        out7 = tmp4 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_64(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in2 ^ in3;
-        out1 = in3 ^ in4;
-        out7 = in1 ^ in2;
-        tmp0 = in4 ^ in5;
-        tmp1 = in0 ^ in7;
-        out4 = in5 ^ in6 ^ in7;
-        out2 = tmp0 ^ out0 ^ in0;
-        out3 = tmp0 ^ out7 ^ in6;
-        out5 = tmp1 ^ in6;
-        out6 = tmp1 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_65(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3;
-        tmp1 = in4 ^ in5;
-        tmp2 = in6 ^ in7;
-        out7 = in1 ^ in2 ^ in7;
-        out1 = in1 ^ in3 ^ in4;
-        out0 = tmp0 ^ in2;
-        out2 = tmp0 ^ tmp1;
-        out4 = tmp1 ^ tmp2;
-        tmp3 = tmp2 ^ in0;
-        out3 = out4 ^ out7 ^ in3;
-        out5 = tmp3 ^ in5;
-        out6 = tmp3 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_66(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in2;
-        tmp1 = in2 ^ in3;
-        tmp2 = in0 ^ in4;
-        out7 = tmp0 ^ in6;
-        out0 = tmp1 ^ in7;
-        out1 = tmp2 ^ in3;
-        tmp3 = tmp2 ^ in6;
-        tmp4 = out1 ^ in5;
-        out5 = tmp3 ^ in7;
-        out4 = tmp3 ^ tmp4;
-        out2 = tmp0 ^ tmp4 ^ in7;
-        out6 = tmp1 ^ out2 ^ in4;
-        out3 = tmp3 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_67(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3;
-        tmp1 = tmp0 ^ in1;
-        tmp2 = tmp0 ^ in7;
-        out1 = tmp1 ^ in4;
-        out0 = tmp2 ^ in2;
-        tmp3 = out1 ^ in7;
-        out2 = tmp3 ^ in5;
-        out3 = out2 ^ in0 ^ in6;
-        out7 = tmp1 ^ out0 ^ in6;
-        out5 = tmp1 ^ out3;
-        out4 = tmp2 ^ out5;
-        out6 = tmp3 ^ out4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_68(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in4;
-        tmp1 = in2 ^ in3 ^ in5;
-        tmp2 = tmp0 ^ in1;
-        tmp3 = tmp0 ^ in6;
-        out0 = tmp1 ^ in6;
-        out6 = tmp2 ^ in0;
-        out7 = tmp1 ^ tmp2;
-        out1 = tmp3 ^ in7;
-        out2 = out1 ^ in2;
-        out4 = tmp2 ^ out2;
-        out3 = out4 ^ out6 ^ in3;
-        out5 = tmp3 ^ out3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_69(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in6 ^ in7;
-        out2 = tmp0 ^ in3 ^ in4;
-        out1 = out2 ^ in1;
-        out3 = out2 ^ in0 ^ in2;
-        out4 = out1 ^ in2 ^ in3;
-        out6 = out1 ^ in0 ^ in7;
-        out7 = out4 ^ in5 ^ in6;
-        out5 = out4 ^ out6 ^ in5;
-        out0 = tmp0 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_6A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in6;
-        out3 = in0 ^ in4 ^ in6;
-        tmp1 = tmp0 ^ in3;
-        out4 = tmp1 ^ in1;
-        tmp2 = tmp1 ^ in7;
-        out2 = out4 ^ in4;
-        out0 = tmp2 ^ in5;
-        out5 = tmp2 ^ out3;
-        out7 = out2 ^ in3 ^ in5;
-        out1 = tmp0 ^ out5;
-        out6 = tmp1 ^ out7 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_6B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in4 ^ in6;
-        out2 = tmp0 ^ in1 ^ in3;
-        out4 = out2 ^ in2;
-        tmp1 = out2 ^ in0;
-        out7 = out4 ^ in3 ^ in5 ^ in7;
-        out1 = tmp1 ^ in7;
-        out3 = tmp1 ^ in1;
-        out6 = tmp1 ^ in5;
-        out0 = tmp1 ^ out7 ^ in6;
-        out5 = tmp0 ^ out0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_6C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in1;
-        tmp0 = in2 ^ in3;
-        out5 = in0 ^ in2;
-        out1 = in3 ^ in4 ^ in6;
-        tmp1 = out5 ^ in1;
-        out0 = tmp0 ^ in5;
-        out6 = tmp0 ^ tmp1;
-        out3 = tmp1 ^ in4;
-        out7 = out3 ^ in0;
-        out2 = out6 ^ out7 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_6D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in1 ^ in4;
-        tmp0 = in0 ^ in2;
-        tmp1 = out4 ^ in3;
-        out7 = out4 ^ in2 ^ in7;
-        out5 = tmp0 ^ in5;
-        out3 = tmp0 ^ tmp1;
-        out1 = tmp1 ^ in6;
-        out0 = out5 ^ in3;
-        out2 = out3 ^ out7 ^ in4;
-        out6 = out1 ^ in0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_6E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in3;
-        tmp1 = in0 ^ in4;
-        out4 = tmp0 ^ in7;
-        out6 = tmp0 ^ in0 ^ in5;
-        out5 = tmp1 ^ in2;
-        tmp2 = tmp1 ^ in3;
-        out3 = tmp2 ^ out4;
-        out1 = tmp2 ^ in6;
-        out2 = tmp0 ^ out5;
-        out0 = out2 ^ out3 ^ in5;
-        out7 = out1 ^ out2 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_6F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in7;
-        tmp1 = tmp0 ^ in4;
-        tmp2 = tmp0 ^ in0 ^ in2;
-        out4 = tmp1 ^ in1;
-        out0 = tmp2 ^ in5;
-        out3 = out4 ^ in0;
-        out2 = out3 ^ in7;
-        out1 = out2 ^ in6;
-        out6 = out1 ^ in4 ^ in5;
-        out7 = tmp2 ^ out1;
-        out5 = tmp1 ^ out0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_70(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in2;
-        tmp0 = in2 ^ in4;
-        out2 = in2 ^ in3 ^ in5;
-        tmp1 = tmp0 ^ in6;
-        tmp2 = out2 ^ in7;
-        out0 = tmp1 ^ in3;
-        out4 = tmp1 ^ in0;
-        out7 = tmp2 ^ in1;
-        out6 = out4 ^ in1;
-        out5 = out7 ^ in0 ^ in2;
-        out1 = tmp0 ^ tmp2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_71(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in3 ^ in5;
-        out3 = in2 ^ in3;
-        tmp0 = in0 ^ in2;
-        tmp1 = out2 ^ in1;
-        out4 = tmp0 ^ in6;
-        tmp2 = tmp0 ^ in1;
-        out7 = tmp1 ^ in2;
-        out1 = tmp1 ^ in4 ^ in7;
-        out0 = out4 ^ in3 ^ in4;
-        out6 = tmp2 ^ in4;
-        out5 = tmp2 ^ out3 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_72(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in7;
-        tmp0 = in0 ^ in4;
-        tmp1 = tmp0 ^ in3 ^ in7;
-        out1 = tmp1 ^ in5;
-        out5 = out1 ^ in1;
-        tmp2 = tmp0 ^ out5;
-        out2 = tmp2 ^ in2;
-        out7 = out2 ^ in6;
-        out6 = tmp1 ^ out7;
-        out4 = tmp2 ^ out6;
-        out0 = out4 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_73(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in3 ^ in7;
-        out2 = out3 ^ in1 ^ in5;
-        out1 = out2 ^ in0 ^ in4;
-        out5 = out1 ^ in5;
-        out6 = out1 ^ out3 ^ in2;
-        out0 = out2 ^ out6 ^ in6;
-        out7 = out0 ^ out1 ^ in3;
-        out4 = out0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_74(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in4;
-        tmp1 = in1 ^ in2 ^ in6;
-        out4 = in0 ^ in4 ^ in7;
-        out5 = in0 ^ in1 ^ in5;
-        out0 = tmp0 ^ in2;
-        out1 = tmp0 ^ in5;
-        out3 = tmp1 ^ in7;
-        out6 = tmp1 ^ in0;
-        out2 = tmp1 ^ out5 ^ in3;
-        out7 = out3 ^ in3 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_75(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in0 ^ in7;
-        tmp0 = in1 ^ in3;
-        out5 = in0 ^ in1;
-        out7 = tmp0 ^ in2;
-        tmp1 = tmp0 ^ in4;
-        out6 = out5 ^ in2;
-        tmp2 = out7 ^ in6;
-        out1 = tmp1 ^ in5;
-        out0 = tmp1 ^ out6;
-        out3 = tmp2 ^ in7;
-        out2 = tmp2 ^ out6 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_76(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in1 ^ in6;
-        tmp0 = in0 ^ in5;
-        tmp1 = in3 ^ in7;
-        tmp2 = tmp0 ^ in4;
-        tmp3 = tmp1 ^ in2;
-        out5 = tmp2 ^ in1;
-        out1 = tmp2 ^ in3;
-        out0 = tmp3 ^ in4;
-        out4 = out1 ^ in5;
-        out7 = tmp3 ^ out3;
-        out2 = tmp0 ^ out7;
-        out6 = tmp1 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_77(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in0 ^ in3;
-        tmp0 = in1 ^ in4;
-        tmp1 = in1 ^ in6;
-        tmp2 = out4 ^ in5;
-        out5 = tmp0 ^ in0;
-        out1 = tmp0 ^ tmp2;
-        out3 = tmp1 ^ in3;
-        out2 = tmp1 ^ tmp2 ^ in7;
-        out7 = out3 ^ in2;
-        tmp3 = out7 ^ in6;
-        out6 = tmp2 ^ tmp3;
-        out0 = tmp3 ^ out5 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_78(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3;
-        tmp1 = in2 ^ in7;
-        tmp2 = in0 ^ in5 ^ in6;
-        out2 = tmp1 ^ in3;
-        out3 = tmp2 ^ in2;
-        out5 = out3 ^ in1 ^ in3;
-        out0 = tmp0 ^ out3 ^ in4;
-        out1 = tmp1 ^ out0;
-        out4 = out1 ^ out5 ^ in5;
-        out7 = tmp0 ^ out4;
-        out6 = tmp2 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_79(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in3 ^ in7;
-        tmp0 = in3 ^ in4;
-        tmp1 = in1 ^ in5;
-        tmp2 = tmp1 ^ in2;
-        out4 = tmp2 ^ in0 ^ in7;
-        tmp3 = out4 ^ in5;
-        out5 = tmp3 ^ out2 ^ in6;
-        out7 = tmp0 ^ tmp2;
-        out6 = tmp0 ^ tmp3;
-        out3 = tmp1 ^ out5;
-        out0 = out3 ^ in4;
-        out1 = tmp3 ^ out0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_7A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in2;
-        out2 = tmp0 ^ in3;
-        tmp1 = out2 ^ in4;
-        out4 = tmp1 ^ in0 ^ in5;
-        out5 = out4 ^ in6;
-        out6 = out5 ^ in7;
-        out7 = out6 ^ in0;
-        out0 = out7 ^ in1;
-        out1 = tmp0 ^ out6;
-        out3 = tmp1 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_7B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in1 ^ in3;
-        tmp0 = in0 ^ in5;
-        out4 = tmp0 ^ out2 ^ in2;
-        tmp1 = out4 ^ in4;
-        out6 = tmp1 ^ in7;
-        out5 = tmp1 ^ in5 ^ in6;
-        out0 = out6 ^ in1 ^ in6;
-        tmp2 = out0 ^ in2;
-        out1 = tmp2 ^ in1;
-        out3 = tmp2 ^ in4;
-        out7 = tmp0 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_7C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in5;
-        tmp1 = tmp0 ^ in4;
-        out0 = tmp1 ^ in2;
-        out1 = tmp1 ^ in6;
-        out7 = out0 ^ in1 ^ in5 ^ in7;
-        out5 = out1 ^ out7 ^ in0;
-        out3 = out5 ^ in6;
-        out6 = tmp0 ^ out5;
-        out2 = out6 ^ in1;
-        out4 = out2 ^ out7 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_7D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in2;
-        tmp1 = tmp0 ^ in3;
-        tmp2 = tmp0 ^ in6;
-        out7 = tmp1 ^ in4;
-        tmp3 = tmp2 ^ in0;
-        out5 = tmp3 ^ in7;
-        out4 = tmp3 ^ in2 ^ in5;
-        out2 = tmp1 ^ out5;
-        out6 = tmp2 ^ out2;
-        out0 = out4 ^ out7 ^ in6;
-        out1 = tmp3 ^ out0;
-        out3 = out6 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_7E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in4;
-        tmp1 = in0 ^ in5;
-        out1 = tmp0 ^ tmp1 ^ in6;
-        out3 = tmp1 ^ in1;
-        out4 = out1 ^ in1 ^ in7;
-        tmp2 = out4 ^ in3;
-        out5 = tmp2 ^ in2;
-        out6 = tmp0 ^ out5;
-        out7 = tmp1 ^ out4 ^ in2;
-        out2 = out6 ^ in5 ^ in7;
-        out0 = tmp2 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_7F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in7;
-        tmp1 = tmp0 ^ in3 ^ in5;
-        tmp2 = tmp1 ^ in0;
-        out0 = tmp2 ^ in4;
-        out6 = tmp2 ^ in1;
-        out3 = tmp0 ^ out6;
-        tmp3 = out3 ^ in6;
-        out1 = tmp3 ^ in4;
-        out2 = tmp3 ^ in5;
-        out4 = tmp3 ^ in7;
-        out5 = tmp1 ^ out1;
-        out7 = out0 ^ out4 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_80(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in3;
-        tmp1 = in4 ^ in5;
-        out1 = in2 ^ in6 ^ in7;
-        out5 = tmp0 ^ in4;
-        tmp2 = tmp0 ^ in1;
-        out6 = tmp1 ^ in3;
-        out7 = tmp1 ^ in0 ^ in6;
-        out4 = tmp2 ^ in7;
-        out3 = tmp2 ^ out6;
-        out2 = out3 ^ out5 ^ in6;
-        out0 = out2 ^ in3 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_81(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in4 ^ in6;
-        tmp1 = tmp0 ^ in3;
-        out6 = tmp1 ^ in5;
-        out5 = out6 ^ in2 ^ in6;
-        out3 = out5 ^ in1;
-        out2 = tmp0 ^ out3;
-        out1 = out3 ^ out6 ^ in7;
-        out4 = tmp1 ^ out1;
-        out7 = out2 ^ out4 ^ in0;
-        out0 = out7 ^ in1 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_82(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in1 ^ in2;
-        tmp0 = in6 ^ in7;
-        out5 = in2 ^ in3;
-        out6 = in3 ^ in4;
-        out7 = in0 ^ in4 ^ in5;
-        out0 = in1 ^ in5 ^ in6;
-        out1 = tmp0 ^ in0 ^ in2;
-        out2 = tmp0 ^ in3 ^ in5;
-        out3 = tmp0 ^ out0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_83(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        tmp1 = in2 ^ in5;
-        tmp2 = in3 ^ in6;
-        out4 = in1 ^ in2 ^ in4;
-        out0 = tmp0 ^ in5 ^ in6;
-        out5 = tmp1 ^ in3;
-        tmp3 = tmp1 ^ in7;
-        out6 = tmp2 ^ in4;
-        out2 = tmp2 ^ tmp3;
-        tmp4 = tmp3 ^ out4;
-        out1 = tmp3 ^ out0;
-        out3 = tmp4 ^ in3;
-        out7 = tmp0 ^ tmp4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_84(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in2 ^ in6;
-        out6 = in3 ^ in5;
-        out0 = in1 ^ in5 ^ in7;
-        out7 = in0 ^ in4 ^ in6;
-        out4 = in1 ^ in3 ^ in6;
-        out5 = in2 ^ in4 ^ in7;
-        out2 = out6 ^ in0 ^ in1;
-        out3 = out5 ^ in5 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_85(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in6;
-        tmp1 = in3 ^ in6;
-        tmp2 = tmp0 ^ in4;
-        out1 = tmp0 ^ in2;
-        out6 = tmp1 ^ in5;
-        out4 = tmp2 ^ in3;
-        tmp3 = out1 ^ out6;
-        out2 = tmp3 ^ in0;
-        out3 = tmp2 ^ tmp3 ^ in7;
-        out7 = out2 ^ out3 ^ in1;
-        out5 = tmp1 ^ out3;
-        out0 = tmp2 ^ out7 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_86(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out6 = in3;
-        out7 = in0 ^ in4;
-        out0 = in1 ^ in5;
-        out5 = in2 ^ in7;
-        out3 = in4 ^ in5 ^ in6;
-        out1 = in0 ^ in2 ^ in6;
-        out4 = in1 ^ in6 ^ in7;
-        out2 = in0 ^ in3 ^ in5 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_87(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out6 = in3 ^ in6;
-        tmp0 = in0 ^ in1;
-        out7 = in0 ^ in4 ^ in7;
-        out5 = in2 ^ in5 ^ in7;
-        out3 = out6 ^ in4 ^ in5;
-        out0 = tmp0 ^ in5;
-        tmp1 = tmp0 ^ in6;
-        out2 = out5 ^ in0 ^ in3;
-        out1 = tmp1 ^ in2;
-        out4 = tmp1 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_88(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in2 ^ in7;
-        tmp0 = in5 ^ in6;
-        out0 = in1 ^ in6 ^ in7;
-        out6 = in4 ^ in5 ^ in7;
-        out3 = out0 ^ out1 ^ in0 ^ in4;
-        out7 = tmp0 ^ in0;
-        tmp1 = tmp0 ^ in3;
-        out2 = out0 ^ in3;
-        out4 = tmp1 ^ in2;
-        out5 = tmp1 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_89(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in7;
-        tmp1 = in2 ^ in7;
-        tmp2 = tmp0 ^ in6;
-        out1 = tmp1 ^ in1;
-        out7 = tmp2 ^ in5;
-        out0 = tmp2 ^ in1;
-        out2 = out1 ^ in3 ^ in6;
-        out6 = out7 ^ in0 ^ in4;
-        out5 = out6 ^ in3;
-        out3 = tmp0 ^ out2 ^ in4;
-        out4 = tmp1 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_8A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in1 ^ in6;
-        out7 = in0 ^ in5;
-        out2 = in3 ^ in6;
-        out6 = in4 ^ in7;
-        out1 = in0 ^ in2 ^ in7;
-        out3 = out0 ^ out6 ^ in0;
-        out4 = out1 ^ out7 ^ in6;
-        out5 = out2 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_8B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        tmp1 = in3 ^ in6;
-        tmp2 = in5 ^ in7;
-        tmp3 = tmp0 ^ in7;
-        out0 = tmp0 ^ in6;
-        out2 = tmp1 ^ in2;
-        out5 = tmp1 ^ tmp2;
-        out7 = tmp2 ^ in0;
-        tmp4 = tmp3 ^ in4;
-        out1 = tmp3 ^ in2;
-        out6 = tmp4 ^ out0;
-        out4 = out6 ^ in2 ^ in5;
-        out3 = tmp1 ^ tmp4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_8C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in2;
-        out0 = in1 ^ in7;
-        out7 = in0 ^ in6;
-        out5 = in4 ^ in6;
-        out6 = in5 ^ in7;
-        out2 = out0 ^ in0 ^ in3;
-        out3 = out5 ^ out7 ^ in2 ^ in7;
-        out4 = out6 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_8D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in1 ^ in2;
-        tmp0 = in6 ^ in7;
-        out0 = in0 ^ in1 ^ in7;
-        out5 = in4 ^ in5 ^ in6;
-        out6 = tmp0 ^ in5;
-        out7 = tmp0 ^ in0;
-        out4 = tmp0 ^ out5 ^ in3;
-        out2 = out0 ^ in2 ^ in3;
-        out3 = out2 ^ in1 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_8E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in1;
-        out4 = in5;
-        out7 = in0;
-        out5 = in6;
-        out6 = in7;
-        out3 = in0 ^ in4;
-        out1 = in0 ^ in2;
-        out2 = in0 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_8F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in0 ^ in1;
-        tmp0 = in0 ^ in3;
-        out4 = in4 ^ in5;
-        out7 = in0 ^ in7;
-        out5 = in5 ^ in6;
-        out6 = in6 ^ in7;
-        out1 = out0 ^ in2;
-        out2 = tmp0 ^ in2;
-        out3 = tmp0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_90(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in2;
-        tmp1 = in2 ^ in6 ^ in7;
-        out3 = tmp0 ^ in7;
-        out1 = tmp1 ^ in5;
-        tmp2 = out1 ^ in4;
-        out6 = tmp2 ^ in3;
-        out5 = out6 ^ in1;
-        out4 = out5 ^ in0;
-        out0 = tmp0 ^ tmp2;
-        out7 = tmp0 ^ out4;
-        out2 = tmp1 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_91(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in4;
-        tmp1 = tmp0 ^ in3 ^ in5;
-        out2 = tmp1 ^ in1;
-        out6 = tmp1 ^ in7;
-        tmp2 = out2 ^ in5 ^ in7;
-        out3 = tmp2 ^ in4;
-        out5 = tmp2 ^ in6;
-        out1 = tmp1 ^ out5 ^ in2;
-        tmp3 = out1 ^ in0;
-        out4 = tmp3 ^ in3;
-        out0 = tmp0 ^ tmp3;
-        out7 = tmp2 ^ tmp3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_92(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in1;
-        tmp0 = in4 ^ in5;
-        tmp1 = tmp0 ^ in1;
-        out2 = tmp0 ^ in3 ^ in7;
-        out0 = tmp1 ^ in6;
-        out7 = out2 ^ in0;
-        out4 = out0 ^ in0 ^ in2;
-        out5 = out4 ^ out7 ^ in5;
-        out6 = tmp1 ^ out5;
-        out1 = out6 ^ out7 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_93(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in1 ^ in3;
-        tmp0 = in2 ^ in7;
-        tmp1 = out3 ^ in6;
-        tmp2 = tmp0 ^ in4;
-        out5 = tmp0 ^ tmp1;
-        out6 = tmp2 ^ in3;
-        out2 = out6 ^ in5;
-        out0 = out2 ^ out5 ^ in0;
-        out7 = tmp1 ^ out0;
-        out1 = tmp2 ^ out0;
-        out4 = out1 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_94(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in2 ^ in6;
-        tmp0 = in1 ^ in4 ^ in5;
-        out1 = out3 ^ in5;
-        out5 = tmp0 ^ out3;
-        out0 = tmp0 ^ in7;
-        out4 = tmp0 ^ in0 ^ in3;
-        out6 = out1 ^ in3 ^ in7;
-        out2 = out4 ^ in6;
-        out7 = out0 ^ out2 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_95(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in3;
-        out3 = tmp0 ^ in6;
-        tmp1 = tmp0 ^ in7;
-        tmp2 = out3 ^ in0;
-        out6 = tmp1 ^ in5;
-        tmp3 = tmp2 ^ in4;
-        out7 = tmp3 ^ in2;
-        tmp4 = tmp3 ^ in5;
-        out2 = tmp4 ^ in1;
-        tmp5 = out2 ^ in6;
-        out0 = tmp1 ^ tmp5;
-        out1 = tmp5 ^ out7;
-        out4 = tmp2 ^ out1;
-        out5 = tmp4 ^ out4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_96(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in6 ^ in7;
-        tmp0 = in1 ^ in5;
-        tmp1 = in5 ^ in6;
-        out6 = out3 ^ in2 ^ in3;
-        out0 = tmp0 ^ in4;
-        tmp2 = tmp1 ^ in2;
-        out4 = out0 ^ in0 ^ in7;
-        out1 = tmp2 ^ in0;
-        out5 = tmp2 ^ in1;
-        out7 = tmp0 ^ out4 ^ in3;
-        out2 = tmp1 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_97(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in4;
-        tmp1 = in2 ^ in6;
-        out3 = in3 ^ in6 ^ in7;
-        out7 = tmp0 ^ in3;
-        tmp2 = tmp0 ^ in5;
-        out5 = tmp1 ^ in1;
-        out6 = tmp1 ^ out3;
-        out0 = tmp2 ^ in1;
-        out2 = tmp2 ^ out3 ^ in2;
-        tmp3 = out0 ^ in4;
-        out4 = tmp3 ^ in7;
-        out1 = tmp1 ^ tmp3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_98(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in5 ^ in7;
-        tmp1 = in1 ^ in4 ^ in7;
-        out1 = tmp0 ^ in2;
-        out0 = tmp1 ^ in6;
-        out2 = tmp1 ^ in3;
-        out6 = out0 ^ out1 ^ in1;
-        out5 = tmp0 ^ out2;
-        out3 = tmp1 ^ out6 ^ in0;
-        out7 = out0 ^ out5 ^ in0;
-        out4 = out6 ^ out7 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_99(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3;
-        out5 = in1 ^ in3 ^ in4;
-        out6 = in2 ^ in4 ^ in5;
-        out4 = tmp0 ^ in2;
-        tmp1 = tmp0 ^ in6;
-        tmp2 = out5 ^ in7;
-        out7 = tmp1 ^ in5;
-        out0 = tmp1 ^ tmp2;
-        out2 = tmp2 ^ in2;
-        out3 = out0 ^ out6 ^ in3;
-        out1 = tmp1 ^ out3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_9A(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in3 ^ in4;
-        tmp0 = in0 ^ in5;
-        tmp1 = in1 ^ in6;
-        out5 = in1 ^ in3 ^ in5;
-        tmp2 = tmp0 ^ in7;
-        out3 = tmp0 ^ tmp1;
-        out0 = tmp1 ^ in4;
-        out7 = tmp2 ^ in3;
-        out1 = tmp2 ^ in2;
-        out6 = out0 ^ in1 ^ in2;
-        out4 = out1 ^ in4 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_9B(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out5 = in1 ^ in3;
-        tmp0 = in3 ^ in5;
-        out6 = in2 ^ in4;
-        out4 = in0 ^ in2 ^ in7;
-        out7 = tmp0 ^ in0;
-        out2 = out6 ^ in3;
-        out1 = out4 ^ in1 ^ in5;
-        out3 = out7 ^ in1 ^ in6;
-        out0 = tmp0 ^ out3 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_9C(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out1 = in2 ^ in5;
-        tmp0 = in0 ^ in3 ^ in6;
-        out3 = out1 ^ in0;
-        out6 = out1 ^ in6;
-        out7 = tmp0 ^ in7;
-        out4 = out7 ^ in4;
-        out2 = out4 ^ in1;
-        out0 = tmp0 ^ out2;
-        out5 = out0 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_9D(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out6 = in2 ^ in5;
-        tmp0 = in0 ^ in3;
-        out5 = in1 ^ in4 ^ in7;
-        out1 = out6 ^ in1;
-        out3 = tmp0 ^ out6;
-        out7 = tmp0 ^ in6;
-        out0 = out5 ^ in0;
-        out4 = out7 ^ in7;
-        out2 = out5 ^ out7 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_9E(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in1 ^ in4;
-        tmp0 = in0 ^ in5;
-        out6 = in2 ^ in6;
-        out7 = in0 ^ in3 ^ in7;
-        out4 = in0 ^ in4 ^ in6;
-        out5 = in1 ^ in5 ^ in7;
-        out1 = tmp0 ^ in2;
-        out3 = tmp0 ^ in7;
-        out2 = out4 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_9F(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out6 = in2;
-        out7 = in0 ^ in3;
-        tmp0 = in0 ^ in1;
-        out4 = in0 ^ in6;
-        out5 = in1 ^ in7;
-        out1 = tmp0 ^ in2 ^ in5;
-        out2 = out7 ^ in2 ^ in4 ^ in6;
-        out3 = out7 ^ in5 ^ in7;
-        out0 = tmp0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A0(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in6;
-        out2 = tmp0 ^ in7;
-        tmp1 = tmp0 ^ in5;
-        out6 = out2 ^ in3 ^ in4;
-        out0 = tmp1 ^ in3;
-        tmp2 = out0 ^ in2;
-        out3 = tmp2 ^ in7;
-        tmp3 = tmp2 ^ in1;
-        out5 = tmp3 ^ in0;
-        out4 = tmp3 ^ out6;
-        out7 = out5 ^ out6 ^ in1;
-        out1 = tmp1 ^ out4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A1(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in5;
-        tmp1 = tmp0 ^ in1;
-        tmp2 = tmp0 ^ in4;
-        out4 = tmp1 ^ in7;
-        out7 = tmp2 ^ in0;
-        out6 = tmp2 ^ out4 ^ in3;
-        out3 = out4 ^ in6;
-        out2 = out3 ^ in5;
-        out1 = out2 ^ in4;
-        out5 = out1 ^ out6 ^ in0;
-        out0 = tmp1 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A2(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in6;
-        tmp0 = in1 ^ in3 ^ in5;
-        out3 = tmp0 ^ in6;
-        out4 = tmp0 ^ in2 ^ in4;
-        out0 = out3 ^ in7;
-        out6 = out0 ^ in4;
-        out1 = out0 ^ out4 ^ in0;
-        out7 = out1 ^ in5;
-        out5 = out7 ^ in3 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A3(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in2 ^ in6;
-        out3 = in1 ^ in5 ^ in6;
-        tmp0 = out2 ^ in0;
-        out4 = out2 ^ out3 ^ in3;
-        tmp1 = tmp0 ^ in4;
-        out0 = tmp0 ^ out4 ^ in7;
-        out5 = tmp1 ^ in3;
-        out7 = tmp1 ^ in5;
-        out1 = tmp1 ^ in1 ^ in7;
-        out6 = tmp1 ^ out0 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A4(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in3;
-        tmp1 = in2 ^ in4;
-        tmp2 = in2 ^ in5;
-        tmp3 = in0 ^ in7;
-        out0 = tmp0 ^ in5;
-        out6 = tmp0 ^ in6 ^ in7;
-        out1 = tmp1 ^ in6;
-        out7 = tmp1 ^ tmp3;
-        out3 = tmp2 ^ in3;
-        tmp4 = tmp2 ^ out1;
-        out2 = tmp3 ^ in1;
-        out5 = tmp4 ^ out7;
-        out4 = tmp4 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A5(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in2 ^ in5;
-        tmp0 = in1 ^ in6;
-        tmp1 = in0 ^ in1;
-        tmp2 = in2 ^ in4;
-        out6 = in1 ^ in3 ^ in7;
-        out4 = tmp0 ^ in5;
-        out1 = tmp0 ^ tmp2;
-        out0 = tmp1 ^ in3 ^ in5;
-        out2 = tmp1 ^ in2 ^ in7;
-        out7 = tmp2 ^ in0;
-        out5 = tmp0 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A6(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in0;
-        out3 = in3 ^ in5 ^ in7;
-        out1 = in0 ^ in2 ^ in4 ^ in6;
-        out0 = out3 ^ in1;
-        out7 = out1 ^ in7;
-        out6 = out0 ^ in6;
-        out5 = out7 ^ in5;
-        out4 = out6 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A7(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in0 ^ in2;
-        out3 = in5 ^ in7;
-        out7 = out2 ^ in4 ^ in6;
-        out6 = out3 ^ in1 ^ in3;
-        out1 = out7 ^ in1;
-        out5 = out7 ^ in7;
-        out0 = out6 ^ in0;
-        out4 = out6 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A8(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in4;
-        tmp1 = in1 ^ in6;
-        tmp2 = in0 ^ in2 ^ in7;
-        out1 = tmp0 ^ in7;
-        out4 = tmp0 ^ in6;
-        out0 = tmp1 ^ in3;
-        out2 = tmp1 ^ in5;
-        out6 = tmp1 ^ in4;
-        out7 = tmp2 ^ in5;
-        out3 = tmp2 ^ out0 ^ in6;
-        out5 = out7 ^ in2 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_A9(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in2 ^ in6;
-        out6 = in1 ^ in4;
-        out7 = in0 ^ in2 ^ in5;
-        out5 = in0 ^ in3 ^ in7;
-        out2 = out4 ^ in1 ^ in5;
-        out1 = out6 ^ in2 ^ in7;
-        out0 = out2 ^ out7 ^ in3;
-        out3 = out1 ^ in0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_AA(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in2;
-        tmp1 = in1 ^ in3;
-        tmp2 = in6 ^ in7;
-        out1 = tmp0 ^ in4 ^ in7;
-        out3 = tmp1 ^ in0;
-        out0 = tmp1 ^ tmp2;
-        out2 = tmp2 ^ in5;
-        out7 = tmp0 ^ out2;
-        out6 = out1 ^ out7 ^ in1;
-        out5 = out0 ^ out6 ^ in0;
-        out4 = out5 ^ out7 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_AB(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in0 ^ in1;
-        tmp0 = in1 ^ in4;
-        tmp1 = in0 ^ in7;
-        out6 = tmp0 ^ in5;
-        out1 = tmp0 ^ tmp1 ^ in2;
-        out5 = tmp1 ^ in3 ^ in4;
-        out0 = tmp0 ^ out5 ^ in6;
-        out4 = out0 ^ out3 ^ in2;
-        out2 = out4 ^ in3 ^ in5;
-        out7 = tmp1 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_AC(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in1 ^ in3;
-        out1 = in2 ^ in4;
-        tmp0 = in0 ^ in2;
-        out4 = in4 ^ in7;
-        out5 = in0 ^ in5;
-        out6 = in1 ^ in6;
-        out7 = tmp0 ^ in7;
-        out3 = tmp0 ^ in3 ^ in6;
-        out2 = out5 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_AD(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in7;
-        out5 = in0;
-        out6 = in1;
-        out7 = in0 ^ in2;
-        out0 = in0 ^ in1 ^ in3;
-        out2 = out7 ^ in1 ^ in5;
-        out1 = in1 ^ in2 ^ in4;
-        out3 = out7 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_AE(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in3 ^ in4;
-        tmp0 = in0 ^ in4;
-        tmp1 = in0 ^ in7;
-        out0 = in1 ^ in3 ^ in7;
-        out1 = tmp0 ^ in2;
-        out5 = tmp0 ^ in5;
-        tmp2 = tmp1 ^ in6;
-        out2 = tmp1 ^ in5;
-        out3 = tmp2 ^ in3;
-        out7 = tmp2 ^ in2;
-        out6 = tmp2 ^ out2 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_AF(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in3;
-        tmp0 = in0 ^ in7;
-        out5 = in0 ^ in4;
-        out6 = in1 ^ in5;
-        out7 = in0 ^ in2 ^ in6;
-        out0 = tmp0 ^ in1 ^ in3;
-        out3 = tmp0 ^ in6;
-        out2 = tmp0 ^ in2 ^ in5;
-        out1 = out5 ^ in1 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B0(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in4;
-        tmp1 = in3 ^ in6;
-        out2 = tmp0 ^ in7;
-        tmp2 = tmp0 ^ tmp1;
-        out0 = tmp2 ^ in5;
-        out3 = tmp2 ^ in2;
-        out6 = out3 ^ in6;
-        tmp3 = out6 ^ in0 ^ in1;
-        out7 = tmp3 ^ in5;
-        out5 = tmp3 ^ out2;
-        out1 = out0 ^ out5 ^ in0;
-        out4 = tmp1 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B1(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in4;
-        out2 = tmp0 ^ in2 ^ in7;
-        tmp1 = out2 ^ in6;
-        out1 = tmp1 ^ in5;
-        out3 = tmp1 ^ in7;
-        out4 = tmp1 ^ in0;
-        out6 = out3 ^ in3;
-        out0 = out6 ^ in0 ^ in2 ^ in5;
-        out5 = tmp1 ^ out0 ^ in1;
-        out7 = tmp0 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B2(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in4;
-        tmp0 = in4 ^ in7;
-        tmp1 = in1 ^ in3 ^ in6;
-        out3 = tmp0 ^ tmp1;
-        tmp2 = tmp1 ^ in0;
-        out0 = out3 ^ in5;
-        out4 = tmp2 ^ in2;
-        tmp3 = out4 ^ in6;
-        out5 = tmp0 ^ tmp3;
-        out1 = tmp3 ^ out0;
-        tmp4 = out1 ^ in7;
-        out7 = tmp4 ^ in3;
-        out6 = tmp2 ^ tmp4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B3(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in2 ^ in4;
-        tmp0 = in0 ^ in5;
-        tmp1 = in1 ^ in6;
-        out3 = tmp1 ^ in4 ^ in7;
-        tmp2 = tmp0 ^ out3;
-        out0 = tmp2 ^ in3;
-        out1 = tmp2 ^ in2;
-        out5 = out0 ^ in2 ^ in6;
-        out7 = tmp1 ^ out5;
-        out4 = out7 ^ in1 ^ in5 ^ in7;
-        out6 = tmp0 ^ out4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B4(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in0 ^ in1;
-        out5 = out4 ^ in2;
-        tmp0 = out4 ^ in4;
-        out6 = out5 ^ in0 ^ in3;
-        out7 = tmp0 ^ out6;
-        out2 = tmp0 ^ in6 ^ in7;
-        out3 = out7 ^ in0 ^ in7;
-        out0 = out5 ^ out7 ^ in5;
-        out1 = out0 ^ out6 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B5(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        tmp1 = in2 ^ in4;
-        out4 = tmp0 ^ in4;
-        out3 = tmp1 ^ in7;
-        tmp2 = out4 ^ in5;
-        out7 = out3 ^ in0 ^ in3;
-        out0 = tmp2 ^ in3;
-        out2 = tmp0 ^ out3 ^ in6;
-        out5 = tmp1 ^ tmp2;
-        out6 = out2 ^ out7 ^ in2;
-        out1 = tmp0 ^ out0 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B6(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in3 ^ in4;
-        tmp0 = in1 ^ in2;
-        tmp1 = in0 ^ in4;
-        tmp2 = in3 ^ in5;
-        tmp3 = out3 ^ in1 ^ in7;
-        out5 = tmp0 ^ tmp1;
-        out6 = tmp0 ^ tmp2;
-        out2 = tmp1 ^ in6;
-        out4 = tmp1 ^ tmp3;
-        out0 = tmp3 ^ in5;
-        out1 = out2 ^ in2 ^ in5;
-        out7 = tmp2 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B7(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in4;
-        tmp0 = in0 ^ in4;
-        out2 = tmp0 ^ in2 ^ in6;
-        tmp1 = out2 ^ in7;
-        out1 = out2 ^ in1 ^ in5;
-        out7 = tmp1 ^ in3;
-        out5 = out1 ^ in6;
-        out6 = tmp0 ^ out1 ^ in3;
-        out0 = tmp1 ^ out6;
-        out4 = out0 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B8(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in4;
-        tmp1 = in2 ^ in5;
-        out2 = tmp0 ^ in5;
-        out4 = tmp1 ^ in0;
-        tmp2 = tmp1 ^ in7;
-        out6 = tmp2 ^ out2;
-        out7 = out4 ^ in3;
-        out1 = tmp2 ^ in4;
-        out3 = tmp0 ^ out7;
-        out0 = out3 ^ out4 ^ in6;
-        out5 = out0 ^ in0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_B9(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in2;
-        tmp1 = in4 ^ in5;
-        out4 = tmp0 ^ tmp1;
-        tmp2 = tmp0 ^ in3 ^ in7;
-        out3 = out4 ^ in1;
-        out7 = tmp2 ^ in5;
-        out2 = out3 ^ in0;
-        out1 = out2 ^ in7;
-        out6 = out1 ^ in5 ^ in6;
-        out0 = tmp2 ^ out6;
-        out5 = tmp1 ^ out0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_BA(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in5 ^ in7;
-        out2 = tmp0 ^ in4;
-        tmp1 = out2 ^ in2;
-        out1 = tmp1 ^ in0;
-        out6 = tmp1 ^ in1;
-        out4 = out1 ^ in3 ^ in4;
-        tmp2 = out4 ^ out6;
-        out7 = out4 ^ in6 ^ in7;
-        out5 = tmp2 ^ in6;
-        out3 = tmp0 ^ tmp2;
-        out0 = out6 ^ out7 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_BB(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in2 ^ in4 ^ in5 ^ in7;
-        tmp0 = out2 ^ in1;
-        out4 = out2 ^ in0 ^ in3;
-        out1 = tmp0 ^ in0;
-        out6 = tmp0 ^ in6;
-        out3 = out1 ^ in2;
-        tmp1 = out4 ^ out6 ^ in4;
-        out0 = tmp1 ^ in7;
-        out5 = tmp1 ^ in5;
-        out7 = tmp0 ^ tmp1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_BC(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in2;
-        tmp1 = in2 ^ in4;
-        out0 = in1 ^ in3 ^ in4;
-        out6 = in1 ^ in2 ^ in7;
-        out7 = tmp0 ^ in3;
-        out5 = tmp0 ^ out6 ^ in6;
-        out1 = tmp1 ^ in5;
-        tmp2 = out1 ^ out5 ^ in1;
-        out3 = tmp2 ^ in3;
-        out4 = tmp1 ^ tmp2;
-        out2 = tmp2 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_BD(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3;
-        tmp1 = in1 ^ in4;
-        out0 = tmp0 ^ tmp1;
-        out7 = tmp0 ^ in2 ^ in7;
-        out1 = tmp1 ^ in2 ^ in5;
-        tmp2 = out1 ^ in0;
-        out2 = tmp2 ^ in6;
-        out3 = out2 ^ in1 ^ in7;
-        out4 = out3 ^ in2;
-        out5 = tmp1 ^ out4;
-        out6 = tmp2 ^ out4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_BE(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3 ^ in6;
-        out4 = tmp0 ^ in5;
-        out7 = tmp0 ^ in2;
-        out3 = out4 ^ in4;
-        out1 = out3 ^ out7 ^ in0;
-        out2 = out3 ^ in3 ^ in7;
-        out0 = out2 ^ out4 ^ in1;
-        out5 = tmp0 ^ out0;
-        out6 = out1 ^ out5 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_BF(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in4;
-        out3 = tmp0 ^ in5 ^ in6;
-        out4 = out3 ^ in3;
-        tmp1 = out3 ^ in7;
-        out2 = tmp1 ^ in2;
-        out5 = tmp1 ^ in1;
-        tmp2 = out2 ^ in5;
-        out7 = tmp2 ^ in3 ^ in4;
-        tmp3 = tmp0 ^ out5;
-        out0 = tmp3 ^ out4;
-        out1 = tmp2 ^ tmp3;
-        out6 = tmp3 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C0(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out5 = in2 ^ in5;
-        tmp0 = in1 ^ in4;
-        tmp1 = in3 ^ in6;
-        out0 = out5 ^ in1;
-        out4 = tmp0 ^ in7;
-        out3 = tmp0 ^ tmp1;
-        out1 = tmp1 ^ in2;
-        out6 = tmp1 ^ in0;
-        out7 = out4 ^ in0;
-        out2 = out4 ^ out5 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C1(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out5 = in2;
-        tmp0 = in0 ^ in1;
-        out4 = in1 ^ in7;
-        out6 = in0 ^ in3;
-        out3 = in1 ^ in4 ^ in6;
-        tmp1 = tmp0 ^ in2;
-        out7 = tmp0 ^ in4;
-        out0 = tmp1 ^ in5;
-        out1 = tmp1 ^ out6 ^ in6;
-        out2 = out6 ^ out7 ^ in5 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C2(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in1 ^ in3 ^ in4;
-        tmp0 = in0 ^ in3 ^ in6;
-        out5 = in2 ^ in4 ^ in5;
-        tmp1 = out4 ^ in7;
-        out1 = tmp0 ^ in2;
-        out6 = tmp0 ^ in5;
-        out2 = out5 ^ in3;
-        out7 = tmp0 ^ tmp1;
-        out3 = tmp1 ^ in2 ^ in6;
-        out0 = tmp1 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C3(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in1 ^ in3;
-        tmp0 = in0 ^ in2;
-        tmp1 = in3 ^ in5;
-        out5 = in2 ^ in4;
-        tmp2 = tmp0 ^ out4;
-        out2 = tmp1 ^ in4;
-        out6 = tmp1 ^ in0;
-        out0 = tmp1 ^ tmp2 ^ in7;
-        out1 = tmp2 ^ in6;
-        out7 = out1 ^ out5 ^ in3;
-        out3 = tmp0 ^ out7 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C4(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in7;
-        out3 = tmp0 ^ in4;
-        tmp1 = tmp0 ^ in2;
-        out1 = tmp1 ^ in6;
-        out5 = tmp1 ^ in5;
-        out4 = out1 ^ out3 ^ in1;
-        out0 = out4 ^ in4 ^ in5;
-        out2 = out0 ^ out3 ^ in0;
-        out7 = out1 ^ out2 ^ in7;
-        out6 = tmp1 ^ out0 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C5(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in4 ^ in7;
-        tmp0 = in3 ^ in7;
-        out4 = in1 ^ in2 ^ in6;
-        out6 = in0 ^ in3 ^ in4;
-        out5 = tmp0 ^ in2;
-        out1 = tmp0 ^ out4;
-        out0 = out4 ^ in0 ^ in5;
-        out2 = out0 ^ out5 ^ in4;
-        out7 = tmp0 ^ out2 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C6(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in5 ^ in6;
-        tmp1 = in1 ^ in7;
-        tmp2 = tmp0 ^ in0;
-        tmp3 = tmp0 ^ tmp1;
-        tmp4 = tmp2 ^ in4;
-        out0 = tmp3 ^ in2;
-        out6 = tmp4 ^ in3;
-        out2 = out6 ^ in2;
-        out7 = tmp1 ^ tmp4;
-        out3 = tmp2 ^ out2;
-        tmp5 = out3 ^ in5;
-        out5 = tmp5 ^ in7;
-        out4 = tmp3 ^ tmp5;
-        out1 = tmp4 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C7(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in2 ^ in4;
-        tmp0 = in3 ^ in5;
-        tmp1 = out3 ^ in7;
-        out6 = tmp0 ^ in0 ^ in4;
-        out5 = tmp1 ^ in3;
-        out2 = out6 ^ in6;
-        out7 = out2 ^ in1 ^ in3;
-        out0 = tmp1 ^ out7;
-        out1 = tmp0 ^ out0;
-        out4 = out1 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C8(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out0 = in1 ^ in2;
-        out1 = in2 ^ in3;
-        tmp0 = in5 ^ in6;
-        tmp1 = in0 ^ in7;
-        out2 = out1 ^ in1 ^ in4;
-        out4 = tmp0 ^ in4;
-        out5 = tmp0 ^ in7;
-        out6 = tmp1 ^ in6;
-        out7 = tmp1 ^ in1;
-        out3 = out2 ^ in0 ^ in2 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_C9(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in5 ^ in6;
-        out7 = in0 ^ in1;
-        tmp0 = in1 ^ in3;
-        out5 = in6 ^ in7;
-        out6 = in0 ^ in7;
-        out0 = out7 ^ in2;
-        out3 = out7 ^ in4 ^ in5;
-        out1 = tmp0 ^ in2;
-        out2 = tmp0 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_CA(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in7;
-        tmp1 = in2 ^ in7;
-        tmp2 = tmp0 ^ in6;
-        out0 = tmp1 ^ in1;
-        tmp3 = tmp1 ^ in3;
-        out6 = tmp2 ^ in5;
-        out7 = tmp2 ^ in1;
-        out2 = tmp3 ^ in4;
-        out5 = out6 ^ in0 ^ in4;
-        out4 = out5 ^ in3;
-        out1 = tmp0 ^ tmp3;
-        out3 = tmp3 ^ out5 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_CB(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in4 ^ in7;
-        tmp1 = in5 ^ in7;
-        out7 = in0 ^ in1 ^ in6;
-        out5 = tmp0 ^ in6;
-        out2 = tmp0 ^ in3;
-        out6 = tmp1 ^ in0;
-        out4 = tmp1 ^ in3 ^ in6;
-        tmp2 = out5 ^ out7 ^ in2;
-        out1 = tmp2 ^ out2;
-        out0 = tmp2 ^ in4;
-        out3 = tmp2 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_CC(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in5;
-        tmp1 = in1 ^ in6;
-        out1 = in2 ^ in3 ^ in7;
-        out5 = tmp0 ^ in6;
-        out0 = tmp1 ^ in2;
-        tmp2 = out5 ^ in0 ^ in7;
-        out3 = tmp2 ^ in4;
-        out6 = tmp0 ^ out3;
-        out7 = tmp1 ^ tmp2 ^ in3;
-        tmp3 = out1 ^ out6;
-        out4 = tmp2 ^ tmp3;
-        out2 = tmp3 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_CD(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out5 = in3 ^ in6;
-        tmp0 = in0 ^ in1;
-        tmp1 = in2 ^ in7;
-        out6 = in0 ^ in4 ^ in7;
-        out2 = tmp0 ^ out5 ^ in4;
-        out7 = tmp0 ^ in5;
-        out0 = tmp0 ^ in2 ^ in6;
-        out4 = tmp1 ^ in5;
-        out1 = tmp1 ^ in1 ^ in3;
-        out3 = out6 ^ in5 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_CE(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in5;
-        tmp1 = tmp0 ^ in3;
-        out4 = tmp1 ^ in4;
-        tmp2 = out4 ^ in6;
-        out3 = tmp2 ^ in0;
-        out5 = tmp2 ^ in2;
-        out2 = out3 ^ in5 ^ in7;
-        out6 = tmp1 ^ out2;
-        out7 = out2 ^ out4 ^ in1;
-        out1 = tmp2 ^ out6;
-        out0 = tmp0 ^ out7 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_CF(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in6;
-        tmp1 = in0 ^ in1 ^ in5;
-        out4 = in2 ^ in3 ^ in5;
-        out5 = tmp0 ^ in4;
-        out7 = tmp1 ^ in6;
-        out1 = tmp1 ^ out4 ^ in7;
-        tmp2 = out5 ^ in0;
-        out2 = tmp2 ^ in7;
-        out3 = tmp2 ^ out4;
-        out6 = tmp0 ^ out2 ^ in5;
-        out0 = tmp0 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D0(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3;
-        tmp1 = in1 ^ in4;
-        tmp2 = in2 ^ in5;
-        out7 = tmp0 ^ tmp1;
-        out0 = tmp1 ^ tmp2;
-        tmp3 = tmp2 ^ in3;
-        out1 = tmp3 ^ in6;
-        tmp4 = out1 ^ in1;
-        out2 = tmp4 ^ in7;
-        out3 = out2 ^ in2;
-        out4 = tmp0 ^ out3;
-        out5 = tmp3 ^ out3;
-        out6 = tmp4 ^ out4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D1(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in5 ^ in6;
-        tmp1 = tmp0 ^ in1;
-        out1 = tmp1 ^ in2;
-        out2 = tmp1 ^ in7;
-        out3 = out2 ^ in3;
-        out5 = out3 ^ in2;
-        tmp2 = out3 ^ in0;
-        out4 = tmp2 ^ in4;
-        out7 = tmp0 ^ out4;
-        out6 = tmp2 ^ out1 ^ in6;
-        out0 = out2 ^ out6 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D2(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in5 ^ in6;
-        out2 = tmp0 ^ in2 ^ in3;
-        out1 = out2 ^ in0;
-        out3 = out2 ^ in1;
-        out4 = out1 ^ in1 ^ in2;
-        out6 = out1 ^ in6 ^ in7;
-        out7 = out4 ^ in4 ^ in5;
-        out5 = out4 ^ out6 ^ in4;
-        out0 = tmp0 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D3(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in3 ^ in5 ^ in6;
-        tmp0 = out2 ^ in2;
-        tmp1 = tmp0 ^ in1;
-        out1 = tmp1 ^ in0;
-        out3 = tmp1 ^ in3;
-        out4 = out1 ^ in2 ^ in4;
-        tmp2 = out4 ^ in5;
-        out7 = tmp2 ^ in7;
-        out0 = tmp0 ^ out7;
-        tmp3 = out0 ^ in0;
-        out5 = tmp3 ^ in6;
-        out6 = tmp2 ^ tmp3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D4(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in3 ^ in5;
-        tmp0 = in1 ^ in5;
-        tmp1 = tmp0 ^ in2;
-        out4 = tmp1 ^ in0;
-        tmp2 = tmp1 ^ in6;
-        out2 = out4 ^ in3 ^ in7;
-        out0 = tmp2 ^ in4;
-        out5 = tmp2 ^ out3;
-        out1 = tmp0 ^ out5 ^ in7;
-        out6 = tmp0 ^ out2 ^ in4;
-        out7 = tmp1 ^ out6 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D5(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in5;
-        tmp0 = in0 ^ in4;
-        tmp1 = tmp0 ^ in1 ^ in5;
-        out4 = tmp1 ^ in2;
-        out0 = out4 ^ in6;
-        tmp2 = tmp0 ^ out0;
-        out5 = tmp2 ^ in3;
-        out1 = out5 ^ in7;
-        out6 = tmp1 ^ out1;
-        out7 = tmp2 ^ out6;
-        out2 = out7 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D6(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in2 ^ in4 ^ in6;
-        out5 = tmp0 ^ in3;
-        out0 = tmp0 ^ in5 ^ in7;
-        out3 = out0 ^ out5 ^ in2;
-        tmp1 = out3 ^ in0;
-        out1 = tmp1 ^ in6;
-        out2 = tmp1 ^ in7;
-        out4 = tmp1 ^ in1;
-        out6 = tmp1 ^ in4;
-        out7 = tmp0 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D7(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in3;
-        out3 = in2 ^ in5 ^ in7;
-        out2 = tmp0 ^ in5;
-        tmp1 = tmp0 ^ out3 ^ in1;
-        out1 = tmp1 ^ in6;
-        out4 = tmp1 ^ in4;
-        tmp2 = out1 ^ in4;
-        out6 = tmp2 ^ in1;
-        out7 = tmp2 ^ in2;
-        out0 = tmp2 ^ in3;
-        out5 = tmp2 ^ in0 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D8(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in0;
-        out5 = in1;
-        tmp0 = in1 ^ in2;
-        out6 = in0 ^ in2;
-        out0 = tmp0 ^ in4;
-        tmp1 = tmp0 ^ in3;
-        out7 = tmp1 ^ out6;
-        out2 = tmp1 ^ in6;
-        out3 = out7 ^ in7;
-        out1 = tmp1 ^ in1 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_D9(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in0 ^ in4;
-        out5 = in1 ^ in5;
-        out2 = in1 ^ in3 ^ in6;
-        out3 = in0 ^ in1 ^ in7;
-        out6 = in0 ^ in2 ^ in6;
-        out0 = out4 ^ in1 ^ in2;
-        out1 = out5 ^ in2 ^ in3;
-        out7 = out3 ^ in3;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_DA(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out5 = in1 ^ in4;
-        tmp0 = in2 ^ in7;
-        tmp1 = in0 ^ in2 ^ in3;
-        out0 = tmp0 ^ out5;
-        out4 = tmp0 ^ tmp1;
-        out2 = tmp0 ^ in3 ^ in6;
-        out1 = tmp1 ^ in5;
-        out3 = tmp1 ^ in1;
-        out6 = out1 ^ in3;
-        out7 = out3 ^ in2 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_DB(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        tmp1 = in1 ^ in5;
-        tmp2 = in3 ^ in7;
-        out3 = tmp0 ^ in2;
-        out5 = tmp1 ^ in4;
-        out6 = tmp1 ^ out3 ^ in6;
-        out2 = tmp2 ^ in6;
-        tmp3 = tmp2 ^ in4;
-        tmp4 = out3 ^ in3;
-        out4 = tmp3 ^ in0;
-        out1 = tmp4 ^ in5;
-        out0 = tmp3 ^ tmp4;
-        out7 = tmp0 ^ out2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_DC(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in2;
-        tmp1 = in0 ^ in3;
-        out6 = tmp0 ^ in4;
-        tmp2 = tmp0 ^ in7;
-        out3 = tmp1 ^ in6;
-        tmp3 = tmp1 ^ in1;
-        out1 = tmp1 ^ tmp2 ^ in5;
-        out4 = tmp2 ^ in6;
-        out2 = tmp3 ^ in2;
-        out7 = tmp3 ^ in5;
-        out5 = tmp2 ^ out2;
-        out0 = out2 ^ out3 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_DD(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in0 ^ in6;
-        out2 = in0 ^ in1 ^ in3;
-        out6 = out3 ^ in2 ^ in4;
-        out7 = out2 ^ in5 ^ in7;
-        out0 = out6 ^ in1;
-        out4 = out6 ^ in7;
-        out5 = out7 ^ in0;
-        out1 = out5 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_DE(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in3 ^ in6;
-        tmp1 = in3 ^ in4 ^ in7;
-        out4 = tmp0 ^ in0;
-        out5 = tmp1 ^ in1;
-        out3 = out4 ^ in7;
-        out2 = out3 ^ in6;
-        out1 = out2 ^ in5;
-        out6 = tmp1 ^ out1;
-        out0 = tmp0 ^ out5;
-        out7 = out0 ^ out1 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_DF(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in0 ^ in3 ^ in7;
-        tmp0 = out2 ^ in1 ^ in5;
-        out1 = tmp0 ^ in2;
-        out7 = tmp0 ^ in6;
-        out5 = tmp0 ^ in0 ^ in4;
-        tmp1 = out1 ^ out5 ^ in6;
-        out4 = tmp1 ^ in3;
-        out6 = tmp1 ^ in5;
-        tmp2 = tmp1 ^ in7;
-        out0 = tmp2 ^ in1;
-        out3 = tmp2 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E0(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in1 ^ in7;
-        tmp0 = in2 ^ in4;
-        out4 = out3 ^ in3 ^ in5;
-        out2 = tmp0 ^ in1;
-        tmp1 = tmp0 ^ in6;
-        out0 = out4 ^ in2;
-        out6 = out4 ^ in0;
-        out1 = tmp1 ^ in3;
-        out5 = tmp1 ^ in0;
-        out7 = out5 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E1(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in1 ^ in4;
-        tmp0 = in1 ^ in7;
-        out3 = tmp0 ^ in3;
-        tmp1 = out3 ^ in5;
-        out4 = tmp1 ^ in4;
-        tmp2 = tmp1 ^ in0;
-        out0 = tmp2 ^ in2;
-        out6 = tmp2 ^ in6;
-        tmp3 = out0 ^ out4 ^ in6;
-        out5 = tmp3 ^ in5;
-        out7 = tmp0 ^ tmp3;
-        out1 = tmp2 ^ out5 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E2(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in1 ^ in2;
-        out4 = in1 ^ in5;
-        out2 = in2 ^ in4 ^ in7;
-        out5 = in0 ^ in2 ^ in6;
-        out0 = out3 ^ in3 ^ in5;
-        out7 = out3 ^ in0 ^ in4;
-        out6 = out2 ^ out7 ^ in3;
-        out1 = out5 ^ in3 ^ in4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E3(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in4 ^ in7;
-        tmp0 = in1 ^ in3;
-        out3 = tmp0 ^ in2;
-        tmp1 = out3 ^ in0;
-        out0 = tmp1 ^ in5;
-        tmp2 = tmp1 ^ in4;
-        out1 = tmp2 ^ in6;
-        tmp3 = tmp2 ^ in3;
-        out7 = tmp3 ^ in7;
-        out6 = out1 ^ out2 ^ in2;
-        tmp4 = tmp0 ^ out0;
-        out5 = tmp4 ^ in6;
-        out4 = tmp3 ^ tmp4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E4(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in6;
-        tmp0 = in0 ^ in4;
-        tmp1 = tmp0 ^ in2 ^ in6;
-        out2 = tmp1 ^ in1;
-        out7 = out2 ^ in5;
-        tmp2 = tmp0 ^ out7;
-        out4 = tmp2 ^ in3;
-        out0 = out4 ^ in7;
-        out6 = tmp1 ^ out0;
-        out5 = tmp2 ^ out6;
-        out1 = out5 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E5(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in3 ^ in6;
-        tmp0 = in0 ^ in1;
-        tmp1 = in5 ^ in7;
-        out2 = tmp0 ^ in4 ^ in6;
-        tmp2 = tmp1 ^ out2;
-        out6 = tmp2 ^ in3;
-        out7 = tmp2 ^ in2;
-        out0 = out6 ^ in2 ^ in4;
-        out5 = out6 ^ in1 ^ in2;
-        out1 = tmp0 ^ out5 ^ in5;
-        out4 = tmp1 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E6(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in2 ^ in6 ^ in7;
-        out2 = out3 ^ in0 ^ in4;
-        out4 = out3 ^ in1 ^ in5;
-        out1 = out2 ^ in3;
-        out7 = out2 ^ out4 ^ in2;
-        out0 = out4 ^ in3 ^ in7;
-        out5 = out1 ^ in4;
-        out6 = out0 ^ out2 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E7(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in3;
-        out3 = tmp0 ^ in6 ^ in7;
-        tmp1 = out3 ^ in0;
-        out5 = tmp1 ^ in5;
-        tmp2 = tmp1 ^ in4;
-        tmp3 = out5 ^ in7;
-        out1 = tmp2 ^ in1;
-        out0 = tmp3 ^ in1;
-        out6 = out1 ^ in2;
-        out2 = tmp0 ^ tmp2;
-        tmp4 = tmp3 ^ out6;
-        out4 = tmp4 ^ in6;
-        out7 = tmp4 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E8(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in3 ^ in6;
-        tmp0 = in4 ^ in7;
-        out1 = in2 ^ in3 ^ in4;
-        out5 = tmp0 ^ in0;
-        tmp1 = tmp0 ^ in1;
-        tmp2 = tmp1 ^ in5;
-        out0 = tmp1 ^ out1;
-        out2 = tmp2 ^ in2;
-        out6 = tmp2 ^ out5;
-        tmp3 = out6 ^ in6;
-        out3 = tmp3 ^ in7;
-        out7 = tmp3 ^ in2 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_E9(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        tmp1 = in3 ^ in6;
-        tmp2 = tmp0 ^ in6;
-        out4 = tmp1 ^ in4;
-        out6 = tmp2 ^ in5;
-        out7 = tmp2 ^ in2 ^ in7;
-        out3 = out6 ^ in3 ^ in7;
-        out0 = tmp1 ^ out7;
-        out2 = out3 ^ out4 ^ in0;
-        out5 = tmp0 ^ out2;
-        out1 = out0 ^ out5 ^ in5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_EA(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in6 ^ in7;
-        out5 = in0 ^ in7;
-        out6 = in0 ^ in1;
-        out0 = in1 ^ in2 ^ in3;
-        out2 = in2 ^ in4 ^ in5;
-        out7 = out6 ^ in2;
-        out1 = out0 ^ out6 ^ in4;
-        out3 = out7 ^ in5 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_EB(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in4 ^ in5;
-        tmp0 = in0 ^ in1;
-        out4 = in4 ^ in6 ^ in7;
-        out5 = in0 ^ in5 ^ in7;
-        out6 = tmp0 ^ in6;
-        tmp1 = tmp0 ^ in2;
-        out0 = tmp1 ^ in3;
-        out7 = tmp1 ^ in7;
-        out1 = out0 ^ in4;
-        out3 = out0 ^ in5 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_EC(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out3 = in0 ^ in5;
-        out4 = in2 ^ in3 ^ in7;
-        out5 = in0 ^ in3 ^ in4;
-        out6 = out3 ^ in1 ^ in4;
-        out1 = out4 ^ in4;
-        out0 = out4 ^ in1 ^ in6;
-        out2 = out0 ^ out5 ^ in5;
-        out7 = out2 ^ in4 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_ED(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in2 ^ in4;
-        tmp1 = in3 ^ in5;
-        out4 = tmp0 ^ in3 ^ in7;
-        out3 = tmp1 ^ in0;
-        out1 = out4 ^ in1;
-        out5 = out3 ^ in4;
-        out7 = out1 ^ out5 ^ in6;
-        out2 = tmp0 ^ out7;
-        out0 = tmp1 ^ out7;
-        out6 = out2 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_EE(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in2;
-        tmp0 = in0 ^ in1;
-        out5 = in0 ^ in3;
-        tmp1 = tmp0 ^ in2;
-        out6 = tmp0 ^ in4;
-        tmp2 = tmp1 ^ out5;
-        out7 = tmp1 ^ in5;
-        out1 = tmp2 ^ out6 ^ in7;
-        out0 = tmp2 ^ in6;
-        tmp3 = out7 ^ in1;
-        out3 = tmp3 ^ in7;
-        out2 = tmp3 ^ in4 ^ in6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_EF(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out4 = in2 ^ in4;
-        tmp0 = in0 ^ in5;
-        tmp1 = in4 ^ in6;
-        out5 = tmp0 ^ in3;
-        out2 = tmp0 ^ tmp1;
-        out6 = tmp1 ^ in0 ^ in1;
-        out3 = out5 ^ in2 ^ in7;
-        out7 = out3 ^ in1 ^ in3;
-        out0 = out4 ^ out6 ^ in3;
-        out1 = tmp1 ^ out0 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F0(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in2;
-        tmp1 = in4 ^ in5;
-        out2 = tmp0 ^ in6;
-        out3 = tmp1 ^ in1;
-        tmp2 = tmp1 ^ in7;
-        out1 = out2 ^ out3 ^ in3;
-        tmp3 = tmp0 ^ tmp2;
-        out0 = tmp3 ^ in3;
-        out5 = tmp3 ^ in0;
-        out4 = out1 ^ out5 ^ in4;
-        out7 = out4 ^ in2;
-        out6 = tmp2 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F1(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in1 ^ in6;
-        tmp0 = in3 ^ in5;
-        out3 = tmp0 ^ in1 ^ in4;
-        tmp1 = out3 ^ in2;
-        out1 = tmp1 ^ in6;
-        tmp2 = tmp1 ^ in0;
-        tmp3 = out1 ^ in5;
-        out0 = tmp2 ^ in7;
-        out6 = tmp2 ^ in4;
-        out7 = tmp3 ^ in0;
-        out5 = tmp0 ^ out0;
-        out4 = tmp3 ^ out5 ^ in1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F2(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in4 ^ in5;
-        out2 = in2 ^ in6 ^ in7;
-        tmp1 = tmp0 ^ in1;
-        tmp2 = tmp1 ^ in2;
-        out0 = tmp2 ^ in3;
-        out3 = tmp2 ^ in7;
-        out5 = out3 ^ in0 ^ in4;
-        tmp3 = tmp0 ^ out5;
-        out7 = tmp3 ^ in3;
-        out4 = tmp3 ^ out2;
-        out1 = out0 ^ out4 ^ in4;
-        out6 = tmp1 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F3(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in6 ^ in7;
-        tmp0 = in0 ^ in1;
-        out4 = tmp0 ^ in6;
-        tmp1 = tmp0 ^ in2;
-        out5 = tmp1 ^ in7;
-        out6 = tmp1 ^ in3;
-        out7 = out6 ^ in4;
-        out0 = out7 ^ in5;
-        out1 = out0 ^ in6;
-        out3 = out0 ^ in0 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F4(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in0 ^ in1 ^ in2;
-        tmp0 = out2 ^ in3;
-        out4 = tmp0 ^ in4;
-        out5 = out4 ^ in5;
-        out6 = out5 ^ in6;
-        out7 = out6 ^ in7;
-        out0 = out7 ^ in0;
-        out1 = out0 ^ in1;
-        out3 = tmp0 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F5(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in0 ^ in1;
-        tmp0 = out2 ^ in2;
-        out4 = tmp0 ^ in3;
-        out5 = out4 ^ in4;
-        out6 = out5 ^ in5;
-        out7 = out6 ^ in6;
-        out0 = out7 ^ in7;
-        out1 = out0 ^ in0;
-        out3 = tmp0 ^ out0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F6(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in7;
-        out2 = tmp0 ^ in2;
-        out4 = out2 ^ in1 ^ in4;
-        out7 = out4 ^ in3 ^ in5;
-        out5 = out7 ^ in4 ^ in7;
-        out0 = tmp0 ^ out7 ^ in6;
-        tmp1 = out0 ^ in1;
-        out6 = out0 ^ in0 ^ in5;
-        out3 = tmp1 ^ in3;
-        out1 = tmp0 ^ tmp1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F7(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in0 ^ in7;
-        tmp0 = out2 ^ in1;
-        out4 = tmp0 ^ in2;
-        out5 = out4 ^ in3 ^ in7;
-        out6 = out5 ^ in4;
-        out7 = out6 ^ in5;
-        out0 = out7 ^ in6;
-        out1 = out0 ^ in7;
-        out3 = tmp0 ^ out1;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F8(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in4;
-        tmp1 = in3 ^ in5;
-        tmp2 = tmp0 ^ in6;
-        out4 = tmp0 ^ tmp1;
-        out1 = tmp1 ^ in2 ^ in4;
-        out3 = tmp2 ^ in1;
-        out5 = out3 ^ in5;
-        out7 = out1 ^ out5 ^ in7;
-        out6 = tmp1 ^ out7;
-        out0 = tmp2 ^ out7;
-        out2 = out6 ^ in0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_F9(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in3 ^ in5;
-        tmp1 = in0 ^ in6;
-        out4 = tmp0 ^ in0;
-        tmp2 = tmp1 ^ in4;
-        tmp3 = tmp1 ^ in2;
-        out5 = tmp2 ^ in1;
-        out3 = out5 ^ in3;
-        tmp4 = tmp3 ^ out3;
-        out1 = tmp4 ^ in5;
-        out0 = tmp4 ^ in0 ^ in7;
-        out6 = tmp0 ^ out0 ^ in4;
-        out7 = tmp2 ^ tmp4;
-        out2 = tmp3 ^ out6;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_FA(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in1;
-        tmp1 = tmp0 ^ in2;
-        tmp2 = tmp0 ^ in5;
-        tmp3 = tmp1 ^ in7;
-        out5 = tmp2 ^ in6;
-        out6 = tmp3 ^ in6;
-        out7 = tmp3 ^ in3;
-        out3 = out6 ^ in4;
-        out2 = tmp1 ^ out5;
-        out4 = out2 ^ out3 ^ in1;
-        out0 = out4 ^ out7 ^ in5;
-        out1 = tmp2 ^ out0;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_FB(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in5 ^ in6;
-        tmp0 = in0 ^ in1;
-        out4 = in0 ^ in5 ^ in7;
-        out5 = tmp0 ^ in6;
-        tmp1 = tmp0 ^ in2;
-        out6 = tmp1 ^ in7;
-        out7 = tmp1 ^ in3;
-        out0 = out7 ^ in4;
-        out1 = out0 ^ in5;
-        out3 = out0 ^ in6 ^ in7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_FC(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in1 ^ in2;
-        tmp1 = in0 ^ in7;
-        out2 = tmp0 ^ tmp1 ^ in5;
-        out3 = tmp1 ^ in4;
-        tmp2 = out2 ^ in6;
-        out6 = tmp2 ^ in4;
-        out7 = tmp2 ^ in3;
-        out4 = out6 ^ in1 ^ in3;
-        tmp3 = out4 ^ in0;
-        out1 = tmp3 ^ in6;
-        out0 = tmp3 ^ in1 ^ in5;
-        out5 = tmp0 ^ out4;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_FD(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in5;
-        tmp1 = in1 ^ in7;
-        out2 = tmp0 ^ tmp1;
-        out6 = out2 ^ in2 ^ in4;
-        tmp2 = out6 ^ in0;
-        out1 = tmp2 ^ in3;
-        out0 = tmp0 ^ out1 ^ in6;
-        out5 = out0 ^ in2;
-        tmp3 = out5 ^ in1;
-        out3 = tmp3 ^ in6;
-        out7 = tmp2 ^ tmp3;
-        out4 = tmp1 ^ out7;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_FE(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3, tmp4;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        tmp0 = in0 ^ in2;
-        out2 = tmp0 ^ in5;
-        out3 = tmp0 ^ in4;
-        tmp1 = out3 ^ in6;
-        out4 = tmp1 ^ in5;
-        tmp2 = tmp1 ^ in1;
-        out6 = tmp2 ^ in7;
-        tmp3 = tmp2 ^ in0;
-        out0 = tmp3 ^ in3;
-        tmp4 = out0 ^ out4 ^ in7;
-        out5 = tmp4 ^ in6;
-        out7 = tmp4 ^ in2;
-        out1 = tmp3 ^ out5;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-static void gf8_muladd_FF(uint8_t * out, uint8_t * in, unsigned int width)
-{
-    unsigned int i;
-    uint64_t * in_ptr = (uint64_t *)in;
-    uint64_t * out_ptr = (uint64_t *)out;
-
-    for (i = 0; i < width; i++)
-    {
-        uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
-        uint64_t tmp0, tmp1, tmp2, tmp3;
-
-        uint64_t in0 = out_ptr[0];
-        uint64_t in1 = out_ptr[width];
-        uint64_t in2 = out_ptr[width * 2];
-        uint64_t in3 = out_ptr[width * 3];
-        uint64_t in4 = out_ptr[width * 4];
-        uint64_t in5 = out_ptr[width * 5];
-        uint64_t in6 = out_ptr[width * 6];
-        uint64_t in7 = out_ptr[width * 7];
-
-        out2 = in0 ^ in5;
-        tmp0 = in4 ^ in7;
-        tmp1 = out2 ^ in2;
-        out4 = tmp1 ^ in6;
-        out7 = tmp1 ^ in1 ^ in3;
-        out1 = tmp0 ^ out7;
-        tmp2 = out1 ^ in5;
-        out6 = tmp2 ^ in3;
-        tmp3 = tmp2 ^ in7;
-        out0 = tmp3 ^ in6;
-        out3 = tmp3 ^ in1;
-        out5 = tmp0 ^ out0 ^ in2;
-
-        out_ptr[0] = out0 ^ in_ptr[0];
-        out_ptr[width] = out1 ^ in_ptr[width];
-        out_ptr[width * 2] = out2 ^ in_ptr[width * 2];
-        out_ptr[width * 3] = out3 ^ in_ptr[width * 3];
-        out_ptr[width * 4] = out4 ^ in_ptr[width * 4];
-        out_ptr[width * 5] = out5 ^ in_ptr[width * 5];
-        out_ptr[width * 6] = out6 ^ in_ptr[width * 6];
-        out_ptr[width * 7] = out7 ^ in_ptr[width * 7];
-
-        in_ptr++;
-        out_ptr++;
-    }
-}
-
-void (* ec_gf_muladd[])(uint8_t * out, uint8_t * in, unsigned int width) =
-{
-    gf8_muladd_00, gf8_muladd_01, gf8_muladd_02, gf8_muladd_03,
-    gf8_muladd_04, gf8_muladd_05, gf8_muladd_06, gf8_muladd_07,
-    gf8_muladd_08, gf8_muladd_09, gf8_muladd_0A, gf8_muladd_0B,
-    gf8_muladd_0C, gf8_muladd_0D, gf8_muladd_0E, gf8_muladd_0F,
-    gf8_muladd_10, gf8_muladd_11, gf8_muladd_12, gf8_muladd_13,
-    gf8_muladd_14, gf8_muladd_15, gf8_muladd_16, gf8_muladd_17,
-    gf8_muladd_18, gf8_muladd_19, gf8_muladd_1A, gf8_muladd_1B,
-    gf8_muladd_1C, gf8_muladd_1D, gf8_muladd_1E, gf8_muladd_1F,
-    gf8_muladd_20, gf8_muladd_21, gf8_muladd_22, gf8_muladd_23,
-    gf8_muladd_24, gf8_muladd_25, gf8_muladd_26, gf8_muladd_27,
-    gf8_muladd_28, gf8_muladd_29, gf8_muladd_2A, gf8_muladd_2B,
-    gf8_muladd_2C, gf8_muladd_2D, gf8_muladd_2E, gf8_muladd_2F,
-    gf8_muladd_30, gf8_muladd_31, gf8_muladd_32, gf8_muladd_33,
-    gf8_muladd_34, gf8_muladd_35, gf8_muladd_36, gf8_muladd_37,
-    gf8_muladd_38, gf8_muladd_39, gf8_muladd_3A, gf8_muladd_3B,
-    gf8_muladd_3C, gf8_muladd_3D, gf8_muladd_3E, gf8_muladd_3F,
-    gf8_muladd_40, gf8_muladd_41, gf8_muladd_42, gf8_muladd_43,
-    gf8_muladd_44, gf8_muladd_45, gf8_muladd_46, gf8_muladd_47,
-    gf8_muladd_48, gf8_muladd_49, gf8_muladd_4A, gf8_muladd_4B,
-    gf8_muladd_4C, gf8_muladd_4D, gf8_muladd_4E, gf8_muladd_4F,
-    gf8_muladd_50, gf8_muladd_51, gf8_muladd_52, gf8_muladd_53,
-    gf8_muladd_54, gf8_muladd_55, gf8_muladd_56, gf8_muladd_57,
-    gf8_muladd_58, gf8_muladd_59, gf8_muladd_5A, gf8_muladd_5B,
-    gf8_muladd_5C, gf8_muladd_5D, gf8_muladd_5E, gf8_muladd_5F,
-    gf8_muladd_60, gf8_muladd_61, gf8_muladd_62, gf8_muladd_63,
-    gf8_muladd_64, gf8_muladd_65, gf8_muladd_66, gf8_muladd_67,
-    gf8_muladd_68, gf8_muladd_69, gf8_muladd_6A, gf8_muladd_6B,
-    gf8_muladd_6C, gf8_muladd_6D, gf8_muladd_6E, gf8_muladd_6F,
-    gf8_muladd_70, gf8_muladd_71, gf8_muladd_72, gf8_muladd_73,
-    gf8_muladd_74, gf8_muladd_75, gf8_muladd_76, gf8_muladd_77,
-    gf8_muladd_78, gf8_muladd_79, gf8_muladd_7A, gf8_muladd_7B,
-    gf8_muladd_7C, gf8_muladd_7D, gf8_muladd_7E, gf8_muladd_7F,
-    gf8_muladd_80, gf8_muladd_81, gf8_muladd_82, gf8_muladd_83,
-    gf8_muladd_84, gf8_muladd_85, gf8_muladd_86, gf8_muladd_87,
-    gf8_muladd_88, gf8_muladd_89, gf8_muladd_8A, gf8_muladd_8B,
-    gf8_muladd_8C, gf8_muladd_8D, gf8_muladd_8E, gf8_muladd_8F,
-    gf8_muladd_90, gf8_muladd_91, gf8_muladd_92, gf8_muladd_93,
-    gf8_muladd_94, gf8_muladd_95, gf8_muladd_96, gf8_muladd_97,
-    gf8_muladd_98, gf8_muladd_99, gf8_muladd_9A, gf8_muladd_9B,
-    gf8_muladd_9C, gf8_muladd_9D, gf8_muladd_9E, gf8_muladd_9F,
-    gf8_muladd_A0, gf8_muladd_A1, gf8_muladd_A2, gf8_muladd_A3,
-    gf8_muladd_A4, gf8_muladd_A5, gf8_muladd_A6, gf8_muladd_A7,
-    gf8_muladd_A8, gf8_muladd_A9, gf8_muladd_AA, gf8_muladd_AB,
-    gf8_muladd_AC, gf8_muladd_AD, gf8_muladd_AE, gf8_muladd_AF,
-    gf8_muladd_B0, gf8_muladd_B1, gf8_muladd_B2, gf8_muladd_B3,
-    gf8_muladd_B4, gf8_muladd_B5, gf8_muladd_B6, gf8_muladd_B7,
-    gf8_muladd_B8, gf8_muladd_B9, gf8_muladd_BA, gf8_muladd_BB,
-    gf8_muladd_BC, gf8_muladd_BD, gf8_muladd_BE, gf8_muladd_BF,
-    gf8_muladd_C0, gf8_muladd_C1, gf8_muladd_C2, gf8_muladd_C3,
-    gf8_muladd_C4, gf8_muladd_C5, gf8_muladd_C6, gf8_muladd_C7,
-    gf8_muladd_C8, gf8_muladd_C9, gf8_muladd_CA, gf8_muladd_CB,
-    gf8_muladd_CC, gf8_muladd_CD, gf8_muladd_CE, gf8_muladd_CF,
-    gf8_muladd_D0, gf8_muladd_D1, gf8_muladd_D2, gf8_muladd_D3,
-    gf8_muladd_D4, gf8_muladd_D5, gf8_muladd_D6, gf8_muladd_D7,
-    gf8_muladd_D8, gf8_muladd_D9, gf8_muladd_DA, gf8_muladd_DB,
-    gf8_muladd_DC, gf8_muladd_DD, gf8_muladd_DE, gf8_muladd_DF,
-    gf8_muladd_E0, gf8_muladd_E1, gf8_muladd_E2, gf8_muladd_E3,
-    gf8_muladd_E4, gf8_muladd_E5, gf8_muladd_E6, gf8_muladd_E7,
-    gf8_muladd_E8, gf8_muladd_E9, gf8_muladd_EA, gf8_muladd_EB,
-    gf8_muladd_EC, gf8_muladd_ED, gf8_muladd_EE, gf8_muladd_EF,
-    gf8_muladd_F0, gf8_muladd_F1, gf8_muladd_F2, gf8_muladd_F3,
-    gf8_muladd_F4, gf8_muladd_F5, gf8_muladd_F6, gf8_muladd_F7,
-    gf8_muladd_F8, gf8_muladd_F9, gf8_muladd_FA, gf8_muladd_FB,
-    gf8_muladd_FC, gf8_muladd_FD, gf8_muladd_FE, gf8_muladd_FF
-};
diff --git a/xlators/cluster/ec/src/ec-gf8.c b/xlators/cluster/ec/src/ec-gf8.c
new file mode 100644
index 00000000000..2665632706b
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-gf8.c
@@ -0,0 +1,5959 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#include "ec-gf8.h"
+
+static ec_gf_op_t ec_gf8_mul_00_ops[] = {
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_00 = {
+    0,
+    { 0, },
+    ec_gf8_mul_00_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_01_ops[] = {
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_01 = {
+    8,
+    { 0, 1, 2, 3, 4, 5, 6, 7, },
+    ec_gf8_mul_01_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_02_ops[] = {
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_02 = {
+    8,
+    { 7, 0, 1, 2, 3, 4, 5, 6, },
+    ec_gf8_mul_02_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_03_ops[] = {
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_COPY,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_03 = {
+    9,
+    { 0, 1, 2, 3, 4, 5, 6, 7, 8, },
+    ec_gf8_mul_03_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_04_ops[] = {
+    { EC_GF_OP_XOR3,   8,  6,  7 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_04 = {
+    9,
+    { 6, 7, 0, 1, 2, 3, 4, 5, 8, },
+    ec_gf8_mul_04_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_05_ops[] = {
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_05 = {
+    8,
+    { 0, 1, 2, 6, 7, 3, 4, 5, },
+    ec_gf8_mul_05_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_06_ops[] = {
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_COPY,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_06 = {
+    9,
+    { 7, 0, 1, 2, 8, 3, 4, 5, 6, },
+    ec_gf8_mul_06_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_07_ops[] = {
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_07 = {
+    8,
+    { 6, 0, 1, 3, 2, 4, 5, 7, },
+    ec_gf8_mul_07_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_08_ops[] = {
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR3,   8,  6,  7 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_08 = {
+    9,
+    { 5, 6, 7, 0, 1, 2, 3, 4, 8, },
+    ec_gf8_mul_08_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_09_ops[] = {
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_09 = {
+    8,
+    { 0, 1, 2, 3, 5, 6, 7, 4, },
+    ec_gf8_mul_09_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_0A_ops[] = {
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_0A = {
+    8,
+    { 5, 0, 1, 2, 6, 7, 3, 4, },
+    ec_gf8_mul_0A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_0B_ops[] = {
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_COPY,   9,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_COPY,   8,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR3,   3,  8,  6 },
+    { EC_GF_OP_XOR2,   1,  9,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_0B = {
+    10,
+    { 7, 1, 5, 2, 4, 3, 0, 6, 8, 9, },
+    ec_gf8_mul_0B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_0C_ops[] = {
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_COPY,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_0C = {
+    9,
+    { 5, 7, 0, 1, 8, 2, 3, 4, 6, },
+    ec_gf8_mul_0C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_0D_ops[] = {
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR3,   8,  2,  4 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR3,   2,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_0D = {
+    9,
+    { 5, 6, 7, 3, 1, 0, 2, 4, 8, },
+    ec_gf8_mul_0D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_0E_ops[] = {
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_0E = {
+    8,
+    { 7, 0, 6, 1, 3, 2, 4, 5, },
+    ec_gf8_mul_0E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_0F_ops[] = {
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_0F = {
+    8,
+    { 1, 0, 5, 6, 7, 2, 3, 4, },
+    ec_gf8_mul_0F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_10_ops[] = {
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_10 = {
+    8,
+    { 4, 5, 6, 7, 0, 1, 2, 3, },
+    ec_gf8_mul_10_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_11_ops[] = {
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_11 = {
+    8,
+    { 4, 1, 2, 6, 0, 5, 7, 3, },
+    ec_gf8_mul_11_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_12_ops[] = {
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_12 = {
+    8,
+    { 7, 0, 1, 2, 3, 5, 6, 4, },
+    ec_gf8_mul_12_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_13_ops[] = {
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR3,   8,  3,  7 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_13 = {
+    9,
+    { 4, 5, 2, 6, 0, 1, 7, 3, 8, },
+    ec_gf8_mul_13_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_14_ops[] = {
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_14 = {
+    8,
+    { 6, 7, 0, 1, 2, 4, 5, 3, },
+    ec_gf8_mul_14_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_15_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR3,   5,  8,  7 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_15 = {
+    9,
+    { 0, 1, 2, 4, 7, 6, 5, 3, 8, },
+    ec_gf8_mul_15_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_16_ops[] = {
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_16 = {
+    8,
+    { 6, 7, 4, 1, 2, 3, 5, 0, },
+    ec_gf8_mul_16_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_17_ops[] = {
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_17 = {
+    8,
+    { 5, 7, 0, 1, 3, 2, 4, 6, },
+    ec_gf8_mul_17_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_18_ops[] = {
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_18 = {
+    9,
+    { 4, 5, 7, 6, 0, 1, 2, 3, 8, },
+    ec_gf8_mul_18_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_19_ops[] = {
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_19 = {
+    8,
+    { 0, 5, 2, 6, 7, 1, 3, 4, },
+    ec_gf8_mul_19_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_1A_ops[] = {
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_1A = {
+    8,
+    { 7, 0, 4, 5, 3, 1, 2, 6, },
+    ec_gf8_mul_1A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_1B_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_1B = {
+    8,
+    { 7, 4, 5, 6, 3, 1, 2, 0, },
+    ec_gf8_mul_1B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_1C_ops[] = {
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_1C = {
+    8,
+    { 5, 4, 3, 0, 1, 7, 2, 6, },
+    ec_gf8_mul_1C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_1D_ops[] = {
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR3,   8,  4,  2 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_1D = {
+    9,
+    { 0, 7, 5, 8, 2, 3, 4, 1, 6, },
+    ec_gf8_mul_1D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_1E_ops[] = {
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_1E = {
+    8,
+    { 4, 7, 5, 1, 6, 0, 2, 3, },
+    ec_gf8_mul_1E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_1F_ops[] = {
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR3,   8,  3,  7 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_1F = {
+    9,
+    { 1, 4, 5, 6, 7, 0, 3, 2, 8, },
+    ec_gf8_mul_1F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_20_ops[] = {
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_20 = {
+    8,
+    { 7, 4, 5, 6, 3, 0, 1, 2, },
+    ec_gf8_mul_20_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_21_ops[] = {
+    { EC_GF_OP_COPY,   9,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR3,   8,  7,  5 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   4,  9,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_21 = {
+    10,
+    { 0, 1, 2, 7, 5, 4, 3, 6, 8, 9, },
+    ec_gf8_mul_21_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_22_ops[] = {
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_22 = {
+    8,
+    { 3, 0, 5, 2, 6, 4, 1, 7, },
+    ec_gf8_mul_22_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_23_ops[] = {
+    { EC_GF_OP_COPY,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_23 = {
+    9,
+    { 0, 4, 3, 2, 5, 6, 1, 8, 7, },
+    ec_gf8_mul_23_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_24_ops[] = {
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_24 = {
+    8,
+    { 6, 7, 0, 1, 2, 4, 5, 3, },
+    ec_gf8_mul_24_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_25_ops[] = {
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_25 = {
+    8,
+    { 2, 7, 0, 1, 3, 4, 5, 6, },
+    ec_gf8_mul_25_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_26_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_26 = {
+    8,
+    { 3, 4, 1, 2, 0, 5, 6, 7, },
+    ec_gf8_mul_26_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_27_ops[] = {
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_27 = {
+    8,
+    { 3, 0, 1, 2, 6, 7, 4, 5, },
+    ec_gf8_mul_27_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_28_ops[] = {
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_28 = {
+    8,
+    { 5, 6, 3, 0, 1, 2, 4, 7, },
+    ec_gf8_mul_28_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_29_ops[] = {
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_29 = {
+    8,
+    { 4, 6, 3, 5, 7, 0, 1, 2, },
+    ec_gf8_mul_29_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_2A_ops[] = {
+    { EC_GF_OP_COPY,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR3,   6,  8,  4 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_2A = {
+    9,
+    { 3, 4, 7, 2, 6, 5, 1, 0, 8, },
+    ec_gf8_mul_2A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_2B_ops[] = {
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_2B = {
+    8,
+    { 3, 4, 7, 5, 6, 0, 1, 2, },
+    ec_gf8_mul_2B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_2C_ops[] = {
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_2C = {
+    8,
+    { 5, 6, 7, 0, 2, 3, 4, 1, },
+    ec_gf8_mul_2C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_2D_ops[] = {
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR3,   8,  4,  6 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_2D = {
+    9,
+    { 7, 0, 3, 5, 1, 4, 2, 6, 8, },
+    ec_gf8_mul_2D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_2E_ops[] = {
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_COPY,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   8,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_2E = {
+    9,
+    { 5, 0, 7, 3, 2, 6, 4, 1, 8, },
+    ec_gf8_mul_2E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_2F_ops[] = {
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR3,   8,  7,  6 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_2F = {
+    9,
+    { 6, 3, 2, 5, 7, 0, 1, 4, 8, },
+    ec_gf8_mul_2F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_30_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR3,   6,  8,  7 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_30 = {
+    9,
+    { 3, 4, 7, 5, 0, 6, 1, 2, 8, },
+    ec_gf8_mul_30_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_31_ops[] = {
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_31 = {
+    8,
+    { 7, 1, 4, 5, 6, 0, 2, 3, },
+    ec_gf8_mul_31_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_32_ops[] = {
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_32 = {
+    8,
+    { 3, 4, 6, 7, 5, 0, 1, 2, },
+    ec_gf8_mul_32_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_33_ops[] = {
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_33 = {
+    8,
+    { 5, 4, 3, 0, 2, 1, 6, 7, },
+    ec_gf8_mul_33_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_34_ops[] = {
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_34 = {
+    8,
+    { 7, 5, 3, 0, 2, 4, 1, 6, },
+    ec_gf8_mul_34_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_35_ops[] = {
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_35 = {
+    8,
+    { 6, 7, 5, 4, 2, 0, 1, 3, },
+    ec_gf8_mul_35_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_36_ops[] = {
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_36 = {
+    8,
+    { 6, 7, 4, 1, 2, 3, 0, 5, },
+    ec_gf8_mul_36_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_37_ops[] = {
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR3,   8,  0,  1 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_37 = {
+    9,
+    { 6, 7, 2, 1, 0, 3, 4, 5, 8, },
+    ec_gf8_mul_37_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_38_ops[] = {
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR3,   8,  6,  7 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_38 = {
+    9,
+    { 4, 5, 6, 3, 0, 1, 7, 2, 8, },
+    ec_gf8_mul_38_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_39_ops[] = {
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_39 = {
+    8,
+    { 1, 6, 3, 0, 5, 2, 4, 7, },
+    ec_gf8_mul_39_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_3A_ops[] = {
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_3A = {
+    8,
+    { 3, 4, 7, 0, 5, 6, 1, 2, },
+    ec_gf8_mul_3A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_3B_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR3,   8,  7,  3 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_3B = {
+    9,
+    { 3, 0, 1, 7, 6, 2, 4, 8, 5, },
+    ec_gf8_mul_3B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_3C_ops[] = {
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_3C = {
+    8,
+    { 3, 6, 4, 1, 7, 2, 0, 5, },
+    ec_gf8_mul_3C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_3D_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_3D = {
+    8,
+    { 2, 3, 4, 5, 6, 7, 0, 1, },
+    ec_gf8_mul_3D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_3E_ops[] = {
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_3E = {
+    8,
+    { 6, 1, 2, 7, 0, 3, 5, 4, },
+    ec_gf8_mul_3E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_3F_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_COPY,  10,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_COPY,   9,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR3,   4,  9,  7 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   3, 10,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_3F = {
+    11,
+    { 1, 7, 6, 2, 4, 3, 5, 0, 8, 9, 10, },
+    ec_gf8_mul_3F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_40_ops[] = {
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR3,   8,  7,  6 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_40 = {
+    9,
+    { 5, 7, 4, 6, 2, 3, 0, 1, 8, },
+    ec_gf8_mul_40_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_41_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   8,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_41 = {
+    9,
+    { 0, 7, 6, 5, 3, 4, 8, 1, 2, },
+    ec_gf8_mul_41_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_42_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_42 = {
+    9,
+    { 2, 7, 1, 6, 4, 3, 0, 5, 8, },
+    ec_gf8_mul_42_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_43_ops[] = {
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_43 = {
+    8,
+    { 2, 6, 4, 1, 7, 3, 0, 5, },
+    ec_gf8_mul_43_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_44_ops[] = {
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_44 = {
+    8,
+    { 2, 3, 4, 1, 6, 5, 0, 7, },
+    ec_gf8_mul_44_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_45_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_45 = {
+    8,
+    { 2, 3, 0, 1, 7, 4, 5, 6, },
+    ec_gf8_mul_45_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_46_ops[] = {
+    { EC_GF_OP_XOR3,   8,  2,  4 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_46 = {
+    9,
+    { 2, 0, 1, 3, 4, 5, 6, 7, 8, },
+    ec_gf8_mul_46_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_47_ops[] = {
+    { EC_GF_OP_XOR3,   8,  0,  1 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_47 = {
+    9,
+    { 2, 3, 4, 5, 6, 7, 0, 1, 8, },
+    ec_gf8_mul_47_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_48_ops[] = {
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_48 = {
+    8,
+    { 4, 5, 6, 0, 1, 3, 7, 2, },
+    ec_gf8_mul_48_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_49_ops[] = {
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR3,   8,  0,  6 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR3,   1,  8,  5 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_49 = {
+    9,
+    { 7, 2, 4, 0, 3, 5, 1, 6, 8, },
+    ec_gf8_mul_49_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_4A_ops[] = {
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_4A = {
+    8,
+    { 5, 6, 7, 0, 1, 3, 4, 2, },
+    ec_gf8_mul_4A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_4B_ops[] = {
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR3,   8,  3,  7 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_4B = {
+    9,
+    { 5, 3, 6, 7, 0, 2, 4, 1, 8, },
+    ec_gf8_mul_4B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_4C_ops[] = {
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_4C = {
+    8,
+    { 5, 3, 4, 7, 0, 6, 2, 1, },
+    ec_gf8_mul_4C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_4D_ops[] = {
+    { EC_GF_OP_COPY,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR3,   9,  3,  1 },
+    { EC_GF_OP_XOR2,   5,  9,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR3,   0,  8,  2 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_4D = {
+    10,
+    { 0, 9, 3, 5, 6, 4, 7, 1, 2, 8, },
+    ec_gf8_mul_4D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_4E_ops[] = {
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_4E = {
+    8,
+    { 2, 3, 0, 1, 5, 6, 7, 4, },
+    ec_gf8_mul_4E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_4F_ops[] = {
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_4F = {
+    8,
+    { 0, 3, 5, 6, 1, 2, 7, 4, },
+    ec_gf8_mul_4F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_50_ops[] = {
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_50 = {
+    8,
+    { 4, 5, 7, 3, 0, 1, 2, 6, },
+    ec_gf8_mul_50_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_51_ops[] = {
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_51 = {
+    8,
+    { 0, 1, 7, 2, 3, 4, 5, 6, },
+    ec_gf8_mul_51_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_52_ops[] = {
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_COPY,   9,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR3,   3,  5,  8 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  9,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_52 = {
+    10,
+    { 2, 3, 1, 4, 6, 7, 0, 5, 8, 9, },
+    ec_gf8_mul_52_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_53_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_53 = {
+    8,
+    { 2, 0, 1, 4, 5, 6, 7, 3, },
+    ec_gf8_mul_53_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_54_ops[] = {
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_54 = {
+    8,
+    { 7, 3, 0, 4, 2, 6, 5, 1, },
+    ec_gf8_mul_54_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_55_ops[] = {
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_55 = {
+    8,
+    { 1, 5, 6, 4, 3, 7, 2, 0, },
+    ec_gf8_mul_55_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_56_ops[] = {
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_56 = {
+    8,
+    { 2, 3, 0, 4, 5, 6, 7, 1, },
+    ec_gf8_mul_56_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_57_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_57 = {
+    8,
+    { 2, 3, 0, 1, 4, 5, 6, 7, },
+    ec_gf8_mul_57_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_58_ops[] = {
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_58 = {
+    8,
+    { 4, 3, 2, 7, 0, 1, 5, 6, },
+    ec_gf8_mul_58_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_59_ops[] = {
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_59 = {
+    8,
+    { 7, 3, 5, 6, 1, 2, 0, 4, },
+    ec_gf8_mul_59_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_5A_ops[] = {
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_5A = {
+    8,
+    { 6, 7, 0, 1, 2, 3, 5, 4, },
+    ec_gf8_mul_5A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_5B_ops[] = {
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_5B = {
+    8,
+    { 6, 0, 7, 5, 2, 1, 3, 4, },
+    ec_gf8_mul_5B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_5C_ops[] = {
+    { EC_GF_OP_COPY,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_5C = {
+    9,
+    { 7, 5, 2, 4, 1, 0, 6, 3, 8, },
+    ec_gf8_mul_5C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_5D_ops[] = {
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_5D = {
+    8,
+    { 1, 3, 5, 4, 6, 7, 2, 0, },
+    ec_gf8_mul_5D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_5E_ops[] = {
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_5E = {
+    8,
+    { 4, 3, 6, 2, 5, 7, 0, 1, },
+    ec_gf8_mul_5E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_5F_ops[] = {
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_5F = {
+    8,
+    { 6, 1, 3, 4, 5, 7, 2, 0, },
+    ec_gf8_mul_5F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_60_ops[] = {
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_60 = {
+    8,
+    { 2, 3, 4, 7, 5, 6, 0, 1, },
+    ec_gf8_mul_60_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_61_ops[] = {
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_61 = {
+    8,
+    { 0, 5, 6, 7, 4, 2, 1, 3, },
+    ec_gf8_mul_61_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_62_ops[] = {
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_62 = {
+    8,
+    { 2, 0, 3, 4, 5, 6, 7, 1, },
+    ec_gf8_mul_62_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_63_ops[] = {
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_63 = {
+    8,
+    { 3, 4, 6, 5, 7, 0, 1, 2, },
+    ec_gf8_mul_63_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_64_ops[] = {
+    { EC_GF_OP_COPY,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   8,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_64 = {
+    9,
+    { 2, 3, 4, 6, 5, 7, 8, 1, 0, },
+    ec_gf8_mul_64_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_65_ops[] = {
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_65 = {
+    8,
+    { 2, 5, 1, 3, 4, 0, 6, 7, },
+    ec_gf8_mul_65_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_66_ops[] = {
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_66 = {
+    8,
+    { 2, 3, 1, 4, 5, 7, 0, 6, },
+    ec_gf8_mul_66_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_67_ops[] = {
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_67 = {
+    8,
+    { 2, 4, 5, 6, 7, 3, 1, 0, },
+    ec_gf8_mul_67_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_68_ops[] = {
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_68 = {
+    8,
+    { 5, 7, 2, 3, 0, 6, 4, 1, },
+    ec_gf8_mul_68_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_69_ops[] = {
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_69 = {
+    8,
+    { 0, 1, 3, 2, 4, 5, 7, 6, },
+    ec_gf8_mul_69_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_6A_ops[] = {
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_6A = {
+    8,
+    { 5, 7, 4, 6, 1, 2, 0, 3, },
+    ec_gf8_mul_6A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_6B_ops[] = {
+    { EC_GF_OP_COPY,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_6B = {
+    9,
+    { 6, 7, 2, 0, 3, 1, 5, 4, 8, },
+    ec_gf8_mul_6B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_6C_ops[] = {
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_6C = {
+    8,
+    { 5, 6, 7, 0, 1, 2, 3, 4, },
+    ec_gf8_mul_6C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_6D_ops[] = {
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR3,   8,  3,  4 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_6D = {
+    9,
+    { 3, 6, 7, 0, 4, 5, 1, 2, 8, },
+    ec_gf8_mul_6D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_6E_ops[] = {
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_6E = {
+    8,
+    { 5, 6, 3, 1, 7, 2, 0, 4, },
+    ec_gf8_mul_6E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_6F_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR3,   0,  8,  7 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_6F = {
+    9,
+    { 2, 6, 3, 7, 0, 1, 4, 5, 8, },
+    ec_gf8_mul_6F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_70_ops[] = {
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_70 = {
+    8,
+    { 3, 4, 5, 2, 6, 0, 1, 7, },
+    ec_gf8_mul_70_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_71_ops[] = {
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_71 = {
+    8,
+    { 4, 7, 5, 3, 6, 0, 2, 1, },
+    ec_gf8_mul_71_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_72_ops[] = {
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_72 = {
+    8,
+    { 0, 5, 2, 7, 4, 1, 3, 6, },
+    ec_gf8_mul_72_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_73_ops[] = {
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_73 = {
+    8,
+    { 6, 0, 1, 7, 4, 5, 2, 3, },
+    ec_gf8_mul_73_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_74_ops[] = {
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_74 = {
+    8,
+    { 3, 2, 1, 0, 4, 5, 6, 7, },
+    ec_gf8_mul_74_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_75_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_75 = {
+    8,
+    { 4, 5, 6, 7, 0, 1, 2, 3, },
+    ec_gf8_mul_75_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_76_ops[] = {
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR3,   8,  6,  2 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_76 = {
+    9,
+    { 2, 3, 0, 6, 5, 1, 7, 8, 4, },
+    ec_gf8_mul_76_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_77_ops[] = {
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_77 = {
+    8,
+    { 7, 4, 3, 6, 0, 1, 5, 2, },
+    ec_gf8_mul_77_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_78_ops[] = {
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR3,   8,  0,  2 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_78 = {
+    9,
+    { 4, 7, 3, 2, 5, 1, 6, 0, 8, },
+    ec_gf8_mul_78_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_79_ops[] = {
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR3,   8,  4,  7 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_79 = {
+    9,
+    { 4, 5, 7, 3, 1, 6, 2, 0, 8, },
+    ec_gf8_mul_79_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_7A_ops[] = {
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_7A = {
+    8,
+    { 1, 2, 3, 4, 5, 6, 7, 0, },
+    ec_gf8_mul_7A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_7B_ops[] = {
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR3,   8,  5,  3 },
+    { EC_GF_OP_XOR2,   8,  0,  0 },
+    { EC_GF_OP_COPY,   9,  4,  0 },
+    { EC_GF_OP_XOR2,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR3,   4,  1,  9 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_7B = {
+    10,
+    { 1, 2, 3, 4, 8, 5, 6, 0, 7, 9, },
+    ec_gf8_mul_7B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_7C_ops[] = {
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_7C = {
+    8,
+    { 2, 4, 1, 6, 3, 5, 7, 0, },
+    ec_gf8_mul_7C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_7D_ops[] = {
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_7D = {
+    8,
+    { 1, 0, 3, 5, 6, 7, 2, 4, },
+    ec_gf8_mul_7D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_7E_ops[] = {
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR3,   6,  2,  7 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_7E = {
+    9,
+    { 5, 1, 2, 0, 7, 3, 4, 6, 8, },
+    ec_gf8_mul_7E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_7F_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR3,   9,  7,  5 },
+    { EC_GF_OP_XOR2,   2,  9,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  9,  0 },
+    { EC_GF_OP_XOR3,   9,  6,  4 },
+    { EC_GF_OP_XOR2,   7,  9,  0 },
+    { EC_GF_OP_XOR2,   3,  9,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_7F = {
+    10,
+    { 4, 1, 0, 5, 6, 7, 2, 3, 8, 9, },
+    ec_gf8_mul_7F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_80_ops[] = {
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_80 = {
+    8,
+    { 7, 5, 6, 4, 1, 2, 3, 0, },
+    ec_gf8_mul_80_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_81_ops[] = {
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_81 = {
+    8,
+    { 2, 7, 4, 1, 5, 6, 3, 0, },
+    ec_gf8_mul_81_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_82_ops[] = {
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_COPY,   8,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR3,   5,  8,  7 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_82 = {
+    9,
+    { 6, 2, 7, 5, 1, 3, 4, 0, 8, },
+    ec_gf8_mul_82_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_83_ops[] = {
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_83 = {
+    8,
+    { 3, 5, 6, 7, 1, 2, 4, 0, },
+    ec_gf8_mul_83_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_84_ops[] = {
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_84 = {
+    8,
+    { 7, 6, 0, 4, 1, 5, 3, 2, },
+    ec_gf8_mul_84_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_85_ops[] = {
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_85 = {
+    8,
+    { 7, 6, 0, 3, 2, 4, 5, 1, },
+    ec_gf8_mul_85_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_86_ops[] = {
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_86 = {
+    8,
+    { 1, 2, 6, 4, 5, 7, 3, 0, },
+    ec_gf8_mul_86_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_87_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_COPY,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   8,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR3,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_87 = {
+    9,
+    { 1, 2, 3, 4, 5, 7, 6, 0, 8, },
+    ec_gf8_mul_87_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_88_ops[] = {
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_88 = {
+    8,
+    { 6, 7, 3, 1, 2, 4, 5, 0, },
+    ec_gf8_mul_88_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_89_ops[] = {
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR3,   8,  5,  2 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_89 = {
+    9,
+    { 2, 1, 6, 5, 7, 3, 4, 0, 8, },
+    ec_gf8_mul_89_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_8A_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_8A = {
+    8,
+    { 1, 2, 3, 0, 6, 7, 4, 5, },
+    ec_gf8_mul_8A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_8B_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_8B = {
+    8,
+    { 6, 1, 2, 3, 5, 7, 4, 0, },
+    ec_gf8_mul_8B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_8C_ops[] = {
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_8C = {
+    8,
+    { 1, 2, 0, 7, 3, 4, 5, 6, },
+    ec_gf8_mul_8C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_8D_ops[] = {
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_8D = {
+    8,
+    { 7, 1, 3, 2, 4, 5, 0, 6, },
+    ec_gf8_mul_8D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_8E_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_8E = {
+    8,
+    { 1, 2, 3, 4, 5, 6, 7, 0, },
+    ec_gf8_mul_8E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_8F_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_8F = {
+    8,
+    { 1, 2, 3, 4, 5, 6, 7, 0, },
+    ec_gf8_mul_8F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_90_ops[] = {
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_90 = {
+    8,
+    { 4, 5, 6, 7, 0, 1, 3, 2, },
+    ec_gf8_mul_90_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_91_ops[] = {
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_COPY,   9,  1,  0 },
+    { EC_GF_OP_COPY,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  9,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR3,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_91 = {
+    10,
+    { 2, 3, 1, 4, 0, 6, 7, 5, 8, 9, },
+    ec_gf8_mul_91_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_92_ops[] = {
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_92 = {
+    8,
+    { 6, 7, 0, 1, 2, 3, 5, 4, },
+    ec_gf8_mul_92_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_93_ops[] = {
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_93 = {
+    8,
+    { 6, 4, 5, 1, 7, 2, 3, 0, },
+    ec_gf8_mul_93_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_94_ops[] = {
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_94 = {
+    8,
+    { 7, 5, 0, 2, 6, 1, 3, 4, },
+    ec_gf8_mul_94_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_95_ops[] = {
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_95 = {
+    8,
+    { 7, 6, 1, 3, 0, 4, 5, 2, },
+    ec_gf8_mul_95_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_96_ops[] = {
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR3,   8,  0,  4 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_96 = {
+    9,
+    { 4, 0, 1, 6, 7, 2, 3, 5, 8, },
+    ec_gf8_mul_96_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_97_ops[] = {
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_COPY,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   8,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_97 = {
+    9,
+    { 4, 5, 3, 6, 7, 1, 2, 0, 8, },
+    ec_gf8_mul_97_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_98_ops[] = {
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_98 = {
+    8,
+    { 4, 2, 3, 6, 7, 5, 1, 0, },
+    ec_gf8_mul_98_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_99_ops[] = {
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_99 = {
+    8,
+    { 6, 5, 3, 7, 0, 1, 4, 2, },
+    ec_gf8_mul_99_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_9A_ops[] = {
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR3,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_9A = {
+    9,
+    { 6, 3, 4, 0, 5, 1, 2, 7, 8, },
+    ec_gf8_mul_9A_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_9B_ops[] = {
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_COPY,   9,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR3,   8,  3,  2 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  9,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_9B = {
+    10,
+    { 4, 5, 8, 6, 7, 1, 2, 0, 3, 9, },
+    ec_gf8_mul_9B_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_9C_ops[] = {
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_9C = {
+    8,
+    { 3, 2, 1, 0, 4, 5, 6, 7, },
+    ec_gf8_mul_9C_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_9D_ops[] = {
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_9D = {
+    8,
+    { 0, 1, 2, 3, 7, 4, 5, 6, },
+    ec_gf8_mul_9D_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_9E_ops[] = {
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_COPY,   8,  7,  0 },
+    { EC_GF_OP_XOR2,   8,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_9E = {
+    9,
+    { 4, 5, 3, 8, 6, 0, 2, 7, 1, },
+    ec_gf8_mul_9E_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_9F_ops[] = {
+    { EC_GF_OP_XOR3,   8,  1,  2 },
+    { EC_GF_OP_XOR2,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_9F = {
+    9,
+    { 4, 5, 6, 7, 0, 1, 2, 3, 8, },
+    ec_gf8_mul_9F_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A0_ops[] = {
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A0 = {
+    8,
+    { 3, 1, 6, 7, 5, 2, 4, 0, },
+    ec_gf8_mul_A0_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A1_ops[] = {
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR3,   8,  0,  6 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A1 = {
+    9,
+    { 7, 4, 1, 5, 6, 0, 2, 3, 8, },
+    ec_gf8_mul_A1_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A2_ops[] = {
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A2 = {
+    8,
+    { 7, 0, 6, 3, 2, 1, 4, 5, },
+    ec_gf8_mul_A2_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A3_ops[] = {
+    { EC_GF_OP_COPY,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A3 = {
+    9,
+    { 3, 7, 2, 6, 1, 4, 0, 5, 8, },
+    ec_gf8_mul_A3_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A4_ops[] = {
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A4 = {
+    8,
+    { 5, 6, 7, 2, 4, 3, 0, 1, },
+    ec_gf8_mul_A4_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A5_ops[] = {
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR3,   8,  5,  6 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A5 = {
+    9,
+    { 1, 4, 2, 5, 6, 7, 3, 0, 8, },
+    ec_gf8_mul_A5_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A6_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A6 = {
+    8,
+    { 1, 2, 0, 3, 4, 5, 6, 7, },
+    ec_gf8_mul_A6_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A7_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A7 = {
+    8,
+    { 0, 1, 2, 5, 6, 7, 3, 4, },
+    ec_gf8_mul_A7_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A8_ops[] = {
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_COPY,   9,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  9,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A8 = {
+    10,
+    { 1, 7, 5, 8, 6, 3, 4, 0, 2, 9, },
+    ec_gf8_mul_A8_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_A9_ops[] = {
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_A9 = {
+    8,
+    { 3, 7, 6, 1, 2, 0, 4, 5, },
+    ec_gf8_mul_A9_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_AA_ops[] = {
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_AA = {
+    8,
+    { 0, 4, 5, 3, 6, 7, 1, 2, },
+    ec_gf8_mul_AA_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_AB_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_COPY,   9,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   8,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR3,   3,  9,  7 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_AB = {
+    10,
+    { 2, 3, 8, 0, 5, 6, 1, 4, 7, 9, },
+    ec_gf8_mul_AB_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_AC_ops[] = {
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_AC = {
+    8,
+    { 3, 2, 1, 0, 4, 5, 6, 7, },
+    ec_gf8_mul_AC_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_AD_ops[] = {
+    { EC_GF_OP_XOR3,   8,  1,  2 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_AD = {
+    9,
+    { 3, 4, 5, 6, 7, 0, 1, 2, 8, },
+    ec_gf8_mul_AD_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_AE_ops[] = {
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_COPY,   8,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_AE = {
+    9,
+    { 7, 0, 5, 6, 3, 4, 1, 2, 8, },
+    ec_gf8_mul_AE_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_AF_ops[] = {
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_AF = {
+    8,
+    { 0, 1, 2, 7, 3, 4, 5, 6, },
+    ec_gf8_mul_AF_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B0_ops[] = {
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B0 = {
+    8,
+    { 4, 0, 7, 2, 3, 1, 6, 5, },
+    ec_gf8_mul_B0_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B1_ops[] = {
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_COPY,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR3,   5,  8,  1 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B1 = {
+    9,
+    { 2, 6, 4, 7, 0, 1, 3, 5, 8, },
+    ec_gf8_mul_B1_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B2_ops[] = {
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR3,   8,  4,  5 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B2 = {
+    9,
+    { 0, 7, 4, 5, 6, 1, 2, 3, 8, },
+    ec_gf8_mul_B2_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B3_ops[] = {
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_COPY,   9,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR3,   8,  6,  4 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   8,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR3,   1,  9,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B3 = {
+    10,
+    { 2, 3, 4, 5, 1, 6, 0, 7, 8, 9, },
+    ec_gf8_mul_B3_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B4_ops[] = {
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B4 = {
+    8,
+    { 5, 6, 7, 0, 1, 2, 3, 4, },
+    ec_gf8_mul_B4_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B5_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_COPY,   8,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR3,   4,  8,  3 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B5 = {
+    9,
+    { 3, 4, 0, 7, 1, 5, 6, 2, 8, },
+    ec_gf8_mul_B5_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B6_ops[] = {
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B6 = {
+    8,
+    { 5, 3, 6, 4, 7, 0, 1, 2, },
+    ec_gf8_mul_B6_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B7_ops[] = {
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B7 = {
+    8,
+    { 5, 0, 1, 4, 2, 6, 7, 3, },
+    ec_gf8_mul_B7_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B8_ops[] = {
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B8 = {
+    8,
+    { 6, 4, 5, 1, 2, 0, 7, 3, },
+    ec_gf8_mul_B8_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_B9_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR3,   0,  8,  2 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_B9 = {
+    9,
+    { 6, 7, 0, 2, 1, 4, 5, 3, 8, },
+    ec_gf8_mul_B9_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_BA_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_BA = {
+    8,
+    { 1, 2, 4, 3, 5, 6, 0, 7, },
+    ec_gf8_mul_BA_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_BB_ops[] = {
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_COPY,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   8,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   8,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_BB = {
+    9,
+    { 7, 2, 1, 8, 3, 5, 6, 4, 0, },
+    ec_gf8_mul_BB_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_BC_ops[] = {
+    { EC_GF_OP_COPY,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR3,   2,  8,  4 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_BC = {
+    9,
+    { 2, 6, 3, 4, 5, 1, 7, 0, 8, },
+    ec_gf8_mul_BC_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_BD_ops[] = {
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_BD = {
+    8,
+    { 4, 5, 0, 2, 7, 1, 6, 3, },
+    ec_gf8_mul_BD_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_BE_ops[] = {
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_BE = {
+    8,
+    { 0, 6, 7, 4, 5, 1, 3, 2, },
+    ec_gf8_mul_BE_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_BF_ops[] = {
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_BF = {
+    8,
+    { 5, 6, 1, 7, 3, 0, 2, 4, },
+    ec_gf8_mul_BF_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C0_ops[] = {
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C0 = {
+    8,
+    { 1, 2, 3, 4, 7, 5, 6, 0, },
+    ec_gf8_mul_C0_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C1_ops[] = {
+    { EC_GF_OP_XOR3,   8,  1,  2 },
+    { EC_GF_OP_XOR2,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C1 = {
+    9,
+    { 5, 6, 7, 4, 1, 2, 3, 0, 8, },
+    ec_gf8_mul_C1_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C2_ops[] = {
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C2 = {
+    8,
+    { 7, 6, 3, 0, 1, 4, 5, 2, },
+    ec_gf8_mul_C2_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C3_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR3,   0,  2,  6 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR3,   9,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  9,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C3 = {
+    10,
+    { 5, 6, 4, 7, 1, 2, 3, 0, 8, 9, },
+    ec_gf8_mul_C3_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C4_ops[] = {
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C4 = {
+    8,
+    { 0, 2, 1, 3, 4, 5, 6, 7, },
+    ec_gf8_mul_C4_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C5_ops[] = {
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C5 = {
+    8,
+    { 4, 3, 5, 7, 6, 2, 0, 1, },
+    ec_gf8_mul_C5_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C6_ops[] = {
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_COPY,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR3,   9,  5,  4 },
+    { EC_GF_OP_XOR2,   6,  9,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  9,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C6 = {
+    10,
+    { 6, 3, 0, 4, 5, 7, 2, 1, 8, 9, },
+    ec_gf8_mul_C6_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C7_ops[] = {
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C7 = {
+    8,
+    { 7, 0, 6, 2, 5, 3, 4, 1, },
+    ec_gf8_mul_C7_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C8_ops[] = {
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C8 = {
+    8,
+    { 1, 3, 2, 4, 6, 7, 5, 0, },
+    ec_gf8_mul_C8_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_C9_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_C9 = {
+    8,
+    { 2, 3, 4, 5, 6, 7, 0, 1, },
+    ec_gf8_mul_C9_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_CA_ops[] = {
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_CA = {
+    8,
+    { 1, 2, 5, 7, 3, 4, 0, 6, },
+    ec_gf8_mul_CA_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_CB_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_CB = {
+    8,
+    { 2, 3, 4, 5, 7, 6, 0, 1, },
+    ec_gf8_mul_CB_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_CC_ops[] = {
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_CC = {
+    8,
+    { 2, 7, 1, 0, 5, 6, 3, 4, },
+    ec_gf8_mul_CC_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_CD_ops[] = {
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_CD = {
+    8,
+    { 0, 6, 1, 2, 7, 3, 4, 5, },
+    ec_gf8_mul_CD_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_CE_ops[] = {
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_COPY,   8,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR3,   3,  6,  8 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR3,   8,  2,  3 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_CE = {
+    9,
+    { 5, 7, 3, 0, 2, 6, 4, 1, 8, },
+    ec_gf8_mul_CE_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_CF_ops[] = {
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_CF = {
+    8,
+    { 3, 6, 7, 0, 2, 4, 5, 1, },
+    ec_gf8_mul_CF_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D0_ops[] = {
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D0 = {
+    8,
+    { 5, 6, 7, 2, 0, 3, 1, 4, },
+    ec_gf8_mul_D0_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D1_ops[] = {
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR3,   8,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D1 = {
+    9,
+    { 5, 6, 3, 2, 0, 7, 4, 1, 8, },
+    ec_gf8_mul_D1_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D2_ops[] = {
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D2 = {
+    8,
+    { 7, 0, 2, 1, 3, 4, 6, 5, },
+    ec_gf8_mul_D2_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D3_ops[] = {
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_COPY,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   8,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D3 = {
+    9,
+    { 0, 3, 2, 8, 4, 6, 7, 1, 5, },
+    ec_gf8_mul_D3_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D4_ops[] = {
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_COPY,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR3,   1,  7,  8 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D4 = {
+    9,
+    { 4, 1, 7, 5, 0, 6, 3, 2, 8, },
+    ec_gf8_mul_D4_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D5_ops[] = {
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D5 = {
+    8,
+    { 6, 7, 4, 5, 2, 3, 1, 0, },
+    ec_gf8_mul_D5_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D6_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D6 = {
+    9,
+    { 0, 6, 2, 7, 1, 3, 4, 5, 8, },
+    ec_gf8_mul_D6_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D7_ops[] = {
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR3,   8,  3,  5 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR3,   6,  7,  8 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D7 = {
+    9,
+    { 3, 4, 6, 5, 0, 7, 1, 2, 8, },
+    ec_gf8_mul_D7_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D8_ops[] = {
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D8 = {
+    8,
+    { 4, 5, 6, 7, 0, 1, 2, 3, },
+    ec_gf8_mul_D8_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_D9_ops[] = {
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_D9 = {
+    8,
+    { 1, 2, 6, 7, 4, 5, 0, 3, },
+    ec_gf8_mul_D9_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_DA_ops[] = {
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR3,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_DA = {
+    9,
+    { 2, 5, 7, 1, 0, 4, 3, 6, 8, },
+    ec_gf8_mul_DA_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_DB_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_DB = {
+    9,
+    { 7, 5, 6, 2, 3, 4, 1, 0, 8, },
+    ec_gf8_mul_DB_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_DC_ops[] = {
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_DC = {
+    8,
+    { 4, 5, 2, 6, 7, 1, 0, 3, },
+    ec_gf8_mul_DC_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_DD_ops[] = {
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_DD = {
+    8,
+    { 1, 2, 3, 6, 7, 0, 4, 5, },
+    ec_gf8_mul_DD_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_DE_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_DE = {
+    8,
+    { 0, 5, 2, 6, 7, 1, 3, 4, },
+    ec_gf8_mul_DE_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_DF_ops[] = {
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   8,  3,  0 },
+    { EC_GF_OP_COPY,   9,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   8,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR3,   1,  9,  2 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_DF = {
+    10,
+    { 7, 2, 8, 4, 3, 1, 0, 6, 5, 9, },
+    ec_gf8_mul_DF_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E0_ops[] = {
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E0 = {
+    8,
+    { 2, 3, 4, 7, 5, 6, 0, 1, },
+    ec_gf8_mul_E0_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E1_ops[] = {
+    { EC_GF_OP_COPY,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   8,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR3,   9,  5,  3 },
+    { EC_GF_OP_XOR2,   0,  9,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  9,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E1 = {
+    10,
+    { 0, 7, 1, 3, 4, 5, 6, 2, 8, 9, },
+    ec_gf8_mul_E1_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E2_ops[] = {
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E2 = {
+    8,
+    { 2, 3, 7, 1, 5, 6, 0, 4, },
+    ec_gf8_mul_E2_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E3_ops[] = {
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR3,   8,  2,  7 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  8,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR3,   6,  8,  4 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E3 = {
+    9,
+    { 5, 4, 7, 2, 1, 3, 6, 0, 8, },
+    ec_gf8_mul_E3_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E4_ops[] = {
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E4 = {
+    8,
+    { 7, 0, 1, 6, 3, 4, 2, 5, },
+    ec_gf8_mul_E4_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E5_ops[] = {
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E5 = {
+    9,
+    { 4, 5, 3, 6, 7, 1, 0, 2, 8, },
+    ec_gf8_mul_E5_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E6_ops[] = {
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E6 = {
+    8,
+    { 5, 4, 3, 6, 7, 0, 1, 2, },
+    ec_gf8_mul_E6_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E7_ops[] = {
+    { EC_GF_OP_COPY,   8,  6,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR3,   9,  0,  6 },
+    { EC_GF_OP_XOR2,   4,  9,  0 },
+    { EC_GF_OP_XOR2,   5,  9,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E7 = {
+    10,
+    { 1, 4, 3, 6, 7, 5, 2, 0, 8, 9, },
+    ec_gf8_mul_E7_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E8_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E8 = {
+    8,
+    { 1, 4, 2, 7, 3, 0, 5, 6, },
+    ec_gf8_mul_E8_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_E9_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_COPY,   8,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR3,   1,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_E9 = {
+    9,
+    { 6, 2, 0, 3, 4, 1, 5, 7, 8, },
+    ec_gf8_mul_E9_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_EA_ops[] = {
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_EA = {
+    8,
+    { 3, 4, 5, 6, 7, 0, 1, 2, },
+    ec_gf8_mul_EA_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_EB_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_EB = {
+    8,
+    { 3, 4, 5, 6, 7, 0, 1, 2, },
+    ec_gf8_mul_EB_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_EC_ops[] = {
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR3,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_EC = {
+    9,
+    { 7, 4, 3, 0, 2, 5, 1, 6, 8, },
+    ec_gf8_mul_EC_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_ED_ops[] = {
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_ED = {
+    8,
+    { 5, 6, 7, 0, 1, 4, 3, 2, },
+    ec_gf8_mul_ED_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_EE_ops[] = {
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR3,   8,  2,  3 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   8,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_EE = {
+    9,
+    { 6, 4, 5, 7, 2, 3, 0, 1, 8, },
+    ec_gf8_mul_EE_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_EF_ops[] = {
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_COPY,   8,  0,  0 },
+    { EC_GF_OP_XOR2,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   6,  8,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_EF = {
+    9,
+    { 6, 4, 5, 7, 2, 0, 3, 1, 8, },
+    ec_gf8_mul_EF_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F0_ops[] = {
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR3,   8,  3,  6 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   1,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F0 = {
+    9,
+    { 3, 4, 6, 1, 2, 0, 5, 7, 8, },
+    ec_gf8_mul_F0_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F1_ops[] = {
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_COPY,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_COPY,   9,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   9,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  9,  0 },
+    { EC_GF_OP_XOR2,   4,  9,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR3,   9,  8,  7 },
+    { EC_GF_OP_XOR2,   1,  9,  0 },
+    { EC_GF_OP_XOR2,   5,  9,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F1 = {
+    10,
+    { 7, 2, 6, 3, 5, 1, 4, 0, 8, 9, },
+    ec_gf8_mul_F1_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F2_ops[] = {
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_XOR3,   8,  6,  4 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F2 = {
+    9,
+    { 1, 0, 6, 7, 4, 5, 2, 3, 8, },
+    ec_gf8_mul_F2_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F3_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F3 = {
+    8,
+    { 5, 6, 7, 0, 1, 2, 3, 4, },
+    ec_gf8_mul_F3_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F4_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F4 = {
+    8,
+    { 0, 1, 2, 3, 4, 5, 6, 7, },
+    ec_gf8_mul_F4_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F5_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F5 = {
+    8,
+    { 7, 0, 1, 2, 3, 4, 5, 6, },
+    ec_gf8_mul_F5_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F6_ops[] = {
+    { EC_GF_OP_XOR2,   3,  1,  0 },
+    { EC_GF_OP_COPY,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_COPY,   9,  3,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   9,  4,  0 },
+    { EC_GF_OP_XOR2,   4,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  9,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR3,   7,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F6 = {
+    10,
+    { 0, 6, 2, 7, 4, 3, 5, 9, 1, 8, },
+    ec_gf8_mul_F6_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F7_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F7 = {
+    8,
+    { 6, 7, 0, 1, 2, 3, 4, 5, },
+    ec_gf8_mul_F7_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F8_ops[] = {
+    { EC_GF_OP_XOR2,   4,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  5,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F8 = {
+    8,
+    { 6, 2, 0, 1, 4, 5, 3, 7, },
+    ec_gf8_mul_F8_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_F9_ops[] = {
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  4,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR3,   8,  7,  1 },
+    { EC_GF_OP_XOR2,   1,  3,  0 },
+    { EC_GF_OP_XOR2,   4,  8,  0 },
+    { EC_GF_OP_XOR2,   5,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_F9 = {
+    9,
+    { 4, 1, 7, 6, 0, 3, 5, 2, 8, },
+    ec_gf8_mul_F9_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_FA_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  2,  0 },
+    { EC_GF_OP_XOR2,   1,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_FA = {
+    8,
+    { 0, 1, 2, 4, 5, 6, 7, 3, },
+    ec_gf8_mul_FA_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_FB_ops[] = {
+    { EC_GF_OP_XOR2,   1,  0,  0 },
+    { EC_GF_OP_XOR2,   2,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   3,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  7,  0 },
+    { EC_GF_OP_XOR2,   2,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  6,  0 },
+    { EC_GF_OP_XOR2,   7,  6,  0 },
+    { EC_GF_OP_XOR2,   4,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  5,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_FB = {
+    8,
+    { 4, 5, 6, 7, 0, 1, 2, 3, },
+    ec_gf8_mul_FB_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_FC_ops[] = {
+    { EC_GF_OP_XOR2,   7,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  1,  0 },
+    { EC_GF_OP_COPY,   9,  3,  0 },
+    { EC_GF_OP_XOR3,   8,  5,  7 },
+    { EC_GF_OP_XOR2,   3,  6,  0 },
+    { EC_GF_OP_XOR2,   8,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  8,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   3,  4,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   6,  0,  0 },
+    { EC_GF_OP_XOR3,   0,  9,  2 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_FC = {
+    10,
+    { 5, 6, 3, 7, 1, 8, 0, 4, 2, 9, },
+    ec_gf8_mul_FC_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_FD_ops[] = {
+    { EC_GF_OP_XOR2,   7,  1,  0 },
+    { EC_GF_OP_COPY,   8,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  0,  0 },
+    { EC_GF_OP_XOR2,   7,  5,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  5,  0 },
+    { EC_GF_OP_XOR2,   1,  2,  0 },
+    { EC_GF_OP_XOR2,   0,  1,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR3,   1,  8,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_FD = {
+    9,
+    { 5, 3, 7, 6, 1, 2, 4, 0, 8, },
+    ec_gf8_mul_FD_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_FE_ops[] = {
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_COPY,   8,  2,  0 },
+    { EC_GF_OP_XOR2,   2,  4,  0 },
+    { EC_GF_OP_XOR2,   6,  2,  0 },
+    { EC_GF_OP_XOR2,   8,  5,  0 },
+    { EC_GF_OP_XOR2,   5,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  1,  0 },
+    { EC_GF_OP_XOR2,   0,  6,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   7,  8,  0 },
+    { EC_GF_OP_XOR2,   3,  0,  0 },
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR2,   0,  4,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_FE = {
+    9,
+    { 3, 4, 8, 2, 5, 0, 6, 1, 7, },
+    ec_gf8_mul_FE_ops
+};
+
+static ec_gf_op_t ec_gf8_mul_FF_ops[] = {
+    { EC_GF_OP_XOR2,   4,  7,  0 },
+    { EC_GF_OP_COPY,   9,  0,  0 },
+    { EC_GF_OP_COPY,   8,  4,  0 },
+    { EC_GF_OP_XOR2,   9,  1,  0 },
+    { EC_GF_OP_XOR2,   4,  2,  0 },
+    { EC_GF_OP_XOR2,   9,  4,  0 },
+    { EC_GF_OP_XOR2,   0,  5,  0 },
+    { EC_GF_OP_XOR2,   2,  0,  0 },
+    { EC_GF_OP_XOR2,   3,  9,  0 },
+    { EC_GF_OP_XOR2,   7,  3,  0 },
+    { EC_GF_OP_XOR2,   2,  6,  0 },
+    { EC_GF_OP_XOR2,   5,  3,  0 },
+    { EC_GF_OP_XOR2,   6,  7,  0 },
+    { EC_GF_OP_XOR2,   1,  7,  0 },
+    { EC_GF_OP_XOR3,   3,  8,  5 },
+    { EC_GF_OP_XOR2,   4,  6,  0 },
+    { EC_GF_OP_END,    0,  0,  0 }
+};
+
+static ec_gf_mul_t ec_gf8_mul_FF = {
+    10,
+    { 6, 5, 0, 1, 2, 4, 9, 3, 7, 8, },
+    ec_gf8_mul_FF_ops
+};
+
+ec_gf_mul_t *ec_gf8_mul[] = {
+    &ec_gf8_mul_00, &ec_gf8_mul_01, &ec_gf8_mul_02, &ec_gf8_mul_03,
+    &ec_gf8_mul_04, &ec_gf8_mul_05, &ec_gf8_mul_06, &ec_gf8_mul_07,
+    &ec_gf8_mul_08, &ec_gf8_mul_09, &ec_gf8_mul_0A, &ec_gf8_mul_0B,
+    &ec_gf8_mul_0C, &ec_gf8_mul_0D, &ec_gf8_mul_0E, &ec_gf8_mul_0F,
+    &ec_gf8_mul_10, &ec_gf8_mul_11, &ec_gf8_mul_12, &ec_gf8_mul_13,
+    &ec_gf8_mul_14, &ec_gf8_mul_15, &ec_gf8_mul_16, &ec_gf8_mul_17,
+    &ec_gf8_mul_18, &ec_gf8_mul_19, &ec_gf8_mul_1A, &ec_gf8_mul_1B,
+    &ec_gf8_mul_1C, &ec_gf8_mul_1D, &ec_gf8_mul_1E, &ec_gf8_mul_1F,
+    &ec_gf8_mul_20, &ec_gf8_mul_21, &ec_gf8_mul_22, &ec_gf8_mul_23,
+    &ec_gf8_mul_24, &ec_gf8_mul_25, &ec_gf8_mul_26, &ec_gf8_mul_27,
+    &ec_gf8_mul_28, &ec_gf8_mul_29, &ec_gf8_mul_2A, &ec_gf8_mul_2B,
+    &ec_gf8_mul_2C, &ec_gf8_mul_2D, &ec_gf8_mul_2E, &ec_gf8_mul_2F,
+    &ec_gf8_mul_30, &ec_gf8_mul_31, &ec_gf8_mul_32, &ec_gf8_mul_33,
+    &ec_gf8_mul_34, &ec_gf8_mul_35, &ec_gf8_mul_36, &ec_gf8_mul_37,
+    &ec_gf8_mul_38, &ec_gf8_mul_39, &ec_gf8_mul_3A, &ec_gf8_mul_3B,
+    &ec_gf8_mul_3C, &ec_gf8_mul_3D, &ec_gf8_mul_3E, &ec_gf8_mul_3F,
+    &ec_gf8_mul_40, &ec_gf8_mul_41, &ec_gf8_mul_42, &ec_gf8_mul_43,
+    &ec_gf8_mul_44, &ec_gf8_mul_45, &ec_gf8_mul_46, &ec_gf8_mul_47,
+    &ec_gf8_mul_48, &ec_gf8_mul_49, &ec_gf8_mul_4A, &ec_gf8_mul_4B,
+    &ec_gf8_mul_4C, &ec_gf8_mul_4D, &ec_gf8_mul_4E, &ec_gf8_mul_4F,
+    &ec_gf8_mul_50, &ec_gf8_mul_51, &ec_gf8_mul_52, &ec_gf8_mul_53,
+    &ec_gf8_mul_54, &ec_gf8_mul_55, &ec_gf8_mul_56, &ec_gf8_mul_57,
+    &ec_gf8_mul_58, &ec_gf8_mul_59, &ec_gf8_mul_5A, &ec_gf8_mul_5B,
+    &ec_gf8_mul_5C, &ec_gf8_mul_5D, &ec_gf8_mul_5E, &ec_gf8_mul_5F,
+    &ec_gf8_mul_60, &ec_gf8_mul_61, &ec_gf8_mul_62, &ec_gf8_mul_63,
+    &ec_gf8_mul_64, &ec_gf8_mul_65, &ec_gf8_mul_66, &ec_gf8_mul_67,
+    &ec_gf8_mul_68, &ec_gf8_mul_69, &ec_gf8_mul_6A, &ec_gf8_mul_6B,
+    &ec_gf8_mul_6C, &ec_gf8_mul_6D, &ec_gf8_mul_6E, &ec_gf8_mul_6F,
+    &ec_gf8_mul_70, &ec_gf8_mul_71, &ec_gf8_mul_72, &ec_gf8_mul_73,
+    &ec_gf8_mul_74, &ec_gf8_mul_75, &ec_gf8_mul_76, &ec_gf8_mul_77,
+    &ec_gf8_mul_78, &ec_gf8_mul_79, &ec_gf8_mul_7A, &ec_gf8_mul_7B,
+    &ec_gf8_mul_7C, &ec_gf8_mul_7D, &ec_gf8_mul_7E, &ec_gf8_mul_7F,
+    &ec_gf8_mul_80, &ec_gf8_mul_81, &ec_gf8_mul_82, &ec_gf8_mul_83,
+    &ec_gf8_mul_84, &ec_gf8_mul_85, &ec_gf8_mul_86, &ec_gf8_mul_87,
+    &ec_gf8_mul_88, &ec_gf8_mul_89, &ec_gf8_mul_8A, &ec_gf8_mul_8B,
+    &ec_gf8_mul_8C, &ec_gf8_mul_8D, &ec_gf8_mul_8E, &ec_gf8_mul_8F,
+    &ec_gf8_mul_90, &ec_gf8_mul_91, &ec_gf8_mul_92, &ec_gf8_mul_93,
+    &ec_gf8_mul_94, &ec_gf8_mul_95, &ec_gf8_mul_96, &ec_gf8_mul_97,
+    &ec_gf8_mul_98, &ec_gf8_mul_99, &ec_gf8_mul_9A, &ec_gf8_mul_9B,
+    &ec_gf8_mul_9C, &ec_gf8_mul_9D, &ec_gf8_mul_9E, &ec_gf8_mul_9F,
+    &ec_gf8_mul_A0, &ec_gf8_mul_A1, &ec_gf8_mul_A2, &ec_gf8_mul_A3,
+    &ec_gf8_mul_A4, &ec_gf8_mul_A5, &ec_gf8_mul_A6, &ec_gf8_mul_A7,
+    &ec_gf8_mul_A8, &ec_gf8_mul_A9, &ec_gf8_mul_AA, &ec_gf8_mul_AB,
+    &ec_gf8_mul_AC, &ec_gf8_mul_AD, &ec_gf8_mul_AE, &ec_gf8_mul_AF,
+    &ec_gf8_mul_B0, &ec_gf8_mul_B1, &ec_gf8_mul_B2, &ec_gf8_mul_B3,
+    &ec_gf8_mul_B4, &ec_gf8_mul_B5, &ec_gf8_mul_B6, &ec_gf8_mul_B7,
+    &ec_gf8_mul_B8, &ec_gf8_mul_B9, &ec_gf8_mul_BA, &ec_gf8_mul_BB,
+    &ec_gf8_mul_BC, &ec_gf8_mul_BD, &ec_gf8_mul_BE, &ec_gf8_mul_BF,
+    &ec_gf8_mul_C0, &ec_gf8_mul_C1, &ec_gf8_mul_C2, &ec_gf8_mul_C3,
+    &ec_gf8_mul_C4, &ec_gf8_mul_C5, &ec_gf8_mul_C6, &ec_gf8_mul_C7,
+    &ec_gf8_mul_C8, &ec_gf8_mul_C9, &ec_gf8_mul_CA, &ec_gf8_mul_CB,
+    &ec_gf8_mul_CC, &ec_gf8_mul_CD, &ec_gf8_mul_CE, &ec_gf8_mul_CF,
+    &ec_gf8_mul_D0, &ec_gf8_mul_D1, &ec_gf8_mul_D2, &ec_gf8_mul_D3,
+    &ec_gf8_mul_D4, &ec_gf8_mul_D5, &ec_gf8_mul_D6, &ec_gf8_mul_D7,
+    &ec_gf8_mul_D8, &ec_gf8_mul_D9, &ec_gf8_mul_DA, &ec_gf8_mul_DB,
+    &ec_gf8_mul_DC, &ec_gf8_mul_DD, &ec_gf8_mul_DE, &ec_gf8_mul_DF,
+    &ec_gf8_mul_E0, &ec_gf8_mul_E1, &ec_gf8_mul_E2, &ec_gf8_mul_E3,
+    &ec_gf8_mul_E4, &ec_gf8_mul_E5, &ec_gf8_mul_E6, &ec_gf8_mul_E7,
+    &ec_gf8_mul_E8, &ec_gf8_mul_E9, &ec_gf8_mul_EA, &ec_gf8_mul_EB,
+    &ec_gf8_mul_EC, &ec_gf8_mul_ED, &ec_gf8_mul_EE, &ec_gf8_mul_EF,
+    &ec_gf8_mul_F0, &ec_gf8_mul_F1, &ec_gf8_mul_F2, &ec_gf8_mul_F3,
+    &ec_gf8_mul_F4, &ec_gf8_mul_F5, &ec_gf8_mul_F6, &ec_gf8_mul_F7,
+    &ec_gf8_mul_F8, &ec_gf8_mul_F9, &ec_gf8_mul_FA, &ec_gf8_mul_FB,
+    &ec_gf8_mul_FC, &ec_gf8_mul_FD, &ec_gf8_mul_FE, &ec_gf8_mul_FF
+};
diff --git a/xlators/cluster/ec/src/ec-gf.h b/xlators/cluster/ec/src/ec-gf8.h
index 23bca91e3b5..4aca91127fc 100644
--- a/xlators/cluster/ec/src/ec-gf.h
+++ b/xlators/cluster/ec/src/ec-gf8.h
@@ -1,5 +1,5 @@
 /*
-  Copyright (c) 2012-2014 DataLab, s.l. <http://www.datalab.es>
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
   This file is part of GlusterFS.
 
   This file is licensed to you under your choice of the GNU Lesser
@@ -11,13 +11,8 @@
 #ifndef __EC_GF8_H__
 #define __EC_GF8_H__
 
-#define EC_GF_BITS 8
-#define EC_GF_MOD 0x11D
+#include "ec-galois.h"
 
-#define EC_GF_SIZE (1 << EC_GF_BITS)
-#define EC_GF_WORD_SIZE sizeof(uint64_t)
-
-extern void (* ec_gf_muladd[])(uint8_t * out, uint8_t * in,
-                               unsigned int width);
+extern ec_gf_mul_t *ec_gf8_mul[];
 
 #endif /* __EC_GF8_H__ */
diff --git a/xlators/cluster/ec/src/ec-heal.c b/xlators/cluster/ec/src/ec-heal.c
index 7fe1b2c4f8b..14255616830 100644
--- a/xlators/cluster/ec/src/ec-heal.c
+++ b/xlators/cluster/ec/src/ec-heal.c
@@ -11,21 +11,22 @@
 #include "xlator.h"
 #include "defaults.h"
 #include "compat-errno.h"
+#include "byte-order.h"
+#include "syncop.h"
+#include "syncop-utils.h"
+#include "cluster-syncop.h"
 
+#include "ec.h"
+#include "ec-mem-types.h"
+#include "ec-types.h"
+#include "ec-messages.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
 #include "ec-method.h"
 #include "ec-fops.h"
 
-#include "ec-mem-types.h"
-#include "ec-data.h"
-#include "byte-order.h"
-#include "ec-messages.h"
-#include "syncop.h"
-#include "syncop-utils.h"
-#include "cluster-syncop.h"
-
+#define alloca0(size) ({void *__ptr; __ptr = alloca(size); memset(__ptr, 0, size); __ptr; })
 #define EC_COUNT(array, max) ({int __i; int __res = 0; for (__i = 0; __i < max; __i++) if (array[__i]) __res++; __res; })
 #define EC_INTERSECT(dst, src1, src2, max) ({int __i; for (__i = 0; __i < max; __i++) dst[__i] = src1[__i] && src2[__i]; })
 #define EC_ADJUST_SOURCE(source, sources, max) ({int __i; if (sources[source] == 0) {source = -1; for (__i = 0; __i < max; __i++) if (sources[__i]) source = __i; } })
diff --git a/xlators/cluster/ec/src/ec-heald.h b/xlators/cluster/ec/src/ec-heald.h
index 0929044d545..4ae02e2df3c 100644
--- a/xlators/cluster/ec/src/ec-heald.h
+++ b/xlators/cluster/ec/src/ec-heald.h
@@ -13,32 +13,7 @@
 
 #include "xlator.h"
 
-struct _ec;
-typedef struct _ec ec_t;
-
-struct subvol_healer {
-        xlator_t        *this;
-        int              subvol;
-        gf_boolean_t     local;
-        gf_boolean_t     running;
-        gf_boolean_t     rerun;
-        pthread_mutex_t  mutex;
-        pthread_cond_t   cond;
-        pthread_t        thread;
-};
-
-struct _ec_self_heald;
-typedef struct _ec_self_heald ec_self_heald_t;
-
-struct _ec_self_heald {
-        gf_boolean_t            iamshd;
-        gf_boolean_t            enabled;
-        int                     timeout;
-        uint32_t                max_threads;
-        uint32_t                wait_qlength;
-        struct subvol_healer   *index_healers;
-        struct subvol_healer   *full_healers;
-};
+#include "ec-types.h"
 
 int
 ec_xl_op (xlator_t *this, dict_t *input, dict_t *output);
@@ -46,4 +21,5 @@ ec_xl_op (xlator_t *this, dict_t *input, dict_t *output);
 int
 ec_selfheal_daemon_init (xlator_t *this);
 void ec_selfheal_childup (ec_t *ec, int child);
+
 #endif /* __EC_HEALD_H__ */
diff --git a/xlators/cluster/ec/src/ec-helpers.c b/xlators/cluster/ec/src/ec-helpers.c
index 7cf8232353d..2391b2de3ae 100644
--- a/xlators/cluster/ec/src/ec-helpers.c
+++ b/xlators/cluster/ec/src/ec-helpers.c
@@ -12,10 +12,12 @@
 
 #include "byte-order.h"
 
+#include "ec.h"
 #include "ec-mem-types.h"
+#include "ec-messages.h"
 #include "ec-fops.h"
+#include "ec-method.h"
 #include "ec-helpers.h"
-#include "ec-messages.h"
 
 static const char * ec_fop_list[] =
 {
@@ -137,6 +139,53 @@ size_t ec_iov_copy_to(void * dst, struct iovec * vector, int32_t count,
     return total;
 }
 
+int32_t ec_buffer_alloc(xlator_t *xl, size_t size, struct iobref **piobref,
+                        void **ptr)
+{
+    struct iobref *iobref = NULL;
+    struct iobuf *iobuf = NULL;
+    int32_t ret = -ENOMEM;
+
+    iobuf = iobuf_get_page_aligned (xl->ctx->iobuf_pool, size,
+                                    EC_METHOD_WORD_SIZE);
+    if (iobuf == NULL) {
+        goto out;
+    }
+
+    iobref = *piobref;
+    if (iobref == NULL) {
+        iobref = iobref_new();
+        if (iobref == NULL) {
+            goto out;
+        }
+    }
+
+    ret = iobref_add(iobref, iobuf);
+    if (ret != 0) {
+        if (iobref != *piobref) {
+            iobref_unref(iobref);
+        }
+        iobref = NULL;
+
+        goto out;
+    }
+
+    GF_ASSERT(EC_ALIGN_CHECK(iobuf->ptr, EC_METHOD_WORD_SIZE));
+
+    *ptr = iobuf->ptr;
+
+out:
+    if (iobuf != NULL) {
+        iobuf_unref(iobuf);
+    }
+
+    if (iobref != NULL) {
+        *piobref = iobref;
+    }
+
+    return ret;
+}
+
 int32_t ec_dict_set_array(dict_t *dict, char *key, uint64_t value[],
                           int32_t size)
 {
diff --git a/xlators/cluster/ec/src/ec-helpers.h b/xlators/cluster/ec/src/ec-helpers.h
index 93d77726089..dfea6fef537 100644
--- a/xlators/cluster/ec/src/ec-helpers.h
+++ b/xlators/cluster/ec/src/ec-helpers.h
@@ -11,7 +11,10 @@
 #ifndef __EC_HELPERS_H__
 #define __EC_HELPERS_H__
 
-#include "ec-data.h"
+#include "ec-types.h"
+
+#define EC_ALIGN_CHECK(_ptr, _align) \
+    ((((uintptr_t)(_ptr)) & ((_align) - 1)) == 0)
 
 const char * ec_bin(char * str, size_t size, uint64_t value, int32_t digits);
 const char * ec_fop_name(int32_t id);
@@ -19,7 +22,8 @@ void ec_trace(const char * event, ec_fop_data_t * fop, const char * fmt, ...);
 int32_t ec_bits_consume(uint64_t * n);
 size_t ec_iov_copy_to(void * dst, struct iovec * vector, int32_t count,
                       off_t offset, size_t size);
-
+int32_t ec_buffer_alloc(xlator_t *xl, size_t size, struct iobref **piobref,
+                        void **ptr);
 int32_t ec_dict_set_array(dict_t *dict, char *key,
                           uint64_t *value, int32_t size);
 int32_t ec_dict_del_array(dict_t *dict, char *key,
diff --git a/xlators/cluster/ec/src/ec-inode-read.c b/xlators/cluster/ec/src/ec-inode-read.c
index c3d9c879eb7..6752b675273 100644
--- a/xlators/cluster/ec/src/ec-inode-read.c
+++ b/xlators/cluster/ec/src/ec-inode-read.c
@@ -11,12 +11,13 @@
 #include "xlator.h"
 #include "defaults.h"
 
+#include "ec.h"
+#include "ec-messages.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
 #include "ec-method.h"
 #include "ec-fops.h"
-#include "ec-messages.h"
 
 /* FOP: access */
 
@@ -1140,12 +1141,12 @@ out:
 
 int32_t ec_readv_rebuild(ec_t * ec, ec_fop_data_t * fop, ec_cbk_data_t * cbk)
 {
-    ec_cbk_data_t * ans = NULL;
-    struct iobref * iobref = NULL;
-    struct iobuf * iobuf = NULL;
-    uint8_t * buff = NULL, * ptr;
+    struct iovec vector[1];
+    ec_cbk_data_t *ans = NULL;
+    struct iobref *iobref = NULL;
+    void *ptr;
     size_t fsize = 0, size = 0, max = 0;
-    int32_t i = 0, err = -ENOMEM;
+    int32_t pos, err = -ENOMEM;
 
     if (cbk->op_ret < 0) {
         err = -cbk->op_errno;
@@ -1157,47 +1158,38 @@ int32_t ec_readv_rebuild(ec_t * ec, ec_fop_data_t * fop, ec_cbk_data_t * cbk)
     GF_ASSERT(ec_get_inode_size(fop, fop->fd->inode, &cbk->iatt[0].ia_size));
 
     if (cbk->op_ret > 0) {
-        struct iovec vector[1];
-        uint8_t * blocks[cbk->count];
+        void *blocks[cbk->count];
         uint32_t values[cbk->count];
 
         fsize = cbk->op_ret;
         size = fsize * ec->fragments;
-        buff = GF_MALLOC(size, gf_common_mt_char);
-        if (buff == NULL) {
-            goto out;
-        }
-        ptr = buff;
-        for (i = 0, ans = cbk; ans != NULL; i++, ans = ans->next) {
-            values[i] = ans->idx;
-            blocks[i] = ptr;
-            ptr += ec_iov_copy_to(ptr, ans->vector, ans->int32, 0, fsize);
+        for (ans = cbk; ans != NULL; ans = ans->next) {
+            pos = gf_bits_count(cbk->mask & ((1 << ans->idx) - 1));
+            values[pos] = ans->idx + 1;
+            blocks[pos] = ans->vector[0].iov_base;
+            if ((ans->int32 != 1) ||
+                !EC_ALIGN_CHECK(blocks[pos], EC_METHOD_WORD_SIZE)) {
+                if (iobref == NULL) {
+                    err = ec_buffer_alloc(ec->xl, size, &iobref, &ptr);
+                    if (err != 0) {
+                        goto out;
+                    }
+                }
+                ec_iov_copy_to(ptr, ans->vector, ans->int32, 0, fsize);
+                blocks[pos] = ptr;
+                ptr += fsize;
+            }
         }
 
-        iobref = iobref_new();
-        if (iobref == NULL) {
-            goto out;
-        }
-        iobuf = iobuf_get2(fop->xl->ctx->iobuf_pool, size);
-        if (iobuf == NULL) {
-            goto out;
-        }
-        err = iobref_add(iobref, iobuf);
+        err = ec_buffer_alloc(ec->xl, size, &iobref, &ptr);
         if (err != 0) {
             goto out;
         }
 
-        vector[0].iov_base = iobuf->ptr;
-        vector[0].iov_len = ec_method_decode(fsize, ec->fragments, values,
-                                             blocks, iobuf->ptr);
+        ec_method_decode(&ec->matrix, fsize, cbk->mask, values, blocks, ptr);
 
-        iobuf_unref(iobuf);
-
-        GF_FREE(buff);
-        buff = NULL;
-
-        vector[0].iov_base += fop->head;
-        vector[0].iov_len -= fop->head;
+        vector[0].iov_base = ptr + fop->head;
+        vector[0].iov_len = size - fop->head;
 
         max = fop->offset * ec->fragments + size;
         if (max > cbk->iatt[0].ia_size) {
@@ -1229,13 +1221,9 @@ int32_t ec_readv_rebuild(ec_t * ec, ec_fop_data_t * fop, ec_cbk_data_t * cbk)
     return 0;
 
 out:
-    if (iobuf != NULL) {
-        iobuf_unref(iobuf);
-    }
     if (iobref != NULL) {
         iobref_unref(iobref);
     }
-    GF_FREE(buff);
 
     return err;
 }
diff --git a/xlators/cluster/ec/src/ec-inode-write.c b/xlators/cluster/ec/src/ec-inode-write.c
index 6aeda5a2481..88145d98c83 100644
--- a/xlators/cluster/ec/src/ec-inode-write.c
+++ b/xlators/cluster/ec/src/ec-inode-write.c
@@ -11,12 +11,13 @@
 #include "xlator.h"
 #include "defaults.h"
 
+#include "ec.h"
+#include "ec-messages.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
 #include "ec-method.h"
 #include "ec-fops.h"
-#include "ec-messages.h"
 
 int
 ec_inode_write_cbk (call_frame_t *frame, xlator_t *this, void *cookie,
@@ -1285,27 +1286,78 @@ out:
     return -1;
 }
 
+static int32_t
+ec_writev_prepare_buffers(ec_t *ec, ec_fop_data_t *fop)
+{
+    struct iobref *iobref = NULL;
+    struct iovec *iov;
+    void *ptr;
+    int32_t err;
+
+    fop->user_size = iov_length(fop->vector, fop->int32);
+    fop->head = ec_adjust_offset(ec, &fop->offset, 0);
+    fop->size = ec_adjust_size(ec, fop->user_size + fop->head, 0);
+
+    if ((fop->int32 != 1) || (fop->head != 0) ||
+        (fop->size > fop->user_size) ||
+        !EC_ALIGN_CHECK(fop->vector[0].iov_base, EC_METHOD_WORD_SIZE)) {
+        err = ec_buffer_alloc(ec->xl, fop->size, &iobref, &ptr);
+        if (err != 0) {
+            goto out;
+        }
+
+        ec_iov_copy_to(ptr + fop->head, fop->vector, fop->int32, 0,
+                       fop->user_size);
+
+        fop->vector[0].iov_base = ptr;
+        fop->vector[0].iov_len = fop->size;
+
+        iobref_unref(fop->buffers);
+        fop->buffers = iobref;
+    }
+
+    if (fop->int32 != 2) {
+        iov = GF_MALLOC(VECTORSIZE(2), gf_common_mt_iovec);
+        if (iov == NULL) {
+            err = -ENOMEM;
+
+            goto out;
+        }
+        iov[0].iov_base = fop->vector[0].iov_base;
+        iov[0].iov_len = fop->vector[0].iov_len;
+
+        GF_FREE(fop->vector);
+        fop->vector = iov;
+    }
+
+    fop->vector[1].iov_len = fop->size / ec->fragments;
+    err = ec_buffer_alloc(ec->xl, fop->vector[1].iov_len * ec->nodes,
+                          &fop->buffers, &fop->vector[1].iov_base);
+    if (err != 0) {
+        goto out;
+    }
+
+    err = 0;
+
+out:
+    return err;
+}
+
 void ec_writev_start(ec_fop_data_t *fop)
 {
     ec_t *ec = fop->xl->private;
-    struct iobref *iobref = NULL;
-    struct iobuf *iobuf = NULL;
-    void *ptr = NULL;
     ec_fd_t *ctx;
     fd_t *fd;
-    size_t tail;
-    uint64_t current;
+    dict_t *xdata = NULL;
+    uint64_t tail, current;
     int32_t err = -ENOMEM;
-    dict_t      *xdata = NULL;
 
     /* This shouldn't fail because we have the inode locked. */
     GF_ASSERT(ec_get_inode_size(fop, fop->fd->inode, &current));
 
     fd = fd_anonymous(fop->fd->inode);
     if (fd == NULL) {
-        ec_fop_set_error(fop, ENOMEM);
-
-        return;
+        goto failed;
     }
 
     fop->frame->root->uid = 0;
@@ -1318,38 +1370,15 @@ void ec_writev_start(ec_fop_data_t *fop)
         }
     }
 
-    fop->user_size = iov_length(fop->vector, fop->int32);
-    fop->head = ec_adjust_offset(ec, &fop->offset, 0);
-    fop->size = ec_adjust_size(ec, fop->user_size + fop->head, 0);
-
-    iobref = iobref_new();
-    if (iobref == NULL) {
-        goto out;
-    }
-    iobuf = iobuf_get2(fop->xl->ctx->iobuf_pool, fop->size);
-    if (iobuf == NULL) {
-        goto out;
-    }
-    err = iobref_add(iobref, iobuf);
+    err = ec_writev_prepare_buffers(ec, fop);
     if (err != 0) {
-        goto out;
+        goto failed_fd;
     }
 
-    ptr = iobuf->ptr + fop->head;
-    ec_iov_copy_to(ptr, fop->vector, fop->int32, 0, fop->user_size);
-
-    fop->vector[0].iov_base = iobuf->ptr;
-    fop->vector[0].iov_len = fop->size;
-
-    iobuf_unref(iobuf);
-
-    iobref_unref(fop->buffers);
-    fop->buffers = iobref;
-
     if (fop->head > 0) {
         if (ec_make_internal_fop_xdata (&xdata)) {
                 err = -ENOMEM;
-                goto out;
+                goto failed_xdata;
         }
         ec_readv(fop->frame, fop->xl, -1, EC_MINIMUM_MIN, ec_writev_merge_head,
                  NULL, fd, ec->stripe_size, fop->offset, 0, xdata);
@@ -1359,7 +1388,7 @@ void ec_writev_start(ec_fop_data_t *fop)
         if (current > fop->offset + fop->head + fop->user_size) {
             if (ec_make_internal_fop_xdata (&xdata)) {
                     err = -ENOMEM;
-                    goto out;
+                    goto failed_xdata;
             }
             ec_readv(fop->frame, fop->xl, -1, EC_MINIMUM_MIN,
                      ec_writev_merge_tail, NULL, fd, ec->stripe_size,
@@ -1369,24 +1398,15 @@ void ec_writev_start(ec_fop_data_t *fop)
         }
     }
 
-    fd_unref(fd);
-    if (xdata)
-            dict_unref (xdata);
-
-    return;
+    err = 0;
 
-out:
-    if (iobuf != NULL) {
-        iobuf_unref(iobuf);
+failed_xdata:
+    if (xdata) {
+        dict_unref(xdata);
     }
-    if (iobref != NULL) {
-        iobref_unref(iobref);
-    }
-
+failed_fd:
     fd_unref(fd);
-    if (xdata)
-            dict_unref (xdata);
-
+failed:
     ec_fop_set_error(fop, -err);
 }
 
@@ -1411,55 +1431,32 @@ void ec_wind_writev(ec_t * ec, ec_fop_data_t * fop, int32_t idx)
     ec_trace("WIND", fop, "idx=%d", idx);
 
     struct iovec vector[1];
-    struct iobref * iobref = NULL;
-    struct iobuf * iobuf = NULL;
-    ssize_t size = 0, bufsize = 0;
-    int32_t err = -ENOMEM;
+    size_t size;
 
-    iobref = iobref_new();
-    if (iobref == NULL) {
-        goto out;
-    }
+    size = fop->vector[1].iov_len;
 
-    size = fop->vector[0].iov_len;
-    bufsize = size / ec->fragments;
-
-    iobuf = iobuf_get2(fop->xl->ctx->iobuf_pool, bufsize);
-    if (iobuf == NULL) {
-        goto out;
-    }
-    err = iobref_add(iobref, iobuf);
-    if (err != 0) {
-        goto out;
-    }
-
-    ec_method_encode(size, ec->fragments, idx, fop->vector[0].iov_base,
-                     iobuf->ptr);
-
-    vector[0].iov_base = iobuf->ptr;
-    vector[0].iov_len = bufsize;
-
-    iobuf_unref(iobuf);
+    vector[0].iov_base = fop->vector[1].iov_base + idx * size;
+    vector[0].iov_len = size;
 
     STACK_WIND_COOKIE(fop->frame, ec_writev_cbk, (void *)(uintptr_t)idx,
                       ec->xl_list[idx], ec->xl_list[idx]->fops->writev,
                       fop->fd, vector, 1, fop->offset / ec->fragments,
-                      fop->uint32, iobref, fop->xdata);
-
-    iobref_unref(iobref);
+                      fop->uint32, fop->buffers, fop->xdata);
+}
 
-    return;
+static void
+ec_writev_encode(ec_fop_data_t *fop)
+{
+    ec_t *ec = fop->xl->private;
+    void *blocks[ec->nodes];
+    uint32_t i;
 
-out:
-    if (iobuf != NULL) {
-        iobuf_unref(iobuf);
+    blocks[0] = fop->vector[1].iov_base;
+    for (i = 1; i < ec->nodes; i++) {
+        blocks[i] = blocks[i - 1] + fop->vector[1].iov_len;
     }
-    if (iobref != NULL) {
-        iobref_unref(iobref);
-    }
-
-    ec_writev_cbk(fop->frame, (void *)(uintptr_t)idx, fop->xl, -1, -err, NULL,
-                  NULL, NULL);
+    ec_method_encode(&ec->matrix, fop->vector[0].iov_len,
+                     fop->vector[0].iov_base, blocks);
 }
 
 int32_t ec_manager_writev(ec_fop_data_t *fop, int32_t state)
@@ -1488,6 +1485,8 @@ int32_t ec_manager_writev(ec_fop_data_t *fop, int32_t state)
             fop->frame->root->uid = fop->uid;
             fop->frame->root->gid = fop->gid;
 
+            ec_writev_encode(fop);
+
             ec_dispatch_all(fop);
 
             return EC_STATE_PREPARE_ANSWER;
diff --git a/xlators/cluster/ec/src/ec-mem-types.h b/xlators/cluster/ec/src/ec-mem-types.h
index df65a031590..9a4b6c58049 100644
--- a/xlators/cluster/ec/src/ec-mem-types.h
+++ b/xlators/cluster/ec/src/ec-mem-types.h
@@ -21,6 +21,10 @@ enum gf_ec_mem_types_
     ec_mt_ec_fd_t,
     ec_mt_ec_heal_t,
     ec_mt_subvol_healer_t,
+    ec_mt_ec_gf_t,
+    ec_mt_ec_code_t,
+    ec_mt_ec_code_builder_t,
+    ec_mt_ec_matrix_t,
     ec_mt_end
 };
 
diff --git a/xlators/cluster/ec/src/ec-messages.h b/xlators/cluster/ec/src/ec-messages.h
index 76678f8f836..dcdf50b9503 100644
--- a/xlators/cluster/ec/src/ec-messages.h
+++ b/xlators/cluster/ec/src/ec-messages.h
@@ -45,7 +45,7 @@
  */
 
 #define GLFS_EC_COMP_BASE       GLFS_MSGID_COMP_EC
-#define GLFS_NUM_MESSAGES       66
+#define GLFS_NUM_MESSAGES       73
 #define GLFS_MSGID_END          (GLFS_EC_COMP_BASE + GLFS_NUM_MESSAGES + 1)
 /* Messaged with message IDs */
 #define glfs_msg_start_x GLFS_EC_COMP_BASE, "Invalid: Start of messages"
@@ -520,6 +520,55 @@
  */
 #define EC_MSG_CONFIG_XATTR_INVALID         (GLFS_EC_COMP_BASE + 66)
 
+/*!
+ * @messageid
+ * @diagnosis
+ * @recommendedaction
+ */
+#define EC_MSG_EXTENSION                    (GLFS_EC_COMP_BASE + 67)
+
+/*!
+ * @messageid
+ * @diagnosis
+ * @recommendedaction
+ */
+#define EC_MSG_EXTENSION_NONE               (GLFS_EC_COMP_BASE + 68)
+
+/*!
+ * @messageid
+ * @diagnosis
+ * @recommendedaction
+ */
+#define EC_MSG_EXTENSION_UNKNOWN            (GLFS_EC_COMP_BASE + 69)
+
+/*!
+ * @messageid
+ * @diagnosis
+ * @recommendedaction
+ */
+#define EC_MSG_EXTENSION_UNSUPPORTED        (GLFS_EC_COMP_BASE + 70)
+
+/*!
+ * @messageid
+ * @diagnosis
+ * @recommendedaction
+ */
+#define EC_MSG_EXTENSION_FAILED             (GLFS_EC_COMP_BASE + 71)
+
+/*!
+ * @messageid
+ * @diagnosis
+ * @recommendedaction
+ */
+#define EC_MSG_NO_GF                        (GLFS_EC_COMP_BASE + 72)
+
+/*!
+ * @messageid
+ * @diagnosis
+ * @recommendedaction
+ */
+#define EC_MSG_MATRIX_FAILED                (GLFS_EC_COMP_BASE + 73)
+
 /*------------*/
 #define glfs_msg_end_x GLFS_MSGID_END, "Invalid: End of messages"
 
diff --git a/xlators/cluster/ec/src/ec-method.c b/xlators/cluster/ec/src/ec-method.c
index faab0115cdd..d1b122fb6a4 100644
--- a/xlators/cluster/ec/src/ec-method.c
+++ b/xlators/cluster/ec/src/ec-method.c
@@ -1,5 +1,5 @@
 /*
-  Copyright (c) 2012-2014 DataLab, s.l. <http://www.datalab.es>
+  Copyright (c) 2012-2015 DataLab, s.l. <http://www.datalab.es>
   This file is part of GlusterFS.
 
   This file is licensed to you under your choice of the GNU Lesser
@@ -11,149 +11,432 @@
 #include <string.h>
 #include <inttypes.h>
 
-#include "ec-gf.h"
+#include "ec-types.h"
+#include "ec-mem-types.h"
+#include "ec-galois.h"
+#include "ec-code.h"
 #include "ec-method.h"
 
-static uint32_t GfPow[EC_GF_SIZE << 1];
-static uint32_t GfLog[EC_GF_SIZE << 1];
+static void
+ec_method_matrix_normal(ec_gf_t *gf, uint32_t *matrix, uint32_t columns,
+                        uint32_t *values, uint32_t count)
+{
+    uint32_t i, j, v, tmp;
+
+    columns--;
+    for (i = 0; i < count; i++) {
+        v = *values++;
+        *matrix++ = tmp = ec_gf_exp(gf, v, columns);
+        for (j = 0; j < columns; j++) {
+            *matrix++ = tmp = ec_gf_div(gf, tmp, v);
+        }
+    }
+}
+
+static void
+ec_method_matrix_inverse(ec_gf_t *gf, uint32_t *matrix, uint32_t *values,
+                         uint32_t count)
+{
+    uint32_t a[count];
+    uint32_t i, j, p, last, tmp;
+
+    last = count - 1;
+    for (i = 0; i < last; i++) {
+        a[i] = 1;
+    }
+    a[i] = values[0];
+    for (i = last; i > 0; i--) {
+        for (j = i - 1; j < last; j++) {
+            a[j] = a[j + 1] ^ ec_gf_mul(gf, values[i], a[j]);
+        }
+        a[j] = ec_gf_mul(gf, values[i], a[j]);
+    }
+    for (i = 0; i < count; i++) {
+        p = a[0];
+        matrix += count;
+        *matrix = tmp = p ^ values[i];
+        for (j = 1; j < last; j++) {
+            matrix += count;
+            *matrix = tmp = a[j] ^ ec_gf_mul(gf, values[i], tmp);
+            p = tmp ^ ec_gf_mul(gf, values[i], p);
+        }
+        for (j = 0; j < last; j++) {
+            *matrix = ec_gf_div(gf, *matrix, p);
+            matrix -= count;
+        }
+        *matrix = ec_gf_div(gf, 1, p);
+        matrix++;
+    }
+}
 
-void ec_method_initialize(void)
+static gf_boolean_t
+ec_method_matrix_init(ec_matrix_list_t *list, ec_matrix_t *matrix,
+                      uintptr_t mask, uint32_t *rows, gf_boolean_t inverse)
 {
     uint32_t i;
 
-    GfPow[0] = 1;
-    GfLog[0] = EC_GF_SIZE;
-    for (i = 1; i < EC_GF_SIZE; i++)
-    {
-        GfPow[i] = GfPow[i - 1] << 1;
-        if (GfPow[i] >= EC_GF_SIZE)
-        {
-            GfPow[i] ^= EC_GF_MOD;
+    matrix->refs = 1;
+    matrix->mask = mask;
+    matrix->code = list->code;
+    matrix->columns = list->columns;
+    INIT_LIST_HEAD(&matrix->lru);
+
+    if (inverse) {
+        matrix->rows = list->columns;
+        ec_method_matrix_inverse(matrix->code->gf, matrix->values, rows,
+                                 matrix->rows);
+        for (i = 0; i < matrix->rows; i++) {
+            matrix->row_data[i].values = matrix->values + i * matrix->columns;
+            matrix->row_data[i].func.interleaved =
+                ec_code_build_interleaved(matrix->code,
+                                          EC_METHOD_WORD_SIZE,
+                                          matrix->row_data[i].values,
+                                          matrix->columns);
+            if (matrix->row_data[i].func.interleaved == NULL) {
+                return _gf_false;
+            }
+        }
+    } else {
+        matrix->rows = list->rows;
+        ec_method_matrix_normal(matrix->code->gf, matrix->values,
+                                matrix->columns, rows, matrix->rows);
+        for (i = 0; i < matrix->rows; i++) {
+            matrix->row_data[i].values = matrix->values + i * matrix->columns;
+            matrix->row_data[i].func.linear =
+                ec_code_build_linear(matrix->code, EC_METHOD_WORD_SIZE,
+                                     matrix->row_data[i].values,
+                                     matrix->columns);
+            if (matrix->row_data[i].func.linear == NULL) {
+                return _gf_false;
+            }
+        }
+    }
+
+    return _gf_true;
+}
+
+static void
+ec_method_matrix_release(ec_matrix_t *matrix)
+{
+    uint32_t i;
+
+    for (i = 0; i < matrix->rows; i++) {
+        if (matrix->row_data[i].func.linear != NULL) {
+            ec_code_release(matrix->code, &matrix->row_data[i].func);
+            matrix->row_data[i].func.linear = NULL;
         }
-        GfPow[i + EC_GF_SIZE - 1] = GfPow[i];
-        GfLog[GfPow[i] + EC_GF_SIZE - 1] = GfLog[GfPow[i]] = i;
     }
 }
 
-static uint32_t ec_method_mul(uint32_t a, uint32_t b)
+static void
+ec_method_matrix_destroy(ec_matrix_list_t *list, ec_matrix_t *matrix)
+{
+    list_del_init(&matrix->lru);
+
+    ec_method_matrix_release(matrix);
+
+    mem_put(matrix);
+
+    list->count--;
+}
+
+static void
+ec_method_matrix_unref(ec_matrix_list_t *list, ec_matrix_t *matrix)
 {
-    if (a && b)
-    {
-        return GfPow[GfLog[a] + GfLog[b]];
+    if (--matrix->refs == 0) {
+        list_add_tail(&matrix->lru, &list->lru);
+        if (list->count > list->max) {
+            matrix = list_first_entry(&list->lru, ec_matrix_t, lru);
+            ec_method_matrix_destroy(list, matrix);
+        }
     }
-    return 0;
 }
 
-static uint32_t ec_method_div(uint32_t a, uint32_t b)
+static ec_matrix_t *
+ec_method_matrix_lookup(ec_matrix_list_t *list, uintptr_t mask, uint32_t *pos)
 {
-    if (b)
-    {
-        if (a)
-        {
-            return GfPow[EC_GF_SIZE - 1 + GfLog[a] - GfLog[b]];
+    ec_matrix_t *matrix;
+    uint32_t i, j, k;
+
+    i = 0;
+    j = list->count;
+    while (i < j) {
+        k = (i + j) >> 1;
+        matrix = list->objects[k];
+        if (matrix->mask == mask) {
+            *pos = k;
+            return matrix;
+        }
+        if (matrix->mask < mask) {
+            i = k + 1;
+        } else {
+            j = k;
         }
-        return 0;
     }
-    return EC_GF_SIZE;
+    *pos = i;
+
+    return NULL;
 }
 
-size_t ec_method_encode(size_t size, uint32_t columns, uint32_t row,
-                        uint8_t * in, uint8_t * out)
+static void
+ec_method_matrix_remove(ec_matrix_list_t *list, uintptr_t mask)
 {
-    uint32_t i, j;
+    uint32_t pos;
 
-    size /= EC_METHOD_CHUNK_SIZE * columns;
-    row++;
-    for (j = 0; j < size; j++)
-    {
-        ec_gf_muladd[0](out, in, EC_METHOD_WIDTH);
-        in += EC_METHOD_CHUNK_SIZE;
-        for (i = 1; i < columns; i++)
-        {
-            ec_gf_muladd[row](out, in, EC_METHOD_WIDTH);
-            in += EC_METHOD_CHUNK_SIZE;
+    if (ec_method_matrix_lookup(list, mask, &pos) != NULL) {
+        list->count--;
+        if (pos < list->count) {
+            memmove(list->objects + pos, list->objects + pos + 1,
+                    sizeof(ec_matrix_t *) * (list->count - pos));
         }
-        out += EC_METHOD_CHUNK_SIZE;
     }
+}
+
+static void
+ec_method_matrix_insert(ec_matrix_list_t *list, ec_matrix_t *matrix)
+{
+    uint32_t pos;
+
+    GF_ASSERT(ec_method_matrix_lookup(list, matrix->mask, &pos) == NULL);
 
-    return size * EC_METHOD_CHUNK_SIZE;
+    if (pos < list->count) {
+        memmove(list->objects + pos + 1, list->objects + pos,
+                sizeof(ec_matrix_t *) * (list->count - pos));
+    }
+    list->objects[pos] = matrix;
+    list->count++;
 }
 
-size_t ec_method_decode(size_t size, uint32_t columns, uint32_t * rows,
-                        uint8_t ** in, uint8_t * out)
+static ec_matrix_t *
+ec_method_matrix_get(ec_matrix_list_t *list, uintptr_t mask, uint32_t *rows)
 {
-    uint32_t i, j, k, off, last, value;
-    uint32_t f;
-    uint8_t inv[EC_METHOD_MAX_FRAGMENTS][EC_METHOD_MAX_FRAGMENTS + 1];
-    uint8_t mtx[EC_METHOD_MAX_FRAGMENTS][EC_METHOD_MAX_FRAGMENTS];
-    uint8_t dummy[EC_METHOD_CHUNK_SIZE];
+    ec_matrix_t *matrix;
+    uint32_t pos;
+
+    LOCK(&list->lock);
 
-    size /= EC_METHOD_CHUNK_SIZE;
+    matrix = ec_method_matrix_lookup(list, mask, &pos);
+    if (matrix != NULL) {
+        list_del_init(&matrix->lru);
+        matrix->refs++;
 
-    memset(inv, 0, sizeof(inv));
-    memset(mtx, 0, sizeof(mtx));
-    memset(dummy, 0, sizeof(dummy));
-    for (i = 0; i < columns; i++)
-    {
-        inv[i][i] = 1;
-        inv[i][columns] = 1;
+        goto out;
     }
-    for (i = 0; i < columns; i++)
-    {
-        mtx[i][columns - 1] = 1;
-        for (j = columns - 1; j > 0; j--)
-        {
-            mtx[i][j - 1] = ec_method_mul(mtx[i][j], rows[i] + 1);
+
+    if ((list->count >= list->max) && !list_empty(&list->lru)) {
+        matrix = list_first_entry(&list->lru, ec_matrix_t, lru);
+        list_del_init(&matrix->lru);
+
+        ec_method_matrix_remove(list, matrix->mask);
+
+        ec_method_matrix_release(matrix);
+    } else {
+        matrix = mem_get0(list->pool);
+        if (matrix == NULL) {
+            goto out;
         }
+        matrix->values = (uint32_t *)((uintptr_t)matrix + sizeof(ec_matrix_t) +
+                                      sizeof(ec_matrix_row_t) * list->columns);
     }
 
-    for (i = 0; i < columns; i++)
-    {
-        f = mtx[i][i];
-        for (j = 0; j < columns; j++)
-        {
-            mtx[i][j] = ec_method_div(mtx[i][j], f);
-            inv[i][j] = ec_method_div(inv[i][j], f);
-        }
-        for (j = 0; j < columns; j++)
-        {
-            if (i != j)
-            {
-                f = mtx[j][i];
-                for (k = 0; k < columns; k++)
-                {
-                    mtx[j][k] ^= ec_method_mul(mtx[i][k], f);
-                    inv[j][k] ^= ec_method_mul(inv[i][k], f);
-                }
-            }
+    if (!ec_method_matrix_init(list, matrix, mask, rows, _gf_true)) {
+        ec_method_matrix_unref(list, matrix);
+
+        matrix = NULL;
+
+        goto out;
+    }
+
+    if (list->count < list->max) {
+        ec_method_matrix_insert(list, matrix);
+    } else {
+        matrix->mask = 0;
+    }
+
+out:
+    UNLOCK(&list->lock);
+
+    return matrix;
+}
+
+static void
+ec_method_matrix_put(ec_matrix_list_t *list, ec_matrix_t *matrix)
+{
+    LOCK(&list->lock);
+
+    ec_method_matrix_unref(list, matrix);
+
+    UNLOCK(&list->lock);
+}
+
+static gf_boolean_t
+ec_method_setup(xlator_t *xl, ec_matrix_list_t *list, const char *gen)
+{
+    ec_matrix_t *matrix;
+    uint32_t values[list->rows];
+    uint32_t i;
+
+    matrix = GF_MALLOC(sizeof(ec_matrix_t) +
+                       sizeof(ec_matrix_row_t) * list->rows +
+                       sizeof(uint32_t) * list->columns * list->rows,
+                       ec_mt_ec_matrix_t);
+    if (matrix == NULL) {
+        goto failed;
+    }
+    memset(matrix, 0, sizeof(ec_matrix_t));
+    matrix->values = (uint32_t *)((uintptr_t)matrix + sizeof(ec_matrix_t) +
+                                  sizeof(ec_matrix_row_t) * list->rows);
+
+    list->code = ec_code_create(list->gf, ec_code_detect(xl, gen));
+    if (list->code == NULL) {
+        goto failed_matrix;
+    }
+    list->width = list->code->width;
+
+    for (i = 0; i < list->rows; i++) {
+        values[i] = i + 1;
+    }
+    if (!ec_method_matrix_init(list, matrix, 0, values, _gf_false)) {
+        goto failed_code;
+    }
+
+    list->encode = matrix;
+
+    return _gf_true;
+
+failed_code:
+    ec_code_destroy(list->code);
+failed_matrix:
+    GF_FREE(matrix);
+failed:
+    return _gf_false;
+}
+
+gf_boolean_t
+ec_method_init(xlator_t *xl, ec_matrix_list_t *list, uint32_t columns,
+               uint32_t rows, uint32_t max, const char *gen)
+{
+    list->columns = columns;
+    list->rows = rows;
+    list->max = max;
+    list->stripe = EC_METHOD_CHUNK_SIZE * list->columns;
+    INIT_LIST_HEAD(&list->lru);
+
+    list->pool = mem_pool_new_fn(sizeof(ec_matrix_t) +
+                                 sizeof(ec_matrix_row_t) * columns +
+                                 sizeof(uint32_t) * columns * columns,
+                                 128, "ec_matrix_t");
+    if (list->pool == NULL) {
+        goto failed;
+    }
+
+    list->objects = GF_MALLOC(sizeof(ec_matrix_t *) * max, ec_mt_ec_matrix_t);
+    if (list->objects == NULL) {
+        goto failed_pool;
+    }
+
+    list->gf = ec_gf_prepare(EC_GF_BITS, EC_GF_MOD);
+    if (list->gf == NULL) {
+        goto failed_objects;
+    }
+
+    if (!ec_method_setup(xl, list, gen)) {
+        goto failed_gf;
+    }
+
+    LOCK_INIT(&list->lock);
+
+    return _gf_true;
+
+failed_gf:
+    ec_gf_destroy(list->gf);
+failed_objects:
+    GF_FREE(list->objects);
+failed_pool:
+    mem_pool_destroy(list->pool);
+failed:
+    list->pool = NULL;
+    list->objects = NULL;
+    list->gf = NULL;
+    return _gf_false;
+}
+
+void
+ec_method_fini(ec_matrix_list_t *list)
+{
+    ec_matrix_t *matrix;
+
+    if (list->encode == NULL) {
+        return;
+    }
+
+    while (!list_empty(&list->lru)) {
+        matrix = list_first_entry(&list->lru, ec_matrix_t, lru);
+        ec_method_matrix_destroy(list, matrix);
+    }
+
+    GF_ASSERT(list->count == 0);
+
+    if (list->pool)/*Init was successful*/
+            LOCK_DESTROY(&list->lock);
+
+    ec_method_matrix_release(list->encode);
+    GF_FREE(list->encode);
+
+    ec_code_destroy(list->code);
+    ec_gf_destroy(list->gf);
+    GF_FREE(list->objects);
+    mem_pool_destroy(list->pool);
+}
+
+gf_boolean_t
+ec_method_update(xlator_t *xl, ec_matrix_list_t *list, const char *gen)
+{
+    /* TODO: Allow changing code generator */
+
+    return _gf_true;
+}
+
+void
+ec_method_encode(ec_matrix_list_t *list, size_t size, void *in, void **out)
+{
+    ec_matrix_t *matrix;
+    size_t pos;
+    uint32_t i;
+
+    matrix = list->encode;
+    for (pos = 0; pos < size; pos += list->stripe) {
+        for (i = 0; i < matrix->rows; i++) {
+            matrix->row_data[i].func.linear(out[i], in, pos,
+                                            matrix->row_data[i].values,
+                                            list->columns);
+            out[i] += EC_METHOD_CHUNK_SIZE;
         }
     }
-    off = 0;
-    for (f = 0; f < size; f++)
-    {
-        for (i = 0; i < columns; i++)
-        {
-            last = 0;
-            j = 0;
-            do
-            {
-                while (inv[i][j] == 0)
-                {
-                    j++;
-                }
-                if (j < columns)
-                {
-                    value = ec_method_div(last, inv[i][j]);
-                    last = inv[i][j];
-                    ec_gf_muladd[value](out, in[j] + off, EC_METHOD_WIDTH);
-                    j++;
-                }
-            } while (j < columns);
-            ec_gf_muladd[last](out, dummy, EC_METHOD_WIDTH);
+}
+
+gf_boolean_t
+ec_method_decode(ec_matrix_list_t *list, size_t size, uintptr_t mask,
+                 uint32_t *rows, void **in, void *out)
+{
+    ec_matrix_t *matrix;
+    size_t pos;
+    uint32_t i;
+
+    matrix = ec_method_matrix_get(list, mask, rows);
+    if (matrix == NULL) {
+        return _gf_false;
+    }
+    for (pos = 0; pos < size; pos += EC_METHOD_CHUNK_SIZE) {
+        for (i = 0; i < matrix->rows; i++) {
+            matrix->row_data[i].func.interleaved(out, in, pos,
+                                                 matrix->row_data[i].values,
+                                                 list->columns);
             out += EC_METHOD_CHUNK_SIZE;
         }
-        off += EC_METHOD_CHUNK_SIZE;
     }
 
-    return size * EC_METHOD_CHUNK_SIZE * columns;
+    ec_method_matrix_put(list, matrix);
+
+    return _gf_true;
 }
diff --git a/xlators/cluster/ec/src/ec-method.h b/xlators/cluster/ec/src/ec-method.h
index 29b46e10443..818b54de872 100644
--- a/xlators/cluster/ec/src/ec-method.h
+++ b/xlators/cluster/ec/src/ec-method.h
@@ -1,5 +1,5 @@
 /*
-  Copyright (c) 2012-2014 DataLab, s.l. <http://www.datalab.es>
+  Copyright (c) 2012-2015 DataLab, s.l. <http://www.datalab.es>
   This file is part of GlusterFS.
 
   This file is licensed to you under your choice of the GNU Lesser
@@ -11,7 +11,15 @@
 #ifndef __EC_METHOD_H__
 #define __EC_METHOD_H__
 
-#include "ec-gf.h"
+#include "xlator.h"
+
+#include "ec-types.h"
+#include "ec-galois.h"
+
+#define EC_GF_BITS 8
+#define EC_GF_MOD 0x11D
+
+#define EC_GF_SIZE (1 << EC_GF_BITS)
 
 /* Determines the maximum size of the matrix used to encode/decode data */
 #define EC_METHOD_MAX_FRAGMENTS 16
@@ -21,12 +29,18 @@
 #define EC_METHOD_WORD_SIZE 64
 
 #define EC_METHOD_CHUNK_SIZE (EC_METHOD_WORD_SIZE * EC_GF_BITS)
-#define EC_METHOD_WIDTH (EC_METHOD_WORD_SIZE / EC_GF_WORD_SIZE)
 
-void ec_method_initialize(void);
-size_t ec_method_encode(size_t size, uint32_t columns, uint32_t row,
-                        uint8_t * in, uint8_t * out);
-size_t ec_method_decode(size_t size, uint32_t columns, uint32_t * rows,
-                        uint8_t ** in, uint8_t * out);
+gf_boolean_t ec_method_init(xlator_t *xl, ec_matrix_list_t *list,
+                            uint32_t columns, uint32_t rows, uint32_t max,
+                            const char *gen);
+void ec_method_fini(ec_matrix_list_t *list);
+gf_boolean_t ec_method_update(xlator_t *xl, ec_matrix_list_t *list,
+                              const char *gen);
+
+void ec_method_encode(ec_matrix_list_t *list, size_t size, void *in,
+                      void **out);
+gf_boolean_t ec_method_decode(ec_matrix_list_t *list, size_t size,
+                              uintptr_t mask, uint32_t *rows, void **in,
+                              void *out);
 
 #endif /* __EC_METHOD_H__ */
diff --git a/xlators/cluster/ec/src/ec-types.h b/xlators/cluster/ec/src/ec-types.h
new file mode 100644
index 00000000000..29f892f01be
--- /dev/null
+++ b/xlators/cluster/ec/src/ec-types.h
@@ -0,0 +1,580 @@
+/*
+  Copyright (c) 2015 DataLab, s.l. <http://www.datalab.es>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#ifndef __EC_TYPES_H__
+#define __EC_TYPES_H__
+
+#include "xlator.h"
+#include "timer.h"
+#include "libxlator.h"
+
+#define EC_GF_MAX_REGS 16
+
+#define EC_CODE_SIZE (1024 * 64)
+
+enum _ec_read_policy;
+typedef enum _ec_read_policy ec_read_policy_t;
+
+struct _ec_config;
+typedef struct _ec_config ec_config_t;
+
+struct _ec_fd;
+typedef struct _ec_fd ec_fd_t;
+
+struct _ec_inode;
+typedef struct _ec_inode ec_inode_t;
+
+union _ec_cbk;
+typedef union _ec_cbk ec_cbk_t;
+
+struct _ec_lock;
+typedef struct _ec_lock ec_lock_t;
+
+struct _ec_lock_link;
+typedef struct _ec_lock_link ec_lock_link_t;
+
+struct _ec_fop_data;
+typedef struct _ec_fop_data ec_fop_data_t;
+
+struct _ec_cbk_data;
+typedef struct _ec_cbk_data ec_cbk_data_t;
+
+enum _ec_gf_opcode;
+typedef enum _ec_gf_opcode ec_gf_opcode_t;
+
+struct _ec_gf_op;
+typedef struct _ec_gf_op ec_gf_op_t;
+
+struct _ec_gf_mul;
+typedef struct _ec_gf_mul ec_gf_mul_t;
+
+struct _ec_gf;
+typedef struct _ec_gf ec_gf_t;
+
+struct _ec_code_gen;
+typedef struct _ec_code_gen ec_code_gen_t;
+
+struct _ec_code;
+typedef struct _ec_code ec_code_t;
+
+struct _ec_code_arg;
+typedef struct _ec_code_arg ec_code_arg_t;
+
+struct _ec_code_op;
+typedef struct _ec_code_op ec_code_op_t;
+
+struct _ec_code_builder;
+typedef struct _ec_code_builder ec_code_builder_t;
+
+struct _ec_code_chunk;
+typedef struct _ec_code_chunk ec_code_chunk_t;
+
+struct _ec_code_space;
+typedef struct _ec_code_space ec_code_space_t;
+
+typedef void (*ec_code_func_linear_t)(void *dst, void *src, uint64_t offset,
+                                      uint32_t *values, uint32_t count);
+
+typedef void (*ec_code_func_interleaved_t)(void *dst, void **src,
+                                           uint64_t offset, uint32_t *values,
+                                           uint32_t count);
+
+union _ec_code_func;
+typedef union _ec_code_func ec_code_func_t;
+
+struct _ec_matrix_row;
+typedef struct _ec_matrix_row ec_matrix_row_t;
+
+struct _ec_matrix;
+typedef struct _ec_matrix ec_matrix_t;
+
+struct _ec_matrix_list;
+typedef struct _ec_matrix_list ec_matrix_list_t;
+
+struct _ec_heal;
+typedef struct _ec_heal ec_heal_t;
+
+struct _ec_self_heald;
+typedef struct _ec_self_heald ec_self_heald_t;
+
+struct _ec;
+typedef struct _ec ec_t;
+
+typedef void (*ec_wind_f)(ec_t *, ec_fop_data_t *, int32_t);
+typedef int32_t (*ec_handler_f)(ec_fop_data_t *, int32_t);
+typedef void (*ec_resume_f)(ec_fop_data_t *, int32_t);
+
+enum _ec_read_policy {
+        EC_ROUND_ROBIN,
+        EC_GFID_HASH,
+        EC_READ_POLICY_MAX
+};
+
+struct _ec_config {
+    uint32_t version;
+    uint8_t  algorithm;
+    uint8_t  gf_word_size;
+    uint8_t  bricks;
+    uint8_t  redundancy;
+    uint32_t chunk_size;
+};
+
+struct _ec_fd {
+    loc_t     loc;
+    uintptr_t open;
+    int32_t   flags;
+};
+
+struct _ec_inode {
+    ec_lock_t        *inode_lock;
+    gf_boolean_t      have_info;
+    gf_boolean_t      have_config;
+    gf_boolean_t      have_version;
+    gf_boolean_t      have_size;
+    ec_config_t       config;
+    uint64_t          pre_version[2];
+    uint64_t          post_version[2];
+    uint64_t          pre_size;
+    uint64_t          post_size;
+    uint64_t          dirty[2];
+    struct list_head  heal;
+};
+
+typedef int32_t (*fop_heal_cbk_t)(call_frame_t *, void *, xlator_t *, int32_t,
+                                  int32_t, uintptr_t, uintptr_t, uintptr_t,
+                                  dict_t *);
+typedef int32_t (*fop_fheal_cbk_t)(call_frame_t *, void *, xlator_t *, int32_t,
+                                   int32_t, uintptr_t, uintptr_t, uintptr_t,
+                                   dict_t *);
+
+union _ec_cbk {
+    fop_access_cbk_t       access;
+    fop_create_cbk_t       create;
+    fop_discard_cbk_t      discard;
+    fop_entrylk_cbk_t      entrylk;
+    fop_fentrylk_cbk_t     fentrylk;
+    fop_fallocate_cbk_t    fallocate;
+    fop_flush_cbk_t        flush;
+    fop_fsync_cbk_t        fsync;
+    fop_fsyncdir_cbk_t     fsyncdir;
+    fop_getxattr_cbk_t     getxattr;
+    fop_fgetxattr_cbk_t    fgetxattr;
+    fop_heal_cbk_t         heal;
+    fop_fheal_cbk_t        fheal;
+    fop_inodelk_cbk_t      inodelk;
+    fop_finodelk_cbk_t     finodelk;
+    fop_link_cbk_t         link;
+    fop_lk_cbk_t           lk;
+    fop_lookup_cbk_t       lookup;
+    fop_mkdir_cbk_t        mkdir;
+    fop_mknod_cbk_t        mknod;
+    fop_open_cbk_t         open;
+    fop_opendir_cbk_t      opendir;
+    fop_readdir_cbk_t      readdir;
+    fop_readdirp_cbk_t     readdirp;
+    fop_readlink_cbk_t     readlink;
+    fop_readv_cbk_t        readv;
+    fop_removexattr_cbk_t  removexattr;
+    fop_fremovexattr_cbk_t fremovexattr;
+    fop_rename_cbk_t       rename;
+    fop_rmdir_cbk_t        rmdir;
+    fop_setattr_cbk_t      setattr;
+    fop_fsetattr_cbk_t     fsetattr;
+    fop_setxattr_cbk_t     setxattr;
+    fop_fsetxattr_cbk_t    fsetxattr;
+    fop_stat_cbk_t         stat;
+    fop_fstat_cbk_t        fstat;
+    fop_statfs_cbk_t       statfs;
+    fop_symlink_cbk_t      symlink;
+    fop_truncate_cbk_t     truncate;
+    fop_ftruncate_cbk_t    ftruncate;
+    fop_unlink_cbk_t       unlink;
+    fop_writev_cbk_t       writev;
+    fop_xattrop_cbk_t      xattrop;
+    fop_fxattrop_cbk_t     fxattrop;
+    fop_zerofill_cbk_t     zerofill;
+    fop_seek_cbk_t         seek;
+};
+
+struct _ec_lock {
+    ec_inode_t        *ctx;
+    gf_timer_t        *timer;
+
+    /* List of owners of this lock. All fops added to this list are running
+     * concurrently. */
+    struct list_head   owners;
+
+    /* List of fops waiting to be an owner of the lock. Fops are added to this
+     * list when the current owner has an incompatible access (shared vs
+     * exclusive) or the lock is not acquired yet. */
+    struct list_head   waiting;
+
+    /* List of fops that will wait until the next unlock/lock cycle. This
+     * happens when the currently acquired lock is decided to be released as
+     * soon as possible. In this case, all frozen fops will be continued only
+     * after the lock is reacquired. */
+    struct list_head   frozen;
+
+    int32_t            exclusive;
+    uintptr_t          mask;
+    uintptr_t          good_mask;
+    uintptr_t          healing;
+    uint32_t           refs_owners;  /* Refs for fops owning the lock */
+    uint32_t           refs_pending; /* Refs assigned to fops being prepared */
+    gf_boolean_t       acquired;
+    gf_boolean_t       getting_size;
+    gf_boolean_t       release;
+    gf_boolean_t       query;
+    fd_t              *fd;
+    loc_t              loc;
+    union {
+        entrylk_type     type;
+        struct gf_flock  flock;
+    };
+};
+
+struct _ec_lock_link {
+    ec_lock_t        *lock;
+    ec_fop_data_t    *fop;
+    struct list_head  owner_list;
+    struct list_head  wait_list;
+    gf_boolean_t      update[2];
+    loc_t            *base;
+    uint64_t          size;
+};
+
+struct _ec_fop_data {
+    int32_t            id;
+    int32_t            refs;
+    int32_t            state;
+    int32_t            minimum;
+    int32_t            expected;
+    int32_t            winds;
+    int32_t            jobs;
+    int32_t            error;
+    ec_fop_data_t     *parent;
+    xlator_t          *xl;
+    call_frame_t      *req_frame;    /* frame of the calling xlator */
+    call_frame_t      *frame;        /* frame used by this fop */
+    struct list_head   cbk_list;     /* sorted list of groups of answers */
+    struct list_head   answer_list;  /* list of answers */
+    struct list_head   pending_list; /* member of ec_t.pending_fops */
+    ec_cbk_data_t     *answer;       /* accepted answer */
+    int32_t            lock_count;
+    int32_t            locked;
+    ec_lock_link_t     locks[2];
+    int32_t            first_lock;
+    gf_lock_t          lock;
+
+    uint32_t           flags;
+    uint32_t           first;
+    uintptr_t          mask;
+    uintptr_t          healing; /*Dispatch is done but call is successful only
+                                  if fop->minimum number of subvolumes succeed
+                                  which are not healing*/
+    uintptr_t          remaining;
+    uintptr_t          received; /* Mask of responses */
+    uintptr_t          good;
+
+    uid_t              uid;
+    gid_t              gid;
+
+    ec_wind_f          wind;
+    ec_handler_f       handler;
+    ec_resume_f        resume;
+    ec_cbk_t           cbks;
+    void              *data;
+    ec_heal_t         *heal;
+    struct list_head   healer;
+
+    uint64_t           user_size;
+    uint32_t           head;
+
+    int32_t            use_fd;
+
+    dict_t            *xdata;
+    dict_t            *dict;
+    int32_t            int32;
+    uint32_t           uint32;
+    uint64_t           size;
+    off_t              offset;
+    mode_t             mode[2];
+    entrylk_cmd        entrylk_cmd;
+    entrylk_type       entrylk_type;
+    gf_xattrop_flags_t xattrop_flags;
+    dev_t              dev;
+    inode_t           *inode;
+    fd_t              *fd;
+    struct iatt        iatt;
+    char              *str[2];
+    loc_t              loc[2];
+    struct gf_flock    flock;
+    struct iovec      *vector;
+    struct iobref     *buffers;
+    gf_seek_what_t     seek;
+};
+
+struct _ec_cbk_data {
+    struct list_head  list;        /* item in the sorted list of groups */
+    struct list_head  answer_list; /* item in the list of answers */
+    ec_fop_data_t    *fop;
+    ec_cbk_data_t    *next;        /* next answer in the same group */
+    int32_t           idx;
+    int32_t           op_ret;
+    int32_t           op_errno;
+    int32_t           count;
+    uintptr_t         mask;
+    uint64_t          dirty[2];
+
+    dict_t           *xdata;
+    dict_t           *dict;
+    int32_t           int32;
+    uintptr_t         uintptr[3];
+    uint64_t          size;
+    uint64_t          version[2];
+    inode_t          *inode;
+    fd_t             *fd;
+    struct statvfs    statvfs;
+    struct iatt       iatt[5];
+    struct gf_flock   flock;
+    struct iovec     *vector;
+    struct iobref    *buffers;
+    char             *str;
+    gf_dirent_t       entries;
+    off_t             offset;
+    gf_seek_what_t    what;
+};
+
+enum _ec_gf_opcode {
+    EC_GF_OP_LOAD,
+    EC_GF_OP_STORE,
+    EC_GF_OP_COPY,
+    EC_GF_OP_XOR2,
+    EC_GF_OP_XOR3,
+    EC_GF_OP_XORM,
+    EC_GF_OP_END
+};
+
+struct _ec_gf_op {
+    ec_gf_opcode_t op;
+    uint32_t       arg1;
+    uint32_t       arg2;
+    uint32_t       arg3;
+};
+
+struct _ec_gf_mul {
+    uint32_t    regs;
+    uint32_t    map[EC_GF_MAX_REGS];
+    ec_gf_op_t *ops;
+};
+
+struct _ec_gf {
+    uint32_t      bits;
+    uint32_t      size;
+    uint32_t      mod;
+    uint32_t      min_ops;
+    uint32_t      max_ops;
+    uint32_t      avg_ops;
+    uint32_t     *log;
+    uint32_t     *pow;
+    ec_gf_mul_t **table;
+};
+
+struct _ec_code_gen {
+    char *name;
+    char **flags;
+    uint32_t width;
+
+    void (*prolog)(ec_code_builder_t *builder);
+    void (*epilog)(ec_code_builder_t *builder);
+    void (*load)(ec_code_builder_t *builder, uint32_t reg, uint32_t offset,
+                 uint32_t bit);
+    void (*store)(ec_code_builder_t *builder, uint32_t reg, uint32_t bit);
+    void (*copy)(ec_code_builder_t *builder, uint32_t dst, uint32_t src);
+    void (*xor2)(ec_code_builder_t *builder, uint32_t dst, uint32_t src);
+    void (*xor3)(ec_code_builder_t *builder, uint32_t dst, uint32_t src1,
+                 uint32_t src2);
+    void (*xorm)(ec_code_builder_t *builder, uint32_t dst, uint32_t offset,
+                 uint32_t bit);
+};
+
+struct _ec_code {
+    gf_lock_t         lock;
+    struct list_head  spaces;
+    uint32_t          width;
+    ec_gf_t          *gf;
+    ec_code_gen_t    *gen;
+};
+
+struct _ec_code_arg {
+    uint32_t          value;
+};
+
+struct _ec_code_op {
+    ec_gf_opcode_t    op;
+    ec_code_arg_t     arg1;
+    ec_code_arg_t     arg2;
+    ec_code_arg_t     arg3;
+};
+
+struct _ec_code_builder {
+    ec_code_t        *code;
+    uint64_t          address;
+    uint8_t          *data;
+    uint32_t          size;
+    int32_t           error;
+    uint32_t          regs;
+    uint32_t          bits;
+    uint32_t          width;
+    uint32_t          count;
+    uint32_t          base;
+    uint32_t          map[EC_GF_MAX_REGS];
+    gf_boolean_t      linear;
+    uint64_t          loop;
+    ec_code_op_t      ops[0];
+};
+
+struct _ec_code_chunk {
+    struct list_head  list;
+    size_t            size;
+    ec_code_space_t  *space;
+};
+
+struct _ec_code_space {
+    struct list_head  list;
+    struct list_head  chunks;
+    ec_code_t        *code;
+    size_t            size;
+};
+
+
+union _ec_code_func {
+    ec_code_func_linear_t      linear;
+    ec_code_func_interleaved_t interleaved;
+};
+
+struct _ec_matrix_row {
+    ec_code_func_t  func;
+    uint32_t       *values;
+};
+
+struct _ec_matrix {
+    struct list_head lru;
+    uint32_t         refs;
+    uint32_t         columns;
+    uint32_t         rows;
+    uintptr_t        mask;
+    ec_code_t       *code;
+    uint32_t        *values;
+    ec_matrix_row_t  row_data[0];
+};
+
+struct _ec_matrix_list {
+    struct list_head   lru;
+    gf_lock_t          lock;
+    uint32_t           columns;
+    uint32_t           rows;
+    uint32_t           max;
+    uint32_t           count;
+    uint32_t           width;
+    uint32_t           stripe;
+    struct mem_pool   *pool;
+    ec_gf_t           *gf;
+    ec_code_t         *code;
+    ec_matrix_t       *encode;
+    ec_matrix_t      **objects;
+};
+
+struct _ec_heal {
+    struct list_head  list;
+    gf_lock_t         lock;
+    xlator_t         *xl;
+    ec_fop_data_t    *fop;
+    void             *data;
+    ec_fop_data_t    *lookup;
+    loc_t             loc;
+    struct iatt       iatt;
+    char             *symlink;
+    fd_t             *fd;
+    int32_t           partial;
+    int32_t           done;
+    int32_t           error;
+    gf_boolean_t      nameheal;
+    uintptr_t         available;
+    uintptr_t         good;
+    uintptr_t         bad;
+    uintptr_t         open;
+    uintptr_t         fixed;
+    uint64_t          offset;
+    uint64_t          size;
+    uint64_t          total_size;
+    uint64_t          version[2];
+    uint64_t          raw_size;
+};
+
+struct subvol_healer {
+        xlator_t        *this;
+        int              subvol;
+        gf_boolean_t     local;
+        gf_boolean_t     running;
+        gf_boolean_t     rerun;
+        pthread_mutex_t  mutex;
+        pthread_cond_t   cond;
+        pthread_t        thread;
+};
+
+struct _ec_self_heald {
+        gf_boolean_t            iamshd;
+        gf_boolean_t            enabled;
+        int                     timeout;
+        uint32_t                max_threads;
+        uint32_t                wait_qlength;
+        struct subvol_healer   *index_healers;
+        struct subvol_healer   *full_healers;
+};
+
+struct _ec {
+    xlator_t          *xl;
+    int32_t            healers;
+    int32_t            heal_waiters;
+    int32_t            nodes;
+    int32_t            bits_for_nodes;
+    int32_t            fragments;
+    int32_t            redundancy;
+    uint32_t           fragment_size;
+    uint32_t           stripe_size;
+    int32_t            up;
+    uint32_t           idx;
+    uint32_t           xl_up_count;
+    uintptr_t          xl_up;
+    uint32_t           xl_notify_count;
+    uintptr_t          xl_notify;
+    uintptr_t          node_mask;
+    xlator_t         **xl_list;
+    gf_lock_t          lock;
+    gf_timer_t        *timer;
+    gf_boolean_t       shutdown;
+    gf_boolean_t       eager_lock;
+    uint32_t           background_heals;
+    uint32_t           heal_wait_qlen;
+    struct list_head   pending_fops;
+    struct list_head   heal_waiting;
+    struct list_head   healing;
+    struct mem_pool   *fop_pool;
+    struct mem_pool   *cbk_pool;
+    struct mem_pool   *lock_pool;
+    ec_self_heald_t    shd;
+    char               vol_uuid[UUID_SIZE + 1];
+    dict_t            *leaf_to_subvolid;
+    ec_read_policy_t   read_policy;
+    ec_matrix_list_t   matrix;
+};
+
+#endif /* __EC_TYPES_H__ */
diff --git a/xlators/cluster/ec/src/ec.c b/xlators/cluster/ec/src/ec.c
index 659e3fd8108..2aff4374b82 100644
--- a/xlators/cluster/ec/src/ec.c
+++ b/xlators/cluster/ec/src/ec.c
@@ -1,5 +1,5 @@
 /*
-  Copyright (c) 2012-2014 DataLab, s.l. <http://www.datalab.es>
+  Copyright (c) 2012-2015 DataLab, s.l. <http://www.datalab.es>
   This file is part of GlusterFS.
 
   This file is licensed to you under your choice of the GNU Lesser
@@ -12,13 +12,15 @@
 #include "statedump.h"
 #include "compat-errno.h"
 
+#include "ec.h"
+#include "ec-messages.h"
 #include "ec-mem-types.h"
+#include "ec-types.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-fops.h"
 #include "ec-method.h"
-#include "ec.h"
-#include "ec-messages.h"
+#include "ec-code.h"
 #include "ec-heald.h"
 #include "events.h"
 
@@ -27,6 +29,7 @@ static char *ec_read_policies[EC_READ_POLICY_MAX + 1] = {
         [EC_GFID_HASH] = "gfid-hash",
         [EC_READ_POLICY_MAX] = NULL
 };
+
 #define EC_MAX_FRAGMENTS EC_METHOD_MAX_FRAGMENTS
 /* The maximum number of nodes is derived from the maximum allowed fragments
  * using the rule that redundancy cannot be equal or greater than the number
@@ -207,6 +210,9 @@ void __ec_destroy_private(xlator_t * this)
 
         if (ec->leaf_to_subvolid)
                 dict_unref (ec->leaf_to_subvolid);
+
+        ec_method_fini(&ec->matrix);
+
         GF_FREE(ec);
     }
 }
@@ -255,8 +261,12 @@ reconfigure (xlator_t *this, dict_t *options)
 {
         ec_t     *ec              = this->private;
         char     *read_policy     = NULL;
+        char     *extensions      = NULL;
         uint32_t heal_wait_qlen   = 0;
         uint32_t background_heals = 0;
+        int32_t  ret              = -1;
+
+        GF_OPTION_RECONF ("cpu-extensions", extensions, options, str, failed);
 
         GF_OPTION_RECONF ("self-heal-daemon", ec->shd.enabled, options, bool,
                           failed);
@@ -272,17 +282,24 @@ reconfigure (xlator_t *this, dict_t *options)
                           int32, failed);
         ec_configure_background_heal_opts (ec, background_heals,
                                            heal_wait_qlen);
-        GF_OPTION_RECONF ("read-policy", read_policy, options, str, failed);
-        if (ec_assign_read_policy (ec, read_policy))
-                goto failed;
         GF_OPTION_RECONF ("shd-max-threads", ec->shd.max_threads,
                           options, uint32, failed);
         GF_OPTION_RECONF ("shd-wait-qlength", ec->shd.wait_qlength,
                           options, uint32, failed);
 
-        return 0;
+        GF_OPTION_RECONF ("read-policy", read_policy, options, str, failed);
+
+        ret = 0;
+        if (ec_assign_read_policy (ec, read_policy)) {
+                ret = -1;
+        }
+
+        if (!ec_method_update(this, &ec->matrix, extensions)) {
+                ret = -1;
+        }
+
 failed:
-        return -1;
+        return ret;
 }
 
 glusterfs_event_t
@@ -554,6 +571,7 @@ init (xlator_t *this)
 {
     ec_t *ec          = NULL;
     char *read_policy = NULL;
+    char *extensions  = NULL;
 
     if (this->parents == NULL)
     {
@@ -608,7 +626,16 @@ init (xlator_t *this)
         goto failed;
     }
 
-    ec_method_initialize();
+    GF_OPTION_INIT("cpu-extensions", extensions, str, failed);
+
+    if (!ec_method_init(this, &ec->matrix, ec->fragments, ec->nodes,
+                        ec->nodes * 2, extensions)) {
+        gf_msg (this->name, GF_LOG_ERROR, 0, EC_MSG_MATRIX_FAILED,
+                "Failed to initialize matrix management");
+
+        goto failed;
+    }
+
     GF_OPTION_INIT ("self-heal-daemon", ec->shd.enabled, bool, failed);
     GF_OPTION_INIT ("iam-self-heal-daemon", ec->shd.iamshd, bool, failed);
     GF_OPTION_INIT ("eager-lock", ec->eager_lock, bool, failed);
@@ -1402,5 +1429,13 @@ struct volume_options options[] =
       .description = "This option can be used to control number of heals"
                      " that can wait in SHD per subvolume"
     },
+    {
+        .key = { "cpu-extensions" },
+        .type = GF_OPTION_TYPE_STR,
+        .value = { "none", "auto", "x64", "sse", "avx" },
+        .default_value = "auto",
+        .description = "force the cpu extensions to be used to accelerate the "
+                       "galois field computations."
+    },
     { }
 };
diff --git a/xlators/cluster/ec/src/ec.h b/xlators/cluster/ec/src/ec.h
index 49af5c2daf2..648d444f595 100644
--- a/xlators/cluster/ec/src/ec.h
+++ b/xlators/cluster/ec/src/ec.h
@@ -11,11 +11,6 @@
 #ifndef __EC_H__
 #define __EC_H__
 
-#include "xlator.h"
-#include "timer.h"
-#include "ec-heald.h"
-#include "libxlator.h"
-
 #define EC_XATTR_PREFIX  "trusted.ec."
 #define EC_XATTR_CONFIG  EC_XATTR_PREFIX"config"
 #define EC_XATTR_SIZE    EC_XATTR_PREFIX"size"
@@ -26,49 +21,4 @@
 #define EC_VERSION_SIZE 2
 #define EC_SHD_INODE_LRU_LIMIT          10
 
-typedef enum {
-        EC_ROUND_ROBIN,
-        EC_GFID_HASH,
-        EC_READ_POLICY_MAX
-} ec_read_policy_t;
-
-struct _ec
-{
-    xlator_t *        xl;
-    int32_t           healers;
-    int32_t           heal_waiters;
-    int32_t           nodes;
-    int32_t           bits_for_nodes;
-    int32_t           fragments;
-    int32_t           redundancy;
-    uint32_t          fragment_size;
-    uint32_t          stripe_size;
-    int32_t           up;
-    uint32_t          idx;
-    uint32_t          xl_up_count;
-    uintptr_t         xl_up;
-    uint32_t          xl_notify_count;
-    uintptr_t         xl_notify;
-    uintptr_t         node_mask;
-    xlator_t **       xl_list;
-    gf_lock_t         lock;
-    gf_timer_t *      timer;
-    gf_boolean_t      shutdown;
-    gf_boolean_t      eager_lock;
-    uint32_t          background_heals;
-    uint32_t          heal_wait_qlen;
-    struct list_head  pending_fops;
-    struct list_head  heal_waiting;
-    struct list_head  healing;
-    struct mem_pool * fop_pool;
-    struct mem_pool * cbk_pool;
-    struct mem_pool * lock_pool;
-    ec_self_heald_t   shd;
-    char              vol_uuid[UUID_SIZE + 1];
-    dict_t           *leaf_to_subvolid;
-    ec_read_policy_t  read_policy;
-};
-
-void ec_pending_fops_completed(ec_t *ec);
-
 #endif /* __EC_H__ */