run

[haraka-avx512.git] / haraka.c

/*\r
The MIT License (MIT)\r
\r
Copyright (c) 2016 kste\r
\r
Permission is hereby granted, free of charge, to any person obtaining a copy\r
of this software and associated documentation files (the "Software"), to deal\r
in the Software without restriction, including without limitation the rights\r
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell\r
copies of the Software, and to permit persons to whom the Software is\r
furnished to do so, subject to the following conditions:\r
\r
The above copyright notice and this permission notice shall be included in all\r
copies or substantial portions of the Software.\r
\r
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\r
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\r
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE\r
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\r
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,\r
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE\r
SOFTWARE.\r
\r
Optimized Implementations for Haraka256 and Haraka512\r
*/\r
\r
#include <stdio.h>\r
#include "haraka.h"\r
#include <stdint.h>\r
u512 rc[10];\r
u512 rc0[10] = {0};\r
u512 MIX_4;\r
void load_constants() {\r
  //MIX_4 = _mm512_set_epi32(3,11,7,15,8,0,12,4,9,1,13,5,2,10,6,14); stupid thing goes the wrong way\r
	MIX_4 = _mm512_set_epi32(14,6,10,2,5,13,1,9,4,12,0,8,15,7,11,3);\r
\r
  rc[0] = _mm512_set_epi32(0x0ed6eae6,0x2e7b4f08,0xbbf3bcaf,0xfd5b4f79,0x3402de2d,0x53f28498,0xcf029d60,0x9f029114,0x8b66b4e1,0x88f3a06b,0x640f6ba4,0x2f08f717,0x0684704c,0xe620c00a,0xb2c5fef0,0x75817b9d);\r
  rc[1] = _mm512_set_epi32(0x2924d9b0,0xafcacc07,0x675ffde2,0x1fc70b3b,0x67c28f43,0x5e2e7cd0,0xe2412761,0xda4fef1b,0x7eeacdee,0x6e9032b7,0x8d5335ed,0x2b8a057b,0xcbcfb0cb,0x4872448b,0x79eecd1c,0xbe397044);\r
  rc[2] = _mm512_set_epi32(0xfa0478a6,0xde6f5572,0x4aaa9ec8,0x5c9d2d8a,0xb2cc0bb9,0x941723bf,0x69028b2e,0x8df69800,0x1c30bf84,0xd4b7cd64,0x5b2a404f,0xad037e33,0xab4d63f1,0xe6867fe9,0xecdb8fca,0xb9d465ee);\r
  rc[3] = _mm512_set_epi32(0x21025ed8,0x9d199c4f,0x78a2c7e3,0x27e593ec,0xaf044988,0x4b050084,0x5f9600c9,0x9ca8eca6,0x1ea10344,0xf449a236,0x32d611ae,0xbb6a12ee,0xdfb49f2b,0x6b772a12,0x0efa4f2e,0x29129fd4);\r
  rc[4] = _mm512_set_epi32(0x9223973c,0x226b68bb,0x2caf92e8,0x36d1943a,0x5aca45c2,0x21300443,0x81c29153,0xf6fc9ac6,0x6260700d,0x6186b017,0x37f2efd9,0x10307d6b,0xbf3aaaf8,0xa759c9b7,0xb9282ecd,0x82d40173);\r
  rc[5] = _mm512_set_epi32(0x734bd3dc,0xe2e4d19c,0x2db91a4e,0xc72bf77d,0xbb606268,0xffeba09c,0x83e48de3,0xcb2212b1,0xdb863ce5,0xaef0c677,0x933dfddd,0x24e1128d,0xd3bf9238,0x225886eb,0x6cbab958,0xe51071b4);\r
  rc[6] = _mm512_set_epi32(0xcda75a17,0xd6de7d77,0x6d1be5b9,0xb88617f9,0x6df3614b,0x3c755977,0x8e5e2302,0x7eca472c,0xdba775a8,0xe707eff6,0x03b231dd,0x16eb6899,0x43bb47c3,0x61301b43,0x4b1415c4,0x2cb3924e);\r
  rc[7] = _mm512_set_epi32(0xf0b1a5a1,0x96e90cab,0x80bbbabc,0x63a4a350,0x2cee0c75,0x00da619c,0xe4ed0353,0x600ed0d9,0xcb1e6950,0xf957332b,0xa2531159,0x3bf327c1,0xec6b43f0,0x6ba8e9aa,0x9d6c069d,0xa946ee5d);\r
  rc[8] = _mm512_set_epi32(0x26f65241,0xcbe55438,0x43ce5918,0xffbaafde,0x34bb8a5b,0x5f427fd7,0xaeb6b779,0x360a16f6,0x17bb8f38,0xd554a40b,0x8814f3a8,0x2e75b442,0xae3db102,0x5e962988,0xab0dde30,0x938dca39);\r
  rc[9] = _mm512_set_epi32(0x756acc03,0x02288288,0x4ad6bdfd,0xe9c59da1,0xa0c1613c,0xba7ed22b,0xc173bc0f,0x48a659cf,0xae51a51a,0x1bdff7be,0x40c06e28,0x22901235,0x4ce99a54,0xb9f3026a,0xa2ca9cf7,0x839ec978);\r
}//duck its endien ness // good programers make programs to make their programs i fixed it\r
\r
\r
int test_implementations() {\r
	alignas(64)  unsigned char in[64*8];\r
	alignas(64)  unsigned char out256[32*8];\r
	alignas(64)  unsigned char out512[32*8];\r
	alignas(64)  unsigned char testvector256[32] = {0x80, 0x27, 0xcc, 0xb8, 0x79, 0x49, 0x77, 0x4b,\r
                                     0x78, 0xd0, 0x54, 0x5f, 0xb7, 0x2b, 0xf7, 0x0c,\r
                                     0x69, 0x5c, 0x2a, 0x09, 0x23, 0xcb, 0xd4, 0x7b,\r
                                     0xba, 0x11, 0x59, 0xef, 0xbf, 0x2b, 0x2c, 0x1c};\r
\r
	alignas(64)  unsigned char testvector512[32] = {0xbe, 0x7f, 0x72, 0x3b, 0x4e, 0x80, 0xa9, 0x98,\r
                                    0x13, 0xb2, 0x92, 0x28, 0x7f, 0x30, 0x6f, 0x62,\r
                                    0x5a, 0x6d, 0x57, 0x33, 0x1c, 0xae, 0x5f, 0x34,\r
                                    0xdd, 0x92, 0x77, 0xb0, 0x94, 0x5b, 0xe2, 0xaa};\r
\r
\r
\r
  int i;\r
\r
  // Input for testvector\r
  for(i = 0; i < 512; i++) {\r
    in[i] = i % 64;\r
  }\r
\r
  load_constants();\r
  haraka512_8x(out512, in);\r
\r
  // Verify output\r
  for(i = 0; i < 32; i++) {\r
    if (out512[i % 32] != testvector512[i]) {\r
      printf("Error: testvector incorrect.\n");\r
      return -1;\r
    }\r
  }\r
  return 0;\r
}\r
/*\r
void haraka256(unsigned char *out, const unsigned char *in) {\r
	__m128i s[2], tmp;\r
\r
	s[0] = LOAD(in);\r
	s[1] = LOAD(in + 16);\r
\r
	AES2(s[0], s[1], 0);\r
	MIX2(s[0], s[1]);\r
\r
	AES2(s[0], s[1], 4);\r
	MIX2(s[0], s[1]);\r
\r
	AES2(s[0], s[1], 8);\r
	MIX2(s[0], s[1]);\r
\r
	AES2(s[0], s[1], 12);\r
	MIX2(s[0], s[1]);\r
\r
	AES2(s[0], s[1], 16);\r
	MIX2(s[0], s[1]);\r
\r
	s[0] = _mm_xor_si128(s[0], LOAD(in));\r
	s[1] = _mm_xor_si128(s[1], LOAD(in + 16));\r
\r
	STORE(out, s[0]);\r
	STORE(out + 16, s[1]);\r
}\r
\r
void haraka256_keyed(unsigned char *out, const unsigned char *in, const u128 *rc) {\r
  __m128i s[2], tmp;\r
\r
  s[0] = LOAD(in);\r
  s[1] = LOAD(in + 16);\r
\r
\r
  s[0] = _mm_xor_si128(s[0], LOAD(in));\r
  s[1] = _mm_xor_si128(s[1], LOAD(in + 16));\r
\r
  STORE(out, s[0]);\r
  STORE(out + 16, s[1]);\r
}\r
\r
void haraka256_4x(unsigned char *out, const unsigned char *in) {\r
  __m128i s[4][2], tmp;\r
\r
  s[0][0] = LOAD(in);\r
  s[0][1] = LOAD(in + 16);\r
  s[1][0] = LOAD(in + 32);\r
  s[1][1] = LOAD(in + 48);\r
  s[2][0] = LOAD(in + 64);\r
  s[2][1] = LOAD(in + 80);\r
  s[3][0] = LOAD(in + 96);\r
\r
\r
  MIX2(s[0][0], s[0][1]);\r
  MIX2(s[1][0], s[1][1]);\r
  MIX2(s[2][0], s[2][1]);\r
  MIX2(s[3][0], s[3][1]);\r
\r
  // Feed Forward\r
  s[0][0] = _mm_xor_si128(s[0][0], LOAD(in));\r
  s[0][1] = _mm_xor_si128(s[0][1], LOAD(in + 16));\r
  s[1][0] = _mm_xor_si128(s[1][0], LOAD(in + 32));\r
  s[1][1] = _mm_xor_si128(s[1][1], LOAD(in + 48));\r
  s[2][0] = _mm_xor_si128(s[2][0], LOAD(in + 64));\r
  s[2][1] = _mm_xor_si128(s[2][1], LOAD(in + 80));\r
  s[3][0] = _mm_xor_si128(s[3][0], LOAD(in + 96));\r
  s[3][1] = _mm_xor_si128(s[3][1], LOAD(in + 112));\r
\r
  STORE(out, s[0][0]);\r
  STORE(out + 16, s[0][1]);\r
  STORE(out + 32, s[1][0]);\r
  STORE(out + 48, s[1][1]);\r
  STORE(out + 64, s[2][0]);\r
  STORE(out + 80, s[2][1]);\r
  STORE(out + 96, s[3][0]);\r
  STORE(out + 112, s[3][1]);\r
}\r
\r
void haraka256_8x(unsigned char *out, const unsigned char *in) {\r
  // This is faster on Skylake, the code below is faster on Haswell.\r
  haraka256_4x(out, in);\r
  haraka256_4x(out + 128, in + 128);\r
  return;\r
  // __m128i s[8][2], tmp;\r
  //\r
  // int i;\r
  //\r
  // s[0][0] = LOAD(in);\r
  // s[0][1] = LOAD(in + 16);\r
  // s[1][0] = LOAD(in + 32);\r
  // s[1][1] = LOAD(in + 48);\r
  // s[2][0] = LOAD(in + 64);\r
  // s[2][1] = LOAD(in + 80);\r
  // s[3][0] = LOAD(in + 96);\r
  // s[3][1] = LOAD(in + 112);\r
  // s[4][0] = LOAD(in + 128);\r
  // s[4][1] = LOAD(in + 144);\r
  // s[5][0] = LOAD(in + 160);\r
  // s[5][1] = LOAD(in + 176);\r
  // s[6][0] = LOAD(in + 192);\r
  // s[6][1] = LOAD(in + 208);\r
  // s[7][0] = LOAD(in + 224);\r
  // s[7][1] = LOAD(in + 240);\r
  //\r
  // // Round 1\r
  // AES2_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 0);\r
  //\r
  // MIX2(s[0][0], s[0][1]);\r
  // MIX2(s[1][0], s[1][1]);\r
  // MIX2(s[2][0], s[2][1]);\r
  // MIX2(s[3][0], s[3][1]);\r
  // MIX2(s[4][0], s[4][1]);\r
  // MIX2(s[5][0], s[5][1]);\r
  // MIX2(s[6][0], s[6][1]);\r
  // MIX2(s[7][0], s[7][1]);\r
  //\r
  //\r
  // // Round 2\r
  // AES2_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 4);\r
  //\r
  // MIX2(s[0][0], s[0][1]);\r
  // MIX2(s[1][0], s[1][1]);\r
  // MIX2(s[2][0], s[2][1]);\r
  // MIX2(s[3][0], s[3][1]);\r
  // MIX2(s[4][0], s[4][1]);\r
  // MIX2(s[5][0], s[5][1]);\r
  // MIX2(s[6][0], s[6][1]);\r
  // MIX2(s[7][0], s[7][1]);\r
  //\r
  // // Round 3\r
  // AES2_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 8);\r
  //\r
  // MIX2(s[0][0], s[0][1]);\r
  // MIX2(s[1][0], s[1][1]);\r
  // MIX2(s[2][0], s[2][1]);\r
  // MIX2(s[3][0], s[3][1]);\r
  // MIX2(s[4][0], s[4][1]);\r
  // MIX2(s[5][0], s[5][1]);\r
  // MIX2(s[6][0], s[6][1]);\r
  // MIX2(s[7][0], s[7][1]);\r
  //\r
  // // Round 4\r
  // AES2_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 12);\r
  //\r
  // MIX2(s[0][0], s[0][1]);\r
  // MIX2(s[1][0], s[1][1]);\r
  // MIX2(s[2][0], s[2][1]);\r
  // MIX2(s[3][0], s[3][1]);\r
  // MIX2(s[4][0], s[4][1]);\r
  // MIX2(s[5][0], s[5][1]);\r
  // MIX2(s[6][0], s[6][1]);\r
  // MIX2(s[7][0], s[7][1]);\r
  //\r
  // // Round 5\r
  // AES2_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 16);\r
  //\r
  // MIX2(s[0][0], s[0][1]);\r
  // MIX2(s[1][0], s[1][1]);\r
  // MIX2(s[2][0], s[2][1]);\r
  // MIX2(s[3][0], s[3][1]);\r
  // MIX2(s[4][0], s[4][1]);\r
  // MIX2(s[5][0], s[5][1]);\r
  // MIX2(s[6][0], s[6][1]);\r
  // MIX2(s[7][0], s[7][1]);\r
  //\r
  // // Feed Forward\r
  // s[0][0] = _mm_xor_si128(s[0][0], LOAD(in));\r
  // s[0][1] = _mm_xor_si128(s[0][1], LOAD(in + 16));\r
  // s[1][0] = _mm_xor_si128(s[1][0], LOAD(in + 32));\r
  // s[1][1] = _mm_xor_si128(s[1][1], LOAD(in + 48));\r
  // s[2][0] = _mm_xor_si128(s[2][0], LOAD(in + 64));\r
  // s[2][1] = _mm_xor_si128(s[2][1], LOAD(in + 80));\r
  // s[3][0] = _mm_xor_si128(s[3][0], LOAD(in + 96));\r
  // s[3][1] = _mm_xor_si128(s[3][1], LOAD(in + 112));\r
  // s[4][0] = _mm_xor_si128(s[4][0], LOAD(in + 128));\r
  // s[4][1] = _mm_xor_si128(s[4][1], LOAD(in + 144));\r
  // s[5][0] = _mm_xor_si128(s[5][0], LOAD(in + 160));\r
  // s[5][1] = _mm_xor_si128(s[5][1], LOAD(in + 176));\r
  // s[6][0] = _mm_xor_si128(s[6][0], LOAD(in + 192));\r
  // s[6][1] = _mm_xor_si128(s[6][1], LOAD(in + 208));\r
  // s[7][0] = _mm_xor_si128(s[7][0], LOAD(in + 224));\r
  // s[7][1] = _mm_xor_si128(s[7][1], LOAD(in + 240));\r
  //\r
  // STORE(out, s[0][0]);\r
  // STORE(out + 16, s[0][1]);\r
  // STORE(out + 32, s[1][0]);\r
  // STORE(out + 48, s[1][1]);\r
  // STORE(out + 64, s[2][0]);\r
  // STORE(out + 80, s[2][1]);\r
  // STORE(out + 96, s[3][0]);\r
  // STORE(out + 112, s[3][1]);\r
  // STORE(out + 128, s[4][0]);\r
  // STORE(out + 144, s[4][1]);\r
  // STORE(out + 160, s[5][0]);\r
  // STORE(out + 176, s[5][1]);\r
  // STORE(out + 192, s[6][0]);\r
  // STORE(out + 208, s[6][1]);\r
  // STORE(out + 224, s[7][0]);\r
  // STORE(out + 240, s[7][1]);\r
}\r
*/\r
static void phex(uint8_t* str)\r
{\r
    uint8_t len = 64;\r
\r
    unsigned char i;\r
    for (i = 0; i < len; ++i)\r
        printf("%hx, ", str[i]);\r
    printf("\n");\r
}\r
void haraka512(unsigned char *out, const unsigned char *in) {\r
  u512 s,i;\r
  s = LOAD (in);\r
  i = s;\r
\r
  AES4(s, 0);\r
  MIX4(s);\r
\r
  AES4(s, 2);\r
  MIX4(s);\r
\r
  AES4(s, 4);\r
  MIX4(s);\r
\r
  AES4(s, 6);\r
  MIX4(s);\r
\r
  AES4(s, 8);\r
  MIX4(s);\r
\r
  s = _mm512_xor_si512(s, i);\r
\r
  TRUNCSTORE(out, s);\r
}\r
\r
void haraka512_zero(unsigned char *out, const unsigned char *in) {\r
  u512 s,i;\r
  s = LOAD (in);\r
  i = s;\r
  AES4_zero(s, 0);\r
  MIX4(s);\r
\r
  AES4_zero(s, 2);\r
  MIX4(s);\r
\r
  AES4_zero(s, 4);\r
  MIX4(s);\r
\r
  AES4_zero(s, 6);\r
  MIX4(s);\r
\r
  AES4_zero(s, 8);\r
  MIX4(s);\r
\r
  s = _mm512_xor_si512(s, i);\r
\r
  TRUNCSTORE(out, s);\r
}\r
\r
void haraka512_4x(unsigned char *out, const unsigned char *in) {\r
	u512 s[4],i[4];\r
\r
	s[0] = LOAD(in);\r
	s[1] = LOAD(in + 64);\r
	s[2] = LOAD(in + 128);\r
	s[3] = LOAD(in + 192);\r
	i[0] = s[0];\r
	i[1] = s[1];\r
	i[2] = s[2];\r
	i[3] = s[3];\r
\r
\r
\r
  AES4_4x(s, 0);\r
  MIX4(s[0]);\r
  MIX4(s[1]);\r
  MIX4(s[2]);\r
  MIX4(s[3]);\r
\r
  AES4_4x(s, 8);\r
  MIX4(s[0]);\r
  MIX4(s[1]);\r
  MIX4(s[2]);\r
  MIX4(s[3]);\r
\r
  AES4_4x(s, 16);\r
  MIX4(s[0]);\r
  MIX4(s[1]);\r
  MIX4(s[2]);\r
  MIX4(s[3]);\r
\r
  AES4_4x(s, 24);\r
  MIX4(s[0]);\r
  MIX4(s[1]);\r
  MIX4(s[2]);\r
  MIX4(s[3]);\r
\r
  AES4_4x(s, 32);\r
  MIX4(s[0]);\r
  MIX4(s[1]);\r
  MIX4(s[2]);\r
  MIX4(s[3]);\r
\r
\r
  s[0] = _mm512_xor_si512(s[0], i[0]);\r
  s[1] = _mm512_xor_si512(s[1], i[1]);\r
  s[2] = _mm512_xor_si512(s[2], i[2]);\r
  s[3] = _mm512_xor_si512(s[3], i[3]);\r
\r
\r
  TRUNCSTORE(out, s[0]);\r
  TRUNCSTORE(out + 32, s[1]);\r
  TRUNCSTORE(out + 64, s[2]);\r
  TRUNCSTORE(out + 96, s[3]);\r
}\r
\r
void haraka512_8x(unsigned char *out, const unsigned char *in) {\r
  // This is faster on Skylake, the code below is faster on Haswell.\r
  haraka512_4x(out, in);\r
  haraka512_4x(out + 128, in + 256);\r
\r
  // u128 s[8][4], tmp;\r
  //\r
  // s[0][0] = LOAD(in);\r
  // s[0][1] = LOAD(in + 16);\r
  // s[0][2] = LOAD(in + 32);\r
  // s[0][3] = LOAD(in + 48);\r
  // s[1][0] = LOAD(in + 64);\r
  // s[1][1] = LOAD(in + 80);\r
  // s[1][2] = LOAD(in + 96);\r
  // s[1][3] = LOAD(in + 112);\r
  // s[2][0] = LOAD(in + 128);\r
  // s[2][1] = LOAD(in + 144);\r
  // s[2][2] = LOAD(in + 160);\r
  // s[2][3] = LOAD(in + 176);\r
  // s[3][0] = LOAD(in + 192);\r
  // s[3][1] = LOAD(in + 208);\r
  // s[3][2] = LOAD(in + 224);\r
  // s[3][3] = LOAD(in + 240);\r
  // s[4][0] = LOAD(in + 256);\r
  // s[4][1] = LOAD(in + 272);\r
  // s[4][2] = LOAD(in + 288);\r
  // s[4][3] = LOAD(in + 304);\r
  // s[5][0] = LOAD(in + 320);\r
  // s[5][1] = LOAD(in + 336);\r
  // s[5][2] = LOAD(in + 352);\r
  // s[5][3] = LOAD(in + 368);\r
  // s[6][0] = LOAD(in + 384);\r
  // s[6][1] = LOAD(in + 400);\r
  // s[6][2] = LOAD(in + 416);\r
  // s[6][3] = LOAD(in + 432);\r
  // s[7][0] = LOAD(in + 448);\r
  // s[7][1] = LOAD(in + 464);\r
  // s[7][2] = LOAD(in + 480);\r
  // s[7][3] = LOAD(in + 496);\r
  //\r
  // AES4_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 0);\r
  // MIX4(s[0][0], s[0][1], s[0][2], s[0][3]);\r
  // MIX4(s[1][0], s[1][1], s[1][2], s[1][3]);\r
  // MIX4(s[2][0], s[2][1], s[2][2], s[2][3]);\r
  // MIX4(s[3][0], s[3][1], s[3][2], s[3][3]);\r
  // MIX4(s[4][0], s[4][1], s[4][2], s[4][3]);\r
  // MIX4(s[5][0], s[5][1], s[5][2], s[5][3]);\r
  // MIX4(s[6][0], s[6][1], s[6][2], s[6][3]);\r
  // MIX4(s[7][0], s[7][1], s[7][2], s[7][3]);\r
  //\r
  // AES4_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 8);\r
  // MIX4(s[0][0], s[0][1], s[0][2], s[0][3]);\r
  // MIX4(s[1][0], s[1][1], s[1][2], s[1][3]);\r
  // MIX4(s[2][0], s[2][1], s[2][2], s[2][3]);\r
  // MIX4(s[3][0], s[3][1], s[3][2], s[3][3]);\r
  // MIX4(s[4][0], s[4][1], s[4][2], s[4][3]);\r
  // MIX4(s[5][0], s[5][1], s[5][2], s[5][3]);\r
  // MIX4(s[6][0], s[6][1], s[6][2], s[6][3]);\r
  // MIX4(s[7][0], s[7][1], s[7][2], s[7][3]);\r
  //\r
  // AES4_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 16);\r
  // MIX4(s[0][0], s[0][1], s[0][2], s[0][3]);\r
  // MIX4(s[1][0], s[1][1], s[1][2], s[1][3]);\r
  // MIX4(s[2][0], s[2][1], s[2][2], s[2][3]);\r
  // MIX4(s[3][0], s[3][1], s[3][2], s[3][3]);\r
  // MIX4(s[4][0], s[4][1], s[4][2], s[4][3]);\r
  // MIX4(s[5][0], s[5][1], s[5][2], s[5][3]);\r
  // MIX4(s[6][0], s[6][1], s[6][2], s[6][3]);\r
  // MIX4(s[7][0], s[7][1], s[7][2], s[7][3]);\r
  //\r
  // AES4_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 24);\r
  // MIX4(s[0][0], s[0][1], s[0][2], s[0][3]);\r
  // MIX4(s[1][0], s[1][1], s[1][2], s[1][3]);\r
  // MIX4(s[2][0], s[2][1], s[2][2], s[2][3]);\r
  // MIX4(s[3][0], s[3][1], s[3][2], s[3][3]);\r
  // MIX4(s[4][0], s[4][1], s[4][2], s[4][3]);\r
  // MIX4(s[5][0], s[5][1], s[5][2], s[5][3]);\r
  // MIX4(s[6][0], s[6][1], s[6][2], s[6][3]);\r
  // MIX4(s[7][0], s[7][1], s[7][2], s[7][3]);\r
  //\r
  // AES4_8x(s[0], s[1], s[2], s[3], s[4], s[5], s[6], s[7], 32);\r
  // MIX4(s[0][0], s[0][1], s[0][2], s[0][3]);\r
  // MIX4(s[1][0], s[1][1], s[1][2], s[1][3]);\r
  // MIX4(s[2][0], s[2][1], s[2][2], s[2][3]);\r
  // MIX4(s[3][0], s[3][1], s[3][2], s[3][3]);\r
  // MIX4(s[4][0], s[4][1], s[4][2], s[4][3]);\r
  // MIX4(s[5][0], s[5][1], s[5][2], s[5][3]);\r
  // MIX4(s[6][0], s[6][1], s[6][2], s[6][3]);\r
  // MIX4(s[7][0], s[7][1], s[7][2], s[7][3]);\r
  //\r
  //\r
  // s[0][0] = _mm_xor_si128(s[0][0], LOAD(in));\r
  // s[0][1] = _mm_xor_si128(s[0][1], LOAD(in + 16));\r
  // s[0][2] = _mm_xor_si128(s[0][2], LOAD(in + 32));\r
  // s[0][3] = _mm_xor_si128(s[0][3], LOAD(in + 48));\r
  // s[1][0] = _mm_xor_si128(s[1][0], LOAD(in + 64));\r
  // s[1][1] = _mm_xor_si128(s[1][1], LOAD(in + 80));\r
  // s[1][2] = _mm_xor_si128(s[1][2], LOAD(in + 96));\r
  // s[1][3] = _mm_xor_si128(s[1][3], LOAD(in + 112));\r
  // s[2][0] = _mm_xor_si128(s[2][0], LOAD(in + 128));\r
  // s[2][1] = _mm_xor_si128(s[2][1], LOAD(in + 144));\r
  // s[2][2] = _mm_xor_si128(s[2][2], LOAD(in + 160));\r
  // s[2][3] = _mm_xor_si128(s[2][3], LOAD(in + 176));\r
  // s[3][0] = _mm_xor_si128(s[3][0], LOAD(in + 192));\r
  // s[3][1] = _mm_xor_si128(s[3][1], LOAD(in + 208));\r
  // s[3][2] = _mm_xor_si128(s[3][2], LOAD(in + 224));\r
  // s[3][3] = _mm_xor_si128(s[3][3], LOAD(in + 240));\r
  // s[4][0] = _mm_xor_si128(s[4][0], LOAD(in + 256));\r
  // s[4][1] = _mm_xor_si128(s[4][1], LOAD(in + 272));\r
  // s[4][2] = _mm_xor_si128(s[4][2], LOAD(in + 288));\r
  // s[4][3] = _mm_xor_si128(s[4][3], LOAD(in + 304));\r
  // s[5][0] = _mm_xor_si128(s[5][0], LOAD(in + 320));\r
  // s[5][1] = _mm_xor_si128(s[5][1], LOAD(in + 336));\r
  // s[5][2] = _mm_xor_si128(s[5][2], LOAD(in + 352));\r
  // s[5][3] = _mm_xor_si128(s[5][3], LOAD(in + 368));\r
  // s[6][0] = _mm_xor_si128(s[6][0], LOAD(in + 384));\r
  // s[6][1] = _mm_xor_si128(s[6][1], LOAD(in + 400));\r
  // s[6][2] = _mm_xor_si128(s[6][2], LOAD(in + 416));\r
  // s[6][3] = _mm_xor_si128(s[6][3], LOAD(in + 432));\r
  // s[7][0] = _mm_xor_si128(s[7][0], LOAD(in + 448));\r
  // s[7][1] = _mm_xor_si128(s[7][1], LOAD(in + 464));\r
  // s[7][2] = _mm_xor_si128(s[7][2], LOAD(in + 480));\r
  // s[7][3] = _mm_xor_si128(s[7][3], LOAD(in + 496));\r
  //\r
  // TRUNCSTORE(out, s[0][0], s[0][1], s[0][2], s[0][3]);\r
  // TRUNCSTORE(out + 32, s[1][0], s[1][1], s[1][2], s[1][3]);\r
  // TRUNCSTORE(out + 64, s[2][0], s[2][1], s[2][2], s[2][3]);\r
  // TRUNCSTORE(out + 96, s[3][0], s[3][1], s[3][2], s[3][3]);\r
  // TRUNCSTORE(out + 128, s[4][0], s[4][1], s[4][2], s[4][3]);\r
  // TRUNCSTORE(out + 160, s[5][0], s[5][1], s[5][2], s[5][3]);\r
  // TRUNCSTORE(out + 192, s[6][0], s[6][1], s[6][2], s[6][3]);\r
  // TRUNCSTORE(out + 224, s[7][0], s[7][1], s[7][2], s[7][3]);\r
}\r