/**
* Initializes the Sponge State. The first 512 bits are set to zeros and the remainder
* receive Blake2b's IV as per Blake2b's specification. <b>Note:</b> Even though sponges
* typically have their internal state initialized with zeros, Blake2b's G function
* has a fixed point: if the internal state and message are both filled with zeros. the
* resulting permutation will always be a block filled with zeros; this happens because
* Blake2b does not use the constants originally employed in Blake2 inside its G function,
* relying on the IV for avoiding possible fixed points.
*
* @param state The 1024-bit array to be initialized
*/
inline void initState(uint64_t state[/*16*/]) {
#ifdef __AVX2__
(*(__m256i*)(&state[0])) = _mm256_setzero_si256();
(*(__m256i*)(&state[4])) = _mm256_setzero_si256();
(*(__m256i*)(&state[8])) = _mm256_set_epi64x( blake2b_IV[3],
blake2b_IV[2],
blake2b_IV[1],
blake2b_IV[0] );
(*(__m256i*)(&state[12])) = _mm256_set_epi64x(blake2b_IV[7],
blake2b_IV[6],
blake2b_IV[5],
blake2b_IV[4] );
//AVX is around the same number of instructions as unnoptimized
//#elif defined __AVX__
#else
//First 512 bis are zeros
memset(state, 0, 64);
//Remainder BLOCK_LEN_BLAKE2_SAFE_BYTES are reserved to the IV
state[8] = blake2b_IV[0];
state[9] = blake2b_IV[1];
state[10] = blake2b_IV[2];
state[11] = blake2b_IV[3];
state[12] = blake2b_IV[4];
state[13] = blake2b_IV[5];
state[14] = blake2b_IV[6];
state[15] = blake2b_IV[7];
#endif
}
//int64
vec3l& vec3l::Sub(int64 v) {
__m256i vxmm = _mm256_set_epi64x(0, v, v, v);
__m256i xmm = _mm256_set_epi64x(0, z, y, x);
xmm = _mm256_sub_epi64(xmm, vxmm);
x = M256L(xmm, 0);
y = M256L(xmm, 1);
z = M256L(xmm, 2);
return *this;
}
vec3l& vec3l::Sub(const vec3l& v) {
__m256i vxmm = _mm256_set_epi64x(0, v.z, v.y, v.x);
__m256i xmm = _mm256_set_epi64x(0, z, y, x);
xmm = _mm256_sub_epi64(xmm, vxmm);
x = M256L(xmm, 0);
y = M256L(xmm, 1);
z = M256L(xmm, 2);
return *this;
}
static void
avx2_test (void)
{
union256i_q u, s1, s2;
long long int e[4];
int i;
s1.x = _mm256_set_epi64x (1, 2, 3, 4);
s2.x = _mm256_set_epi64x (88, 44, 3, 220000);
u.x = _mm256_cmpgt_epi64 (s1.x, s2.x);
for (i = 0; i < 4; i++)
e[i] = (s1.a[i] > s2.a[i]) ? -1 : 0;
if (check_union256i_q (u, e))
abort ();
}
void key_schedule(const unsigned char *k, u256 rk[40][16]) {
int i, j;
u256 tk1[32], tmp[32];
unsigned char *tmp_key = malloc(32);
for(i = 0; i < 2; i++)
memcpy(tmp_key + 16*i, k, 16);
pack_key(tk1, tmp_key);
for(j = 0; j < 40; j++) {
//Extract round key
for(i = 0; i < 16; i++){
rk[j][i] = tk1[i];
}
//Add constant into key
u256 rc = _mm256_set_epi64x(0x000000FF000000FFull,
0x000000FF000000FFull,
0x000000FF000000FFull,
0x000000FF000000FFull);
if(RC[j]>>5 & 1)
rk[j][14] = XOR(rk[j][14], rc);
if(RC[j]>>4 & 1)
rk[j][15] = XOR(rk[j][15], rc);
if(RC[j]>>3 & 1)
rk[j][4] = XOR(rk[j][4], rc);
if(RC[j]>>2 & 1)
rk[j][5] = XOR(rk[j][5], rc);
if(RC[j]>>1 & 1)
rk[j][6] = XOR(rk[j][6], rc);
if(RC[j]>>0 & 1)
rk[j][7] = XOR(rk[j][7], rc);
//Update TK1
for(i = 0; i < 16; i++){
tmp[16 + i] = tk1[0 + i];
}
//Apply bit permutation
for(i = 0; i < 8; i++){
tmp[0 + i] = XOR(_mm256_shuffle_epi8(tk1[16 + i], _mm256_set_epi8(0xff,28,0xff,29,0xff,24,0xff,25,0xff,20,0xff,21,0xff,16,0xff,17,0xff,12,0xff,13,0xff,8,0xff,9,0xff,4,0xff,5,0xff,0,0xff,1)),
_mm256_shuffle_epi8(tk1[24 + i], _mm256_set_epi8(29,0xff,31,0xff,25,0xff,27,0xff,21,0xff,23,0xff,17,0xff,19,0xff,13,0xff,15,0xff,9,0xff,11,0xff,5,0xff,7,0xff,1,0xff,3,0xff)));
tmp[8 + i] = XOR(_mm256_shuffle_epi8(tk1[16 + i], _mm256_set_epi8(31,0xff,0xff,30,27,0xff,0xff,26,23,0xff,0xff,22,19,0xff,0xff,18,15,0xff,0xff,14,11,0xff,0xff,10,7,0xff,0xff,6,3,0xff,0xff,2)),
_mm256_shuffle_epi8(tk1[24 + i], _mm256_set_epi8(0xff,28,30,0xff,0xff,24,26,0xff,0xff,20,22,0xff,0xff,16,18,0xff,0xff,12,14,0xff,0xff,8,10,0xff,0xff,4,6,0xff,0xff,0,2,0xff)));
}
for(i = 0; i < 32; i++){
tk1[i] = tmp[i];
}
}
free(tmp_key);
}
foo (long long x)
{
return _mm256_set_epi64x (x, x, x, x);
}
void Threefish_512_AVX2::encrypt_n(const byte in[], byte out[], size_t blocks) const
{
const u64bit* K = &get_K()[0];
const u64bit* T_64 = &get_T()[0];
const __m256i ROTATE_1 = _mm256_set_epi64x(37,19,36,46);
const __m256i ROTATE_2 = _mm256_set_epi64x(42,14,27,33);
const __m256i ROTATE_3 = _mm256_set_epi64x(39,36,49,17);
const __m256i ROTATE_4 = _mm256_set_epi64x(56,54, 9,44);
const __m256i ROTATE_5 = _mm256_set_epi64x(24,34,30,39);
const __m256i ROTATE_6 = _mm256_set_epi64x(17,10,50,13);
const __m256i ROTATE_7 = _mm256_set_epi64x(43,39,29,25);
const __m256i ROTATE_8 = _mm256_set_epi64x(22,56,35, 8);
#define THREEFISH_ROUND(X0, X1, SHL) \
do { \
const __m256i SHR = _mm256_sub_epi64(_mm256_set1_epi64x(64), SHL); \
X0 = _mm256_add_epi64(X0, X1); \
X1 = _mm256_or_si256(_mm256_sllv_epi64(X1, SHL), _mm256_srlv_epi64(X1, SHR)); \
X1 = _mm256_xor_si256(X1, X0); \
X0 = _mm256_permute4x64_epi64(X0, _MM_SHUFFLE(0, 3, 2, 1)); \
X1 = _mm256_permute4x64_epi64(X1, _MM_SHUFFLE(1, 2, 3, 0)); \
} while(0)
#define THREEFISH_ROUND_2(X0, X1, X2, X3, SHL) \
do { \
const __m256i SHR = _mm256_sub_epi64(_mm256_set1_epi64x(64), SHL); \
X0 = _mm256_add_epi64(X0, X1); \
X2 = _mm256_add_epi64(X2, X3); \
X1 = _mm256_or_si256(_mm256_sllv_epi64(X1, SHL), _mm256_srlv_epi64(X1, SHR)); \
X3 = _mm256_or_si256(_mm256_sllv_epi64(X3, SHL), _mm256_srlv_epi64(X3, SHR)); \
X1 = _mm256_xor_si256(X1, X0); \
X3 = _mm256_xor_si256(X3, X2); \
X0 = _mm256_permute4x64_epi64(X0, _MM_SHUFFLE(0, 3, 2, 1)); \
X2 = _mm256_permute4x64_epi64(X2, _MM_SHUFFLE(0, 3, 2, 1)); \
X1 = _mm256_permute4x64_epi64(X1, _MM_SHUFFLE(1, 2, 3, 0)); \
X3 = _mm256_permute4x64_epi64(X3, _MM_SHUFFLE(1, 2, 3, 0)); \
} while(0)
#define THREEFISH_INJECT_KEY(X0, X1, R, K0, K1, T0I, T1I) \
do { \
const __m256i T0 = _mm256_permute4x64_epi64(T, _MM_SHUFFLE(T0I, 0, 0, 0)); \
const __m256i T1 = _mm256_permute4x64_epi64(T, _MM_SHUFFLE(0, T1I, 0, 0)); \
X0 = _mm256_add_epi64(X0, K0); \
X1 = _mm256_add_epi64(X1, K1); \
X1 = _mm256_add_epi64(X1, R); \
X0 = _mm256_add_epi64(X0, T0); \
X1 = _mm256_add_epi64(X1, T1); \
R = _mm256_add_epi64(R, ONE); \
} while(0)
#define THREEFISH_INJECT_KEY_2(X0, X1, X2, X3, R, K0, K1, T0I, T1I) \
do { \
const __m256i T0 = _mm256_permute4x64_epi64(T, _MM_SHUFFLE(T0I, 0, 0, 0)); \
__m256i T1 = _mm256_permute4x64_epi64(T, _MM_SHUFFLE(0, T1I, 0, 0)); \
X0 = _mm256_add_epi64(X0, K0); \
X2 = _mm256_add_epi64(X2, K0); \
X1 = _mm256_add_epi64(X1, K1); \
X3 = _mm256_add_epi64(X3, K1); \
T1 = _mm256_add_epi64(T1, R); \
X0 = _mm256_add_epi64(X0, T0); \
X2 = _mm256_add_epi64(X2, T0); \
X1 = _mm256_add_epi64(X1, T1); \
X3 = _mm256_add_epi64(X3, T1); \
R = _mm256_add_epi64(R, ONE); \
} while(0)
#define THREEFISH_ENC_8_ROUNDS(X0, X1, R, K1, K2, K3, T0, T1, T2) \
do { \
THREEFISH_ROUND(X0, X1, ROTATE_1); \
THREEFISH_ROUND(X0, X1, ROTATE_2); \
THREEFISH_ROUND(X0, X1, ROTATE_3); \
THREEFISH_ROUND(X0, X1, ROTATE_4); \
THREEFISH_INJECT_KEY(X0, X1, R, K1, K2, T0, T1); \
\
THREEFISH_ROUND(X0, X1, ROTATE_5); \
THREEFISH_ROUND(X0, X1, ROTATE_6); \
THREEFISH_ROUND(X0, X1, ROTATE_7); \
THREEFISH_ROUND(X0, X1, ROTATE_8); \
THREEFISH_INJECT_KEY(X0, X1, R, K2, K3, T2, T0); \
} while(0)
#define THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K1, K2, K3, T0, T1, T2) \
do { \
THREEFISH_ROUND_2(X0, X1, X2, X3, ROTATE_1); \
THREEFISH_ROUND_2(X0, X1, X2, X3, ROTATE_2); \
THREEFISH_ROUND_2(X0, X1, X2, X3, ROTATE_3); \
THREEFISH_ROUND_2(X0, X1, X2, X3, ROTATE_4); \
THREEFISH_INJECT_KEY_2(X0, X1, X2, X3, R, K1, K2, T0, T1); \
\
THREEFISH_ROUND_2(X0, X1, X2, X3, ROTATE_5); \
THREEFISH_ROUND_2(X0, X1, X2, X3, ROTATE_6); \
THREEFISH_ROUND_2(X0, X1, X2, X3, ROTATE_7); \
THREEFISH_ROUND_2(X0, X1, X2, X3, ROTATE_8); \
THREEFISH_INJECT_KEY_2(X0, X1, X2, X3, R, K2, K3, T2, T0); \
} while(0)
/*
v1.0 key schedule: 9 ymm registers (only need 2 or 3)
(0,1,2,3),(4,5,6,7) [8]
then mutating with vpermq
*/
const __m256i K0 = _mm256_set_epi64x(K[6], K[4], K[2], K[0]);
const __m256i K1 = _mm256_set_epi64x(K[7], K[5], K[3], K[1]);
const __m256i K2 = _mm256_set_epi64x(K[8], K[6], K[4], K[2]);
const __m256i K3 = _mm256_set_epi64x(K[0], K[7], K[5], K[3]);
const __m256i K4 = _mm256_set_epi64x(K[1], K[8], K[6], K[4]);
const __m256i K5 = _mm256_set_epi64x(K[2], K[0], K[7], K[5]);
const __m256i K6 = _mm256_set_epi64x(K[3], K[1], K[8], K[6]);
const __m256i K7 = _mm256_set_epi64x(K[4], K[2], K[0], K[7]);
const __m256i K8 = _mm256_set_epi64x(K[5], K[3], K[1], K[8]);
const __m256i ONE = _mm256_set_epi64x(1, 0, 0, 0);
const __m256i* in_mm = reinterpret_cast<const __m256i*>(in);
__m256i* out_mm = reinterpret_cast<__m256i*>(out);
while(blocks >= 2)
{
__m256i X0 = _mm256_loadu_si256(in_mm++);
__m256i X1 = _mm256_loadu_si256(in_mm++);
__m256i X2 = _mm256_loadu_si256(in_mm++);
__m256i X3 = _mm256_loadu_si256(in_mm++);
const __m256i T = _mm256_set_epi64x(T_64[0], T_64[1], T_64[2], 0);
__m256i R = _mm256_set_epi64x(0, 0, 0, 0);
interleave_epi64(X0, X1);
interleave_epi64(X2, X3);
THREEFISH_INJECT_KEY_2(X0, X1, X2, X3, R, K0, K1, 2, 3);
THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K1,K2,K3, 1, 2, 3);
THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K3,K4,K5, 2, 3, 1);
THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K5,K6,K7, 3, 1, 2);
THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K7,K8,K0, 1, 2, 3);
THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K0,K1,K2, 2, 3, 1);
THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K2,K3,K4, 3, 1, 2);
THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K4,K5,K6, 1, 2, 3);
THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K6,K7,K8, 2, 3, 1);
THREEFISH_ENC_2_8_ROUNDS(X0, X1, X2, X3, R, K8,K0,K1, 3, 1, 2);
deinterleave_epi64(X0, X1);
deinterleave_epi64(X2, X3);
_mm256_storeu_si256(out_mm++, X0);
_mm256_storeu_si256(out_mm++, X1);
_mm256_storeu_si256(out_mm++, X2);
_mm256_storeu_si256(out_mm++, X3);
blocks -= 2;
}
for(size_t i = 0; i != blocks; ++i)
{
__m256i X0 = _mm256_loadu_si256(in_mm++);
__m256i X1 = _mm256_loadu_si256(in_mm++);
const __m256i T = _mm256_set_epi64x(T_64[0], T_64[1], T_64[2], 0);
__m256i R = _mm256_set_epi64x(0, 0, 0, 0);
interleave_epi64(X0, X1);
THREEFISH_INJECT_KEY(X0, X1, R, K0, K1, 2, 3);
THREEFISH_ENC_8_ROUNDS(X0, X1, R, K1,K2,K3, 1, 2, 3);
THREEFISH_ENC_8_ROUNDS(X0, X1, R, K3,K4,K5, 2, 3, 1);
THREEFISH_ENC_8_ROUNDS(X0, X1, R, K5,K6,K7, 3, 1, 2);
THREEFISH_ENC_8_ROUNDS(X0, X1, R, K7,K8,K0, 1, 2, 3);
THREEFISH_ENC_8_ROUNDS(X0, X1, R, K0,K1,K2, 2, 3, 1);
THREEFISH_ENC_8_ROUNDS(X0, X1, R, K2,K3,K4, 3, 1, 2);
THREEFISH_ENC_8_ROUNDS(X0, X1, R, K4,K5,K6, 1, 2, 3);
THREEFISH_ENC_8_ROUNDS(X0, X1, R, K6,K7,K8, 2, 3, 1);
THREEFISH_ENC_8_ROUNDS(X0, X1, R, K8,K0,K1, 3, 1, 2);
deinterleave_epi64(X0, X1);
_mm256_storeu_si256(out_mm++, X0);
_mm256_storeu_si256(out_mm++, X1);
}
#undef THREEFISH_ENC_8_ROUNDS
#undef THREEFISH_ROUND
#undef THREEFISH_INJECT_KEY
#undef THREEFISH_ENC_2_8_ROUNDS
#undef THREEFISH_ROUND_2
#undef THREEFISH_INJECT_KEY_2
}
parasail_result_t *result = parasail_result_new();
#endif
#endif
int32_t i = 0;
int32_t j = 0;
int32_t end_query = 0;
int32_t end_ref = 0;
int64_t score = NEG_INF;
__m256i vNegInf = _mm256_set1_epi64x(NEG_INF);
__m256i vNegInf0 = _mm256_srli_si256_rpl(vNegInf, 8); /* shift in a 0 */
__m256i vOpen = _mm256_set1_epi64x(open);
__m256i vGap = _mm256_set1_epi64x(gap);
__m256i vOne = _mm256_set1_epi64x(1);
__m256i vN = _mm256_set1_epi64x(N);
__m256i vNegOne = _mm256_set1_epi64x(-1);
__m256i vI = _mm256_set_epi64x(0,1,2,3);
__m256i vJreset = _mm256_set_epi64x(0,-1,-2,-3);
__m256i vMaxScore = vNegInf;
__m256i vEndI = vNegInf;
__m256i vEndJ = vNegInf;
__m256i vILimit = _mm256_set1_epi64x(s1Len);
__m256i vILimit1 = _mm256_sub_epi64(vILimit, vOne);
__m256i vJLimit = _mm256_set1_epi64x(s2Len);
__m256i vJLimit1 = _mm256_sub_epi64(vJLimit, vOne);
/* convert _s1 from char to int in range 0-23 */
for (i=0; i<s1Len; ++i) {
s1[i] = matrix->mapper[(unsigned char)_s1[i]];
}
/* pad back of s1 with dummy values */
foo (long long *v)
{
return _mm256_set_epi64x (v[3], v[2], v[1], v[0]);
}
void Initialize() {
/*
Round constants for p_1:
01, 02, 05, 0a, 15, 0b, 17, 0e, 1d, 1b, 16, 0c
Round constants for p_2:
18, 11, 03, 07, 0f, 1f
Round constants for p_3:
1e, 1c, 19, 13, 06, 0d
*/
shuffleControlMaskFirstReg = _mm256_setr_epi8(
0, 1, 2, 3, 4, 5, 6, 7, //0
9, 10, 11, 12, 13, 14, 15, 8, //1
18, 19, 20, 21, 22, 23, 16, 17, //2
28, 29, 30, 31, 24, 25, 26, 27); //4
shuffleControlMaskSecondReg = _mm256_setr_epi8(
7, 0, 1, 2, 3, 4, 5, 6, //7
255, 255, 255, 255, 255, 255, 255, 255, //Setting it to 0xFF makes shuffle zero the bits
255, 255, 255, 255, 255, 255, 255, 255, //Setting it to 0xFF makes shuffle zero the bits
255, 255, 255, 255, 255, 255, 255, 255); //Setting it to 0xFF makes shuffle zero the bits
invShuffleControlMaskFirstReg = _mm256_setr_epi8(
0, 1, 2, 3, 4, 5, 6, 7, //0
15, 8, 9, 10, 11, 12, 13, 14, //1
22, 23, 16, 17, 18, 19, 20, 21, //2
28, 29, 30, 31, 24, 25, 26, 27); //4
invShuffleControlMaskSecondReg = _mm256_setr_epi8(
1, 2, 3, 4, 5, 6, 7, 0, //7
255, 255, 255, 255, 255, 255, 255, 255, //Setting it to 0xFF makes shuffle zero the bits
255, 255, 255, 255, 255, 255, 255, 255,
255, 255, 255, 255, 255, 255, 255, 255);
m256iAllOne = _mm256_set1_epi64x(0xFFFFFFFFFFFFFFFF);
//Set the bits to 1111'1111 in the column two, second row byte, if the roundconstant has a onebit on this indice
//p1
p1_constants_bit0[0] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit0[1] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit0[2] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit0[3] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit0[4] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit0[5] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit0[6] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit0[7] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit0[8] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit0[9] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit0[10] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit0[11] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit1[0] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit1[1] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit1[2] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit1[3] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit1[4] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit1[5] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit1[6] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit1[7] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit1[8] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit1[9] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit1[10] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit1[11] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit2[0] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit2[1] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit2[2] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit2[3] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit2[4] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit2[5] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit2[6] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit2[7] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit2[8] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit2[9] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit2[10] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit2[11] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit3[0] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit3[1] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit3[2] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit3[3] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit3[4] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit3[5] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit3[6] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit3[7] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit3[8] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit3[9] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit3[10] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit3[11] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit4[0] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit4[1] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit4[2] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit4[3] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit4[4] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit4[5] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit4[6] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit4[7] = _mm256_set_epi64x(0, 0, 0, 0);
p1_constants_bit4[8] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit4[9] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit4[10] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p1_constants_bit4[11] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit0[0] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit0[1] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit0[2] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit0[3] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit0[4] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit0[5] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit0[6] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit0[7] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit0[8] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit0[9] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit0[10] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit0[11] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit1[0] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit1[1] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit1[2] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit1[3] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit1[4] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit1[5] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit1[6] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit1[7] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit1[8] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit1[9] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit1[10] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit1[11] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit2[0] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit2[1] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit2[2] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit2[3] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit2[4] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit2[5] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit2[6] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit2[7] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit2[8] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit2[9] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit2[10] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit2[11] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit3[0] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit3[1] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit3[2] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit3[3] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit3[4] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit3[5] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit3[6] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit3[7] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit3[8] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit3[9] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit3[10] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit3[11] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit4[0] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit4[1] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit4[2] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit4[3] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit4[4] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit4[5] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit4[6] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit4[7] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit4[8] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit4[9] = _mm256_set_epi64x(0, 0, 0b0000000000000000000000000000000000000000000000001111111100000000, 0);
p4_constants_bit4[10] = _mm256_set_epi64x(0, 0, 0, 0);
p4_constants_bit4[11] = _mm256_set_epi64x(0, 0, 0, 0);
}
void encrypt_64blocks(u256 x[32], u256 rk[40][16]) {
int i, j;
u256 rc, tmp[8];
rc = _mm256_set_epi64x(0x000000FF000000FFull,
0x000000FF000000FFull,
0x000000FF000000FFull,
0x000000FF000000FFull);
for(i = 0; i < 40; i++){
//SubBytes
for(j = 0; j < 4; j++) {
tmp[7] = XOR(x[2 + 8*j], NOR(XOR(x[3 + 8*j], NOR(x[0 + 8*j],x[1 + 8*j])),XOR(x[7 + 8*j], NOR(x[4 + 8*j],x[5 + 8*j]))));
tmp[6] = XOR(x[3 + 8*j], NOR(x[0 + 8*j],x[1 + 8*j]));
tmp[5] = XOR(x[7 + 8*j], NOR(x[4 + 8*j],x[5 + 8*j]));
tmp[4] = XOR(x[4 + 8*j], NOR(XOR(x[2 + 8*j], NOR(XOR(x[3 + 8*j], NOR(x[0 + 8*j],x[1 + 8*j])),XOR(x[7 + 8*j], NOR(x[4 + 8*j],x[5 + 8*j])))),XOR(x[3 + 8*j], NOR(x[0 + 8*j],x[1 + 8*j]))));
tmp[3] = XOR(x[6 + 8*j], NOR(XOR(x[7 + 8*j], NOR(x[4 + 8*j],x[5 + 8*j])),x[4 + 8*j]));
tmp[2] = XOR(x[1 + 8*j], NOR(x[5 + 8*j],x[6 + 8*j]));
tmp[1] = XOR(x[0 + 8*j], NOR(XOR(x[1 + 8*j], NOR(x[5 + 8*j],x[6 + 8*j])),XOR(x[2 + 8*j], NOR(XOR(x[3 + 8*j], NOR(x[0 + 8*j],x[1 + 8*j])),XOR(x[7 + 8*j], NOR(x[4 + 8*j],x[5 + 8*j]))))));
tmp[0] = XOR(x[5 + 8*j], NOR(XOR(x[6 + 8*j], NOR(XOR(x[7 + 8*j], NOR(x[4 + 8*j],x[5 + 8*j])),x[4 + 8*j])),XOR(x[0 + 8*j], NOR(XOR(x[1 + 8*j], NOR(x[5 + 8*j],x[6 + 8*j])),XOR(x[2 + 8*j], NOR(XOR(x[3 + 8*j], NOR(x[0 + 8*j],x[1 + 8*j])),XOR(x[7 + 8*j], NOR(x[4 + 8*j],x[5 + 8*j]))))))));
x[0 + 8*j] = tmp[7];
x[1 + 8*j] = tmp[6];
x[2 + 8*j] = tmp[5];
x[3 + 8*j] = tmp[4];
x[4 + 8*j] = tmp[3];
x[5 + 8*j] = tmp[2];
x[6 + 8*j] = tmp[1];
x[7 + 8*j] = tmp[0];
}
//AddConstant
//This only adds c2. The other constants are added with the key
x[22] = XOR(x[22], rc);
//AddKey
x[0] = XOR(x[0], rk[i][0]);
x[1] = XOR(x[1], rk[i][1]);
x[2] = XOR(x[2], rk[i][2]);
x[3] = XOR(x[3], rk[i][3]);
x[4] = XOR(x[4], rk[i][4]);
x[5] = XOR(x[5], rk[i][5]);
x[6] = XOR(x[6], rk[i][6]);
x[7] = XOR(x[7], rk[i][7]);
x[8] = XOR(x[8], rk[i][8]);
x[9] = XOR(x[9], rk[i][9]);
x[10] = XOR(x[10], rk[i][10]);
x[11] = XOR(x[11], rk[i][11]);
x[12] = XOR(x[12], rk[i][12]);
x[13] = XOR(x[13], rk[i][13]);
x[14] = XOR(x[14], rk[i][14]);
x[15] = XOR(x[15], rk[i][15]);
//ShiftRows
x[8] = SR1(x[8]); x[16] = SR2(x[16]); x[24] = SR3(x[24]);
x[9] = SR1(x[9]); x[17] = SR2(x[17]); x[25] = SR3(x[25]);
x[10] = SR1(x[10]); x[18] = SR2(x[18]); x[26] = SR3(x[26]);
x[11] = SR1(x[11]); x[19] = SR2(x[19]); x[27] = SR3(x[27]);
x[12] = SR1(x[12]); x[20] = SR2(x[20]); x[28] = SR3(x[28]);
x[13] = SR1(x[13]); x[21] = SR2(x[21]); x[29] = SR3(x[29]);
x[14] = SR1(x[14]); x[22] = SR2(x[22]); x[30] = SR3(x[30]);
x[15] = SR1(x[15]); x[23] = SR2(x[23]); x[31] = SR3(x[31]);
//MixColumns
tmp[0] = x[24]; tmp[1] = x[25]; tmp[2] = x[26]; tmp[3] = x[27];
tmp[4] = x[28]; tmp[5] = x[29]; tmp[6] = x[30]; tmp[7] = x[31];
x[24] = XOR(x[16], x[0]); x[28] = XOR(x[20], x[4]);
x[25] = XOR(x[17], x[1]); x[29] = XOR(x[21], x[5]);
x[26] = XOR(x[18], x[2]); x[30] = XOR(x[22], x[6]);
x[27] = XOR(x[19], x[3]); x[31] = XOR(x[23], x[7]);
x[16] = XOR(x[8], x[16]); x[20] = XOR(x[12], x[20]);
x[17] = XOR(x[9], x[17]); x[21] = XOR(x[13], x[21]);
x[18] = XOR(x[10], x[18]); x[22] = XOR(x[14], x[22]);
x[19] = XOR(x[11], x[19]); x[23] = XOR(x[15], x[23]);
x[8] = x[0]; x[12] = x[4];
x[9] = x[1]; x[13] = x[5];
x[10] = x[2]; x[14] = x[6];
x[11] = x[3]; x[15] = x[7];
x[0] = XOR(tmp[0], x[24]); x[4] = XOR(tmp[4], x[28]);
x[1] = XOR(tmp[1], x[25]); x[5] = XOR(tmp[5], x[29]);
x[2] = XOR(tmp[2], x[26]); x[6] = XOR(tmp[6], x[30]);
x[3] = XOR(tmp[3], x[27]); x[7] = XOR(tmp[7], x[31]);
}
}