mirror of
https://github.com/henrydcase/pqc.git
synced 2024-11-27 09:51:30 +00:00
289 lines
8.7 KiB
C
289 lines
8.7 KiB
C
|
#include "crypto_encode_653x1541round.h"
|
||
|
|
#include <immintrin.h>
|
||
|
|
/* auto-generated; do not edit */
|
||
|
|
|
||
|
|
#define int16 int16_t
|
||
|
|
#define uint16 uint16_t
|
||
|
|
#define uint32 uint32_t
|
||
|
|
|
||
|
|
void PQCLEAN_SNTRUP653_AVX2_crypto_encode_653x1541round(unsigned char *out, const void *v) {
|
||
|
|
const int16 *R0 = v;
|
||
|
|
/* XXX: caller could overlap R with input */
|
||
|
|
uint16 R[327];
|
||
|
|
long i;
|
||
|
|
const uint16 *reading;
|
||
|
|
uint16 *writing;
|
||
|
|
uint16 r0, r1;
|
||
|
|
uint32 r2;
|
||
|
|
uint32 s0;
|
||
|
|
|
||
|
|
reading = (uint16 *) R0;
|
||
|
|
writing = R;
|
||
|
|
i = 41;
|
||
|
|
while (i > 0) {
|
||
|
|
__m256i x, y;
|
||
|
|
--i;
|
||
|
|
if (!i) {
|
||
|
|
reading -= 4;
|
||
|
|
writing -= 2;
|
||
|
|
out -= 2;
|
||
|
|
}
|
||
|
|
x = _mm256_loadu_si256((__m256i *) reading);
|
||
|
|
x = _mm256_mulhrs_epi16(x, _mm256_set1_epi16(10923));
|
||
|
|
x = _mm256_add_epi16(x, _mm256_add_epi16(x, x));
|
||
|
|
x = _mm256_add_epi16(x, _mm256_set1_epi16(2310));
|
||
|
|
x &= _mm256_set1_epi16(16383);
|
||
|
|
x = _mm256_mulhi_epi16(x, _mm256_set1_epi16(21846));
|
||
|
|
y = x & _mm256_set1_epi32(65535);
|
||
|
|
x = _mm256_srli_epi32(x, 16);
|
||
|
|
x = _mm256_mullo_epi32(x, _mm256_set1_epi32(1541));
|
||
|
|
x = _mm256_add_epi32(y, x);
|
||
|
|
x = _mm256_shuffle_epi8(x, _mm256_set_epi8(
|
||
|
|
12, 8, 4, 0, 12, 8, 4, 0, 14, 13, 10, 9, 6, 5, 2, 1,
|
||
|
|
12, 8, 4, 0, 12, 8, 4, 0, 14, 13, 10, 9, 6, 5, 2, 1
|
||
|
|
));
|
||
|
|
x = _mm256_permute4x64_epi64(x, 0xd8);
|
||
|
|
_mm_storeu_si128((__m128i *) writing, _mm256_extractf128_si256(x, 0));
|
||
|
|
s0 = _mm256_extract_epi32(x, 4);
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 = _mm256_extract_epi32(x, 6);
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
reading += 16;
|
||
|
|
writing += 8;
|
||
|
|
}
|
||
|
|
R[326] = (((3 * ((10923 * R0[652] + 16384) >> 15) + 2310) & 16383) * 10923) >> 15;
|
||
|
|
|
||
|
|
reading = (uint16 *) R;
|
||
|
|
writing = R;
|
||
|
|
i = 11;
|
||
|
|
while (i > 0) {
|
||
|
|
__m256i x, x2, y, y2;
|
||
|
|
--i;
|
||
|
|
if (!i) {
|
||
|
|
reading -= 26;
|
||
|
|
writing -= 13;
|
||
|
|
out -= 26;
|
||
|
|
}
|
||
|
|
x = _mm256_loadu_si256((__m256i *) (reading + 0));
|
||
|
|
x2 = _mm256_loadu_si256((__m256i *) (reading + 16));
|
||
|
|
y = x & _mm256_set1_epi32(65535);
|
||
|
|
y2 = x2 & _mm256_set1_epi32(65535);
|
||
|
|
x = _mm256_srli_epi32(x, 16);
|
||
|
|
x2 = _mm256_srli_epi32(x2, 16);
|
||
|
|
x = _mm256_mullo_epi32(x, _mm256_set1_epi32(9277));
|
||
|
|
x2 = _mm256_mullo_epi32(x2, _mm256_set1_epi32(9277));
|
||
|
|
x = _mm256_add_epi32(y, x);
|
||
|
|
x2 = _mm256_add_epi32(y2, x2);
|
||
|
|
x = _mm256_shuffle_epi8(x, _mm256_set_epi8(
|
||
|
|
15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0,
|
||
|
|
15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0
|
||
|
|
));
|
||
|
|
x2 = _mm256_shuffle_epi8(x2, _mm256_set_epi8(
|
||
|
|
15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0,
|
||
|
|
15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0
|
||
|
|
));
|
||
|
|
x = _mm256_permute4x64_epi64(x, 0xd8);
|
||
|
|
x2 = _mm256_permute4x64_epi64(x2, 0xd8);
|
||
|
|
_mm256_storeu_si256((__m256i *) writing, _mm256_permute2f128_si256(x, x2, 0x31));
|
||
|
|
_mm256_storeu_si256((__m256i *) out, _mm256_permute2f128_si256(x, x2, 0x20));
|
||
|
|
reading += 32;
|
||
|
|
writing += 16;
|
||
|
|
out += 32;
|
||
|
|
}
|
||
|
|
R[163] = R[326];
|
||
|
|
|
||
|
|
reading = (uint16 *) R;
|
||
|
|
writing = R;
|
||
|
|
i = 11;
|
||
|
|
while (i > 0) {
|
||
|
|
__m256i x, y;
|
||
|
|
--i;
|
||
|
|
if (!i) {
|
||
|
|
reading -= 12;
|
||
|
|
writing -= 6;
|
||
|
|
out -= 6;
|
||
|
|
}
|
||
|
|
x = _mm256_loadu_si256((__m256i *) reading);
|
||
|
|
y = x & _mm256_set1_epi32(65535);
|
||
|
|
x = _mm256_srli_epi32(x, 16);
|
||
|
|
x = _mm256_mullo_epi32(x, _mm256_set1_epi32(1314));
|
||
|
|
x = _mm256_add_epi32(y, x);
|
||
|
|
x = _mm256_shuffle_epi8(x, _mm256_set_epi8(
|
||
|
|
12, 8, 4, 0, 12, 8, 4, 0, 14, 13, 10, 9, 6, 5, 2, 1,
|
||
|
|
12, 8, 4, 0, 12, 8, 4, 0, 14, 13, 10, 9, 6, 5, 2, 1
|
||
|
|
));
|
||
|
|
x = _mm256_permute4x64_epi64(x, 0xd8);
|
||
|
|
_mm_storeu_si128((__m128i *) writing, _mm256_extractf128_si256(x, 0));
|
||
|
|
s0 = _mm256_extract_epi32(x, 4);
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 = _mm256_extract_epi32(x, 6);
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
reading += 16;
|
||
|
|
writing += 8;
|
||
|
|
}
|
||
|
|
|
||
|
|
reading = (uint16 *) R;
|
||
|
|
writing = R;
|
||
|
|
i = 3;
|
||
|
|
while (i > 0) {
|
||
|
|
__m256i x, x2, y, y2;
|
||
|
|
--i;
|
||
|
|
if (!i) {
|
||
|
|
reading -= 14;
|
||
|
|
writing -= 7;
|
||
|
|
out -= 14;
|
||
|
|
}
|
||
|
|
x = _mm256_loadu_si256((__m256i *) (reading + 0));
|
||
|
|
x2 = _mm256_loadu_si256((__m256i *) (reading + 16));
|
||
|
|
y = x & _mm256_set1_epi32(65535);
|
||
|
|
y2 = x2 & _mm256_set1_epi32(65535);
|
||
|
|
x = _mm256_srli_epi32(x, 16);
|
||
|
|
x2 = _mm256_srli_epi32(x2, 16);
|
||
|
|
x = _mm256_mullo_epi32(x, _mm256_set1_epi32(6745));
|
||
|
|
x2 = _mm256_mullo_epi32(x2, _mm256_set1_epi32(6745));
|
||
|
|
x = _mm256_add_epi32(y, x);
|
||
|
|
x2 = _mm256_add_epi32(y2, x2);
|
||
|
|
x = _mm256_shuffle_epi8(x, _mm256_set_epi8(
|
||
|
|
15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0,
|
||
|
|
15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0
|
||
|
|
));
|
||
|
|
x2 = _mm256_shuffle_epi8(x2, _mm256_set_epi8(
|
||
|
|
15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0,
|
||
|
|
15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0
|
||
|
|
));
|
||
|
|
x = _mm256_permute4x64_epi64(x, 0xd8);
|
||
|
|
x2 = _mm256_permute4x64_epi64(x2, 0xd8);
|
||
|
|
_mm256_storeu_si256((__m256i *) writing, _mm256_permute2f128_si256(x, x2, 0x31));
|
||
|
|
_mm256_storeu_si256((__m256i *) out, _mm256_permute2f128_si256(x, x2, 0x20));
|
||
|
|
reading += 32;
|
||
|
|
writing += 16;
|
||
|
|
out += 32;
|
||
|
|
}
|
||
|
|
|
||
|
|
reading = (uint16 *) R;
|
||
|
|
writing = R;
|
||
|
|
i = 3;
|
||
|
|
while (i > 0) {
|
||
|
|
__m256i x, y;
|
||
|
|
--i;
|
||
|
|
if (!i) {
|
||
|
|
reading -= 8;
|
||
|
|
writing -= 4;
|
||
|
|
out -= 4;
|
||
|
|
}
|
||
|
|
x = _mm256_loadu_si256((__m256i *) reading);
|
||
|
|
y = x & _mm256_set1_epi32(65535);
|
||
|
|
x = _mm256_srli_epi32(x, 16);
|
||
|
|
x = _mm256_mullo_epi32(x, _mm256_set1_epi32(695));
|
||
|
|
x = _mm256_add_epi32(y, x);
|
||
|
|
x = _mm256_shuffle_epi8(x, _mm256_set_epi8(
|
||
|
|
12, 8, 4, 0, 12, 8, 4, 0, 14, 13, 10, 9, 6, 5, 2, 1,
|
||
|
|
12, 8, 4, 0, 12, 8, 4, 0, 14, 13, 10, 9, 6, 5, 2, 1
|
||
|
|
));
|
||
|
|
x = _mm256_permute4x64_epi64(x, 0xd8);
|
||
|
|
_mm_storeu_si128((__m128i *) writing, _mm256_extractf128_si256(x, 0));
|
||
|
|
s0 = _mm256_extract_epi32(x, 4);
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 = _mm256_extract_epi32(x, 6);
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
s0 >>= 8;
|
||
|
|
*out++ = s0;
|
||
|
|
reading += 16;
|
||
|
|
writing += 8;
|
||
|
|
}
|
||
|
|
R[20] = R[40];
|
||
|
|
|
||
|
|
for (i = 0; i < 10; ++i) {
|
||
|
|
r0 = R[2 * i];
|
||
|
|
r1 = R[2 * i + 1];
|
||
|
|
r2 = r0 + r1 * (uint32)1887;
|
||
|
|
*out++ = r2;
|
||
|
|
r2 >>= 8;
|
||
|
|
R[i] = r2;
|
||
|
|
}
|
||
|
|
R[10] = R[20];
|
||
|
|
|
||
|
|
for (i = 0; i < 5; ++i) {
|
||
|
|
r0 = R[2 * i];
|
||
|
|
r1 = R[2 * i + 1];
|
||
|
|
r2 = r0 + r1 * (uint32)13910;
|
||
|
|
*out++ = r2;
|
||
|
|
r2 >>= 8;
|
||
|
|
*out++ = r2;
|
||
|
|
r2 >>= 8;
|
||
|
|
R[i] = r2;
|
||
|
|
}
|
||
|
|
R[5] = R[10];
|
||
|
|
|
||
|
|
for (i = 0; i < 2; ++i) {
|
||
|
|
r0 = R[2 * i];
|
||
|
|
r1 = R[2 * i + 1];
|
||
|
|
r2 = r0 + r1 * (uint32)2953;
|
||
|
|
*out++ = r2;
|
||
|
|
r2 >>= 8;
|
||
|
|
*out++ = r2;
|
||
|
|
r2 >>= 8;
|
||
|
|
R[i] = r2;
|
||
|
|
}
|
||
|
|
r0 = R[4];
|
||
|
|
r1 = R[5];
|
||
|
|
r2 = r0 + r1 * (uint32)2953;
|
||
|
|
*out++ = r2;
|
||
|
|
r2 >>= 8;
|
||
|
|
R[2] = r2;
|
||
|
|
|
||
|
|
r0 = R[0];
|
||
|
|
r1 = R[1];
|
||
|
|
r2 = r0 + r1 * (uint32)134;
|
||
|
|
*out++ = r2;
|
||
|
|
r2 >>= 8;
|
||
|
|
R[0] = r2;
|
||
|
|
R[1] = R[2];
|
||
|
|
|
||
|
|
r0 = R[0];
|
||
|
|
r1 = R[1];
|
||
|
|
r2 = r0 + r1 * (uint32)71;
|
||
|
|
*out++ = r2;
|
||
|
|
r2 >>= 8;
|
||
|
|
R[0] = r2;
|
||
|
|
|
||
|
|
r0 = R[0];
|
||
|
|
*out++ = r0;
|
||
|
|
r0 >>= 8;
|
||
|
|
*out++ = r0; /*clang-analyzer-deadcode.DeadStores*/ /*r0 >>= 8;*/
|
||
|
|
}
|