popov/cpuminer-opt-gpu
1
0
Fork 0
mirror of https://github.com/JayDDee/cpuminer-opt.git synced 2025-09-17 23:44:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

v3.20.1

Browse Source
This commit is contained in:
Jay D Dee
2022-07-26 18:36:40 -04:00
parent 40d07c0097
commit 1321ac474c
17 changed files with 568 additions and 2422 deletions
Download Patch File Download Diff File Expand all files Collapse all files

174
algo/blake/blake2b-hash-4way.c
View File

@@ -52,6 +52,180 @@ static const uint8_t sigma[12][16] =
};
#define Z00 0
#define Z01 1
#define Z02 2
#define Z03 3
#define Z04 4
#define Z05 5
#define Z06 6
#define Z07 7
#define Z08 8
#define Z09 9
#define Z0A A
#define Z0B B
#define Z0C C
#define Z0D D
#define Z0E E
#define Z0F F
#define Z10 E
#define Z11 A
#define Z12 4
#define Z13 8
#define Z14 9
#define Z15 F
#define Z16 D
#define Z17 6
#define Z18 1
#define Z19 C
#define Z1A 0
#define Z1B 2
#define Z1C B
#define Z1D 7
#define Z1E 5
#define Z1F 3
#define Z20 B
#define Z21 8
#define Z22 C
#define Z23 0
#define Z24 5
#define Z25 2
#define Z26 F
#define Z27 D
#define Z28 A
#define Z29 E
#define Z2A 3
#define Z2B 6
#define Z2C 7
#define Z2D 1
#define Z2E 9
#define Z2F 4
#define Z30 7
#define Z31 9
#define Z32 3
#define Z33 1
#define Z34 D
#define Z35 C
#define Z36 B
#define Z37 E
#define Z38 2
#define Z39 6
#define Z3A 5
#define Z3B A
#define Z3C 4
#define Z3D 0
#define Z3E F
#define Z3F 8
#define Z40 9
#define Z41 0
#define Z42 5
#define Z43 7
#define Z44 2
#define Z45 4
#define Z46 A
#define Z47 F
#define Z48 E
#define Z49 1
#define Z4A B
#define Z4B C
#define Z4C 6
#define Z4D 8
#define Z4E 3
#define Z4F D
#define Z50 2
#define Z51 C
#define Z52 6
#define Z53 A
#define Z54 0
#define Z55 B
#define Z56 8
#define Z57 3
#define Z58 4
#define Z59 D
#define Z5A 7
#define Z5B 5
#define Z5C F
#define Z5D E
#define Z5E 1
#define Z5F 9
#define Z60 C
#define Z61 5
#define Z62 1
#define Z63 F
#define Z64 E
#define Z65 D
#define Z66 4
#define Z67 A
#define Z68 0
#define Z69 7
#define Z6A 6
#define Z6B 3
#define Z6C 9
#define Z6D 2
#define Z6E 8
#define Z6F B
#define Z70 D
#define Z71 B
#define Z72 7
#define Z73 E
#define Z74 C
#define Z75 1
#define Z76 3
#define Z77 9
#define Z78 5
#define Z79 0
#define Z7A F
#define Z7B 4
#define Z7C 8
#define Z7D 6
#define Z7E 2
#define Z7F A
#define Z80 6
#define Z81 F
#define Z82 E
#define Z83 9
#define Z84 B
#define Z85 3
#define Z86 0
#define Z87 8
#define Z88 C
#define Z89 2
#define Z8A D
#define Z8B 7
#define Z8C 1
#define Z8D 4
#define Z8E A
#define Z8F 5
#define Z90 A
#define Z91 2
#define Z92 8
#define Z93 4
#define Z94 7
#define Z95 6
#define Z96 1
#define Z97 5
#define Z98 F
#define Z99 B
#define Z9A 9
#define Z9B E
#define Z9C 3
#define Z9D C
#define Z9E D
#define Z9F 0
#define Mx(r, i) Mx_(Z ## r ## i)
#define Mx_(n) Mx__(n)
#define Mx__(n) M ## n
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define B2B8W_G(a, b, c, d, x, y) \

1
algo/blake/blake512-hash-4way.c
View File

@@ -757,7 +757,6 @@ void blake512_8way_final_le( blake_8way_big_context *sc, void *hash,
GB_8WAY(Mx(1, C), Mx(1, D), CBx(1, C), CBx(1, D), V2, V7, V8, VD);
GB_8WAY(Mx(1, E), Mx(1, F), CBx(1, E), CBx(1, F), V3, V4, V9, VE);
// remaining rounds
ROUND_B_8WAY(2);
ROUND_B_8WAY(3);

158
algo/blake/sph_blake2b.c
View File

@@ -30,16 +30,10 @@
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include "simd-utils.h"
#include "algo/sha/sph_types.h"
#include "sph_blake2b.h"
// Cyclic right rotation.
#ifndef ROTR64
#define ROTR64(x, y) (((x) >> (y)) ^ ((x) << (64 - (y))))
#endif
// Little-endian byte access.
#define B2B_GET64(p) \
@@ -54,45 +48,131 @@
// G Mixing function.
#define B2B_G(a, b, c, d, x, y) { \
v[a] = v[a] + v[b] + x; \
v[d] = ROTR64(v[d] ^ v[a], 32); \
v[c] = v[c] + v[d]; \
v[b] = ROTR64(v[b] ^ v[c], 24); \
v[a] = v[a] + v[b] + y; \
v[d] = ROTR64(v[d] ^ v[a], 16); \
v[c] = v[c] + v[d]; \
v[b] = ROTR64(v[b] ^ v[c], 63); }
#if defined(__AVX2__)
#define BLAKE2B_G( R, Sa, Sb, Sc, Sd, Na, Nb ) \
{ \
V[0] = _mm256_add_epi64( V[0], _mm256_add_epi64( V[1], \
_mm256_set_epi64x( m[ sigma[R][Sd] ], m[ sigma[R][Sc] ], \
m[ sigma[R][Sb] ], m[ sigma[R][Sa] ] ) ) ); \
V[3] = mm256_ror_64( _mm256_xor_si256( V[3], V[0] ), Na ); \
V[2] = _mm256_add_epi64( V[2], V[3] ); \
V[1] = mm256_ror_64( _mm256_xor_si256( V[1], V[2] ), Nb ); \
}
#define BLAKE2B_ROUND( R ) \
{ \
__m256i *V = (__m256i*)v; \
BLAKE2B_G( R, 0, 2, 4, 6, 32, 24 ); \
BLAKE2B_G( R, 1, 3, 5, 7, 16, 63 ); \
V[3] = mm256_shufll_64( V[3] ); \
V[2] = mm256_swap_128( V[2] ); \
V[1] = mm256_shuflr_64( V[1] ); \
BLAKE2B_G( R, 8, 10, 12, 14, 32, 24 ); \
BLAKE2B_G( R, 9, 11, 13, 15, 16, 63 ); \
V[3] = mm256_shuflr_64( V[3] ); \
V[2] = mm256_swap_128( V[2] ); \
V[1] = mm256_shufll_64( V[1] ); \
}
#elif defined(__SSSE3__)
#define BLAKE2B_G( R, Va, Vb, Vc, Vd, Sa, Sb, Na, Nb ) \
{ \
Va = _mm_add_epi64( Va, _mm_add_epi64( Vb, \
_mm_set_epi64x( m[ sigma[R][Sb] ], m[ sigma[R][Sa] ] ) ) ); \
Vd = mm128_ror_64( _mm_xor_si128( Vd, Va ), Na ); \
Vc = _mm_add_epi64( Vc, Vd ); \
Vb = mm128_ror_64( _mm_xor_si128( Vb, Vc ), Nb ); \
}
#define BLAKE2B_ROUND( R ) \
{ \
__m128i *V = (__m128i*)v; \
__m128i V2, V3, V6, V7; \
BLAKE2B_G( R, V[0], V[2], V[4], V[6], 0, 2, 32, 24 ); \
BLAKE2B_G( R, V[0], V[2], V[4], V[6], 1, 3, 16, 63 ); \
BLAKE2B_G( R, V[1], V[3], V[5], V[7], 4, 6, 32, 24 ); \
BLAKE2B_G( R, V[1], V[3], V[5], V[7], 5, 7, 16, 63 ); \
V2 = mm128_shufl2r_64( V[2], V[3] ); \
V3 = mm128_shufl2r_64( V[3], V[2] ); \
V6 = mm128_shufl2l_64( V[6], V[7] ); \
V7 = mm128_shufl2l_64( V[7], V[6] ); \
BLAKE2B_G( R, V[0], V2, V[5], V6, 8, 10, 32, 24 ); \
BLAKE2B_G( R, V[0], V2, V[5], V6, 9, 11, 16, 63 ); \
BLAKE2B_G( R, V[1], V3, V[4], V7, 12, 14, 32, 24 ); \
BLAKE2B_G( R, V[1], V3, V[4], V7, 13, 15, 16, 63 ); \
V[2] = mm128_shufl2l_64( V2, V3 ); \
V[3] = mm128_shufl2l_64( V3, V2 ); \
V[6] = mm128_shufl2r_64( V6, V7 ); \
V[7] = mm128_shufl2r_64( V7, V6 ); \
}
#else
#ifndef ROTR64
#define ROTR64(x, y) (((x) >> (y)) ^ ((x) << (64 - (y))))
#endif
#define BLAKE2B_G( R, Va, Vb, Vc, Vd, Sa, Sb ) \
{ \
Va = Va + Vb + m[ sigma[R][Sa] ]; \
Vd = ROTR64( Vd ^ Va, 32 ); \
Vc = Vc + Vd; \
Vb = ROTR64( Vb ^ Vc, 24 ); \
Va = Va + Vb + m[ sigma[R][Sb] ]; \
Vd = ROTR64( Vd ^ Va, 16 ); \
Vc = Vc + Vd; \
Vb = ROTR64( Vb ^ Vc, 63 ); \
}
#define BLAKE2B_ROUND( R ) \
{ \
BLAKE2B_G( R, v[ 0], v[ 4], v[ 8], v[12], 0, 1 ); \
BLAKE2B_G( R, v[ 1], v[ 5], v[ 9], v[13], 2, 3 ); \
BLAKE2B_G( R, v[ 2], v[ 6], v[10], v[14], 4, 5 ); \
BLAKE2B_G( R, v[ 3], v[ 7], v[11], v[15], 6, 7 ); \
BLAKE2B_G( R, v[ 0], v[ 5], v[10], v[15], 8, 9 ); \
BLAKE2B_G( R, v[ 1], v[ 6], v[11], v[12], 10, 11 ); \
BLAKE2B_G( R, v[ 2], v[ 7], v[ 8], v[13], 12, 13 ); \
BLAKE2B_G( R, v[ 3], v[ 4], v[ 9], v[14], 14, 15 ); \
}
#endif
// Initialization Vector.
static const uint64_t blake2b_iv[8] = {
static const uint64_t blake2b_iv[8] __attribute__ ((aligned (32))) =
{
0x6A09E667F3BCC908, 0xBB67AE8584CAA73B,
0x3C6EF372FE94F82B, 0xA54FF53A5F1D36F1,
0x510E527FADE682D1, 0x9B05688C2B3E6C1F,
0x1F83D9ABFB41BD6B, 0x5BE0CD19137E2179
};
static const uint8_t sigma[12][16] __attribute__ ((aligned (32))) =
{
{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
{ 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 },
{ 11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4 },
{ 7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8 },
{ 9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13 },
{ 2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9 },
{ 12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11 },
{ 13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10 },
{ 6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5 },
{ 10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13, 0 },
{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
{ 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 }
};
// Compression function. "last" flag indicates last block.
static void blake2b_compress( sph_blake2b_ctx *ctx, int last )
{
const uint8_t sigma[12][16] = {
{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
{ 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 },
{ 11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4 },
{ 7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8 },
{ 9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13 },
{ 2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9 },
{ 12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11 },
{ 13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10 },
{ 6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5 },
{ 10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13, 0 },
{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
{ 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 }
};
int i;
uint64_t v[16], m[16];
uint64_t v[16] __attribute__ ((aligned (32)));
uint64_t m[16] __attribute__ ((aligned (32)));
int i;
for (i = 0; i < 8; i++) { // init work variables
v[i] = ctx->h[i];
@@ -106,16 +186,8 @@ static void blake2b_compress( sph_blake2b_ctx *ctx, int last )
for (i = 0; i < 16; i++) // get little-endian words
m[i] = B2B_GET64(&ctx->b[8 * i]);
for (i = 0; i < 12; i++) { // twelve rounds
B2B_G( 0, 4, 8, 12, m[sigma[i][ 0]], m[sigma[i][ 1]]);
B2B_G( 1, 5, 9, 13, m[sigma[i][ 2]], m[sigma[i][ 3]]);
B2B_G( 2, 6, 10, 14, m[sigma[i][ 4]], m[sigma[i][ 5]]);
B2B_G( 3, 7, 11, 15, m[sigma[i][ 6]], m[sigma[i][ 7]]);
B2B_G( 0, 5, 10, 15, m[sigma[i][ 8]], m[sigma[i][ 9]]);
B2B_G( 1, 6, 11, 12, m[sigma[i][10]], m[sigma[i][11]]);
B2B_G( 2, 7, 8, 13, m[sigma[i][12]], m[sigma[i][13]]);
B2B_G( 3, 4, 9, 14, m[sigma[i][14]], m[sigma[i][15]]);
}
for (i = 0; i < 12; i++)
BLAKE2B_ROUND( i );
for( i = 0; i < 8; ++i )
ctx->h[i] ^= v[i] ^ v[i + 8];