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

v23.10

Browse Source
This commit is contained in:
Jay D Dee
2023-11-15 11:05:41 -05:00
parent 0a78013cbe
commit f3fde95f27
32 changed files with 1074 additions and 1027 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
algo/blake/blake256-hash.c
View File

@@ -429,7 +429,7 @@ void blake256_transform_le( uint32_t *H, const uint32_t *buf,
#define BLAKE256_4X32_BLOCK_BSWAP32 \
{ \
v128_t shuf_bswap32 = v128_set64( 0x0c0d0e0f08090a0b, \
0x0405060700010203 ); \
0x0405060700010203 ); \
M0 = _mm_shuffle_epi8( buf[ 0], shuf_bswap32 ); \
M1 = _mm_shuffle_epi8( buf[ 1], shuf_bswap32 ); \
M2 = _mm_shuffle_epi8( buf[ 2], shuf_bswap32 ); \
@@ -931,14 +931,14 @@ void blake256_4x32_final_rounds_le( void *final_hash, const void *midstate,
const v128_t shuf_bswap32 =
v128_set64( 0x0c0d0e0f08090a0b, 0x0405060700010203 );
H[0] = _mm_shuffle_epi8( mm128_xor3( V8, V0, h[0] ), shuf_bswap32 );
H[1] = _mm_shuffle_epi8( mm128_xor3( V9, V1, h[1] ), shuf_bswap32 );
H[2] = _mm_shuffle_epi8( mm128_xor3( VA, V2, h[2] ), shuf_bswap32 );
H[3] = _mm_shuffle_epi8( mm128_xor3( VB, V3, h[3] ), shuf_bswap32 );
H[4] = _mm_shuffle_epi8( mm128_xor3( VC, V4, h[4] ), shuf_bswap32 );
H[5] = _mm_shuffle_epi8( mm128_xor3( VD, V5, h[5] ), shuf_bswap32 );
H[6] = _mm_shuffle_epi8( mm128_xor3( VE, V6, h[6] ), shuf_bswap32 );
H[7] = _mm_shuffle_epi8( mm128_xor3( VF, V7, h[7] ), shuf_bswap32 );
H[0] = _mm_shuffle_epi8( v128_xor3( V8, V0, h[0] ), shuf_bswap32 );
H[1] = _mm_shuffle_epi8( v128_xor3( V9, V1, h[1] ), shuf_bswap32 );
H[2] = _mm_shuffle_epi8( v128_xor3( VA, V2, h[2] ), shuf_bswap32 );
H[3] = _mm_shuffle_epi8( v128_xor3( VB, V3, h[3] ), shuf_bswap32 );
H[4] = _mm_shuffle_epi8( v128_xor3( VC, V4, h[4] ), shuf_bswap32 );
H[5] = _mm_shuffle_epi8( v128_xor3( VD, V5, h[5] ), shuf_bswap32 );
H[6] = _mm_shuffle_epi8( v128_xor3( VE, V6, h[6] ), shuf_bswap32 );
H[7] = _mm_shuffle_epi8( v128_xor3( VF, V7, h[7] ), shuf_bswap32 );
#else

40
algo/blake/sph_blake2b.c
View File

@@ -131,47 +131,7 @@
V[7] = v128_alignr64( V6, V7, 1 ); \
}
/*
#elif defined(__SSE2__)
// always true
#define BLAKE2B_G( Va, Vb, Vc, Vd, Sa, Sb, Sc, Sd ) \
{ \
Va = _mm_add_epi64( Va, _mm_add_epi64( Vb, \
_mm_set_epi64x( m[ sigmaR[ Sc ] ], m[ sigmaR[ Sa ] ] ) ) ); \
Vd = mm128_swap64_32( _mm_xor_si128( Vd, Va ) ); \
Vc = _mm_add_epi64( Vc, Vd ); \
Vb = mm128_shuflr64_24( _mm_xor_si128( Vb, Vc ) ); \
\
Va = _mm_add_epi64( Va, _mm_add_epi64( Vb, \
_mm_set_epi64x( m[ sigmaR[ Sd ] ], m[ sigmaR[ Sb ] ] ) ) ); \
Vd = mm128_shuflr64_16( _mm_xor_si128( Vd, Va ) ); \
Vc = _mm_add_epi64( Vc, Vd ); \
Vb = mm128_ror_64( _mm_xor_si128( Vb, Vc ), 63 ); \
}
#define BLAKE2B_ROUND( R ) \
{ \
v128_t *V = (v128_t*)v; \
v128_t V2, V3, V6, V7; \
const uint8_t *sigmaR = sigma[R]; \
BLAKE2B_G( V[0], V[2], V[4], V[6], 0, 1, 2, 3 ); \
BLAKE2B_G( V[1], V[3], V[5], V[7], 4, 5, 6, 7 ); \
V2 = mm128_alignr_64( V[3], V[2], 1 ); \
V3 = mm128_alignr_64( V[2], V[3], 1 ); \
V6 = mm128_alignr_64( V[6], V[7], 1 ); \
V7 = mm128_alignr_64( V[7], V[6], 1 ); \
BLAKE2B_G( V[0], V2, V[5], V6, 8, 9, 10, 11 ); \
BLAKE2B_G( V[1], V3, V[4], V7, 12, 13, 14, 15 ); \
V[2] = mm128_alignr_64( V2, V3, 1 ); \
V[3] = mm128_alignr_64( V3, V2, 1 ); \
V[6] = mm128_alignr_64( V7, V6, 1 ); \
V[7] = mm128_alignr_64( V6, V7, 1 ); \
}
*/
#else
// never used, SSE2 is always available
#ifndef ROTR64
#define ROTR64(x, y) (((x) >> (y)) ^ ((x) << (64 - (y))))