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

v3.23.0

Browse Source
This commit is contained in:
Jay D Dee
2023-08-30 20:15:48 -04:00
parent 57a6b7b58b
commit 4378d2f841
72 changed files with 10184 additions and 2182 deletions
Download Patch File Download Diff File Expand all files Collapse all files

216
algo/skein/skein-hash-4way.c
View File

@@ -285,7 +285,7 @@ static const uint64_t IV512[] = {
#define SKBI(k, s, i) XCAT(k, XCAT(XCAT(XCAT(M9_, s), _), i))
#define SKBT(t, s, v) XCAT(t, XCAT(XCAT(XCAT(M3_, s), _), v))
#define READ_STATE_BIG(sc) do { \
#define READ_STATE_BIG(sc) \
h0 = (sc)->h0; \
h1 = (sc)->h1; \
h2 = (sc)->h2; \
@@ -294,10 +294,9 @@ static const uint64_t IV512[] = {
h5 = (sc)->h5; \
h6 = (sc)->h6; \
h7 = (sc)->h7; \
bcount = sc->bcount; \
} while (0)
bcount = sc->bcount;
#define WRITE_STATE_BIG(sc) do { \
#define WRITE_STATE_BIG(sc) \
(sc)->h0 = h0; \
(sc)->h1 = h1; \
(sc)->h2 = h2; \
@@ -306,62 +305,54 @@ static const uint64_t IV512[] = {
(sc)->h5 = h5; \
(sc)->h6 = h6; \
(sc)->h7 = h7; \
sc->bcount = bcount; \
} while (0)
sc->bcount = bcount;
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define TFBIG_KINIT_8WAY( k0, k1, k2, k3, k4, k5, k6, k7, k8, t0, t1, t2 ) \
do { \
k8 = mm512_xor3( mm512_xor3( k0, k1, k2 ), mm512_xor3( k3, k4, k5 ), \
mm512_xor3( k6, k7, m512_const1_64( 0x1BD11BDAA9FC1A22) ));\
t2 = t0 ^ t1; \
} while (0)
k8 = mm512_xor3( mm512_xor3( k0, k1, k2 ), \
mm512_xor3( k3, k4, k5 ), \
mm512_xor3( k6, k7, \
_mm512_set1_epi64( 0x1BD11BDAA9FC1A22) ) ); \
t2 = t0 ^ t1;
#define TFBIG_ADDKEY_8WAY(w0, w1, w2, w3, w4, w5, w6, w7, k, t, s) \
do { \
w0 = _mm512_add_epi64( w0, SKBI(k,s,0) ); \
w1 = _mm512_add_epi64( w1, SKBI(k,s,1) ); \
w2 = _mm512_add_epi64( w2, SKBI(k,s,2) ); \
w3 = _mm512_add_epi64( w3, SKBI(k,s,3) ); \
w4 = _mm512_add_epi64( w4, SKBI(k,s,4) ); \
w5 = _mm512_add_epi64( w5, _mm512_add_epi64( SKBI(k,s,5), \
m512_const1_64( SKBT(t,s,0) ) ) ); \
_mm512_set1_epi64( SKBT(t,s,0) ) ) ); \
w6 = _mm512_add_epi64( w6, _mm512_add_epi64( SKBI(k,s,6), \
m512_const1_64( SKBT(t,s,1) ) ) ); \
_mm512_set1_epi64( SKBT(t,s,1) ) ) ); \
w7 = _mm512_add_epi64( w7, _mm512_add_epi64( SKBI(k,s,7), \
m512_const1_64( s ) ) ); \
} while (0)
_mm512_set1_epi64( s ) ) );
#define TFBIG_MIX_8WAY(x0, x1, rc) \
do { \
x0 = _mm512_add_epi64( x0, x1 ); \
x1 = _mm512_xor_si512( mm512_rol_64( x1, rc ), x0 ); \
} while (0)
x1 = _mm512_xor_si512( mm512_rol_64( x1, rc ), x0 );
#define TFBIG_MIX8_8WAY(w0, w1, w2, w3, w4, w5, w6, w7, rc0, rc1, rc2, rc3) do { \
#define TFBIG_MIX8_8WAY(w0, w1, w2, w3, w4, w5, w6, w7, rc0, rc1, rc2, rc3) \
TFBIG_MIX_8WAY(w0, w1, rc0); \
TFBIG_MIX_8WAY(w2, w3, rc1); \
TFBIG_MIX_8WAY(w4, w5, rc2); \
TFBIG_MIX_8WAY(w6, w7, rc3); \
} while (0)
TFBIG_MIX_8WAY(w6, w7, rc3);
#define TFBIG_8WAY_4e(s) do { \
#define TFBIG_8WAY_4e(s) \
TFBIG_ADDKEY_8WAY(p0, p1, p2, p3, p4, p5, p6, p7, h, t, s); \
TFBIG_MIX8_8WAY(p0, p1, p2, p3, p4, p5, p6, p7, 46, 36, 19, 37); \
TFBIG_MIX8_8WAY(p2, p1, p4, p7, p6, p5, p0, p3, 33, 27, 14, 42); \
TFBIG_MIX8_8WAY(p4, p1, p6, p3, p0, p5, p2, p7, 17, 49, 36, 39); \
TFBIG_MIX8_8WAY(p6, p1, p0, p7, p2, p5, p4, p3, 44, 9, 54, 56); \
} while (0)
TFBIG_MIX8_8WAY(p6, p1, p0, p7, p2, p5, p4, p3, 44, 9, 54, 56);
#define TFBIG_8WAY_4o(s) do { \
#define TFBIG_8WAY_4o(s) \
TFBIG_ADDKEY_8WAY(p0, p1, p2, p3, p4, p5, p6, p7, h, t, s); \
TFBIG_MIX8_8WAY(p0, p1, p2, p3, p4, p5, p6, p7, 39, 30, 34, 24); \
TFBIG_MIX8_8WAY(p2, p1, p4, p7, p6, p5, p0, p3, 13, 50, 10, 17); \
TFBIG_MIX8_8WAY(p4, p1, p6, p3, p0, p5, p2, p7, 25, 29, 39, 43); \
TFBIG_MIX8_8WAY(p6, p1, p0, p7, p2, p5, p4, p3, 8, 35, 56, 22); \
} while (0)
TFBIG_MIX8_8WAY(p6, p1, p0, p7, p2, p5, p4, p3, 8, 35, 56, 22);
#define UBI_BIG_8WAY(etype, extra) \
do { \
@@ -424,59 +415,48 @@ do { \
#endif // AVX512
#define TFBIG_KINIT_4WAY( k0, k1, k2, k3, k4, k5, k6, k7, k8, t0, t1, t2 ) \
do { \
k8 = _mm256_xor_si256( _mm256_xor_si256( \
_mm256_xor_si256( _mm256_xor_si256( k0, k1 ), \
_mm256_xor_si256( k2, k3 ) ), \
_mm256_xor_si256( _mm256_xor_si256( k4, k5 ), \
_mm256_xor_si256( k6, k7 ) ) ), \
m256_const1_64( 0x1BD11BDAA9FC1A22) ); \
t2 = t0 ^ t1; \
} while (0)
k8 = mm256_xor3( mm256_xor3( k0, k1, k2 ), \
mm256_xor3( k3, k4, k5 ), \
mm256_xor3( k6, k7, \
_mm256_set1_epi64x( 0x1BD11BDAA9FC1A22) ) ); \
t2 = t0 ^ t1;
#define TFBIG_ADDKEY_4WAY(w0, w1, w2, w3, w4, w5, w6, w7, k, t, s) \
do { \
w0 = _mm256_add_epi64( w0, SKBI(k,s,0) ); \
w1 = _mm256_add_epi64( w1, SKBI(k,s,1) ); \
w2 = _mm256_add_epi64( w2, SKBI(k,s,2) ); \
w3 = _mm256_add_epi64( w3, SKBI(k,s,3) ); \
w4 = _mm256_add_epi64( w4, SKBI(k,s,4) ); \
w5 = _mm256_add_epi64( w5, _mm256_add_epi64( SKBI(k,s,5), \
m256_const1_64( SKBT(t,s,0) ) ) ); \
_mm256_set1_epi64x( SKBT(t,s,0) ) ) ); \
w6 = _mm256_add_epi64( w6, _mm256_add_epi64( SKBI(k,s,6), \
m256_const1_64( SKBT(t,s,1) ) ) ); \
_mm256_set1_epi64x( SKBT(t,s,1) ) ) ); \
w7 = _mm256_add_epi64( w7, _mm256_add_epi64( SKBI(k,s,7), \
m256_const1_64( s ) ) ); \
} while (0)
_mm256_set1_epi64x( s ) ) );
#define TFBIG_MIX_4WAY(x0, x1, rc) \
do { \
x0 = _mm256_add_epi64( x0, x1 ); \
x1 = _mm256_xor_si256( mm256_rol_64( x1, rc ), x0 ); \
} while (0)
x1 = _mm256_xor_si256( mm256_rol_64( x1, rc ), x0 );
#define TFBIG_MIX8_4WAY(w0, w1, w2, w3, w4, w5, w6, w7, rc0, rc1, rc2, rc3) do { \
#define TFBIG_MIX8_4WAY(w0, w1, w2, w3, w4, w5, w6, w7, rc0, rc1, rc2, rc3) \
TFBIG_MIX_4WAY(w0, w1, rc0); \
TFBIG_MIX_4WAY(w2, w3, rc1); \
TFBIG_MIX_4WAY(w4, w5, rc2); \
TFBIG_MIX_4WAY(w6, w7, rc3); \
} while (0)
TFBIG_MIX_4WAY(w6, w7, rc3);
#define TFBIG_4WAY_4e(s) do { \
#define TFBIG_4WAY_4e(s) \
TFBIG_ADDKEY_4WAY(p0, p1, p2, p3, p4, p5, p6, p7, h, t, s); \
TFBIG_MIX8_4WAY(p0, p1, p2, p3, p4, p5, p6, p7, 46, 36, 19, 37); \
TFBIG_MIX8_4WAY(p2, p1, p4, p7, p6, p5, p0, p3, 33, 27, 14, 42); \
TFBIG_MIX8_4WAY(p4, p1, p6, p3, p0, p5, p2, p7, 17, 49, 36, 39); \
TFBIG_MIX8_4WAY(p6, p1, p0, p7, p2, p5, p4, p3, 44, 9, 54, 56); \
} while (0)
TFBIG_MIX8_4WAY(p6, p1, p0, p7, p2, p5, p4, p3, 44, 9, 54, 56);
#define TFBIG_4WAY_4o(s) do { \
#define TFBIG_4WAY_4o(s) \
TFBIG_ADDKEY_4WAY(p0, p1, p2, p3, p4, p5, p6, p7, h, t, s); \
TFBIG_MIX8_4WAY(p0, p1, p2, p3, p4, p5, p6, p7, 39, 30, 34, 24); \
TFBIG_MIX8_4WAY(p2, p1, p4, p7, p6, p5, p0, p3, 13, 50, 10, 17); \
TFBIG_MIX8_4WAY(p4, p1, p6, p3, p0, p5, p2, p7, 25, 29, 39, 43); \
TFBIG_MIX8_4WAY(p6, p1, p0, p7, p2, p5, p4, p3, 8, 35, 56, 22); \
} while (0)
TFBIG_MIX8_4WAY(p6, p1, p0, p7, p2, p5, p4, p3, 8, 35, 56, 22);
// scale buf offset by 4
#define UBI_BIG_4WAY(etype, extra) \
@@ -541,28 +521,28 @@ do { \
void skein256_8way_init( skein256_8way_context *sc )
{
sc->h0 = m512_const1_64( 0xCCD044A12FDB3E13 );
sc->h1 = m512_const1_64( 0xE83590301A79A9EB );
sc->h2 = m512_const1_64( 0x55AEA0614F816E6F );
sc->h3 = m512_const1_64( 0x2A2767A4AE9B94DB );
sc->h4 = m512_const1_64( 0xEC06025E74DD7683 );
sc->h5 = m512_const1_64( 0xE7A436CDC4746251 );
sc->h6 = m512_const1_64( 0xC36FBAF9393AD185 );
sc->h7 = m512_const1_64( 0x3EEDBA1833EDFC13 );
sc->h0 = _mm512_set1_epi64( 0xCCD044A12FDB3E13 );
sc->h1 = _mm512_set1_epi64( 0xE83590301A79A9EB );
sc->h2 = _mm512_set1_epi64( 0x55AEA0614F816E6F );
sc->h3 = _mm512_set1_epi64( 0x2A2767A4AE9B94DB );
sc->h4 = _mm512_set1_epi64( 0xEC06025E74DD7683 );
sc->h5 = _mm512_set1_epi64( 0xE7A436CDC4746251 );
sc->h6 = _mm512_set1_epi64( 0xC36FBAF9393AD185 );
sc->h7 = _mm512_set1_epi64( 0x3EEDBA1833EDFC13 );
sc->bcount = 0;
sc->ptr = 0;
}
void skein512_8way_init( skein512_8way_context *sc )
{
sc->h0 = m512_const1_64( 0x4903ADFF749C51CE );
sc->h1 = m512_const1_64( 0x0D95DE399746DF03 );
sc->h2 = m512_const1_64( 0x8FD1934127C79BCE );
sc->h3 = m512_const1_64( 0x9A255629FF352CB1 );
sc->h4 = m512_const1_64( 0x5DB62599DF6CA7B0 );
sc->h5 = m512_const1_64( 0xEABE394CA9D5C3F4 );
sc->h6 = m512_const1_64( 0x991112C71A75B523 );
sc->h7 = m512_const1_64( 0xAE18A40B660FCC33 );
sc->h0 = _mm512_set1_epi64( 0x4903ADFF749C51CE );
sc->h1 = _mm512_set1_epi64( 0x0D95DE399746DF03 );
sc->h2 = _mm512_set1_epi64( 0x8FD1934127C79BCE );
sc->h3 = _mm512_set1_epi64( 0x9A255629FF352CB1 );
sc->h4 = _mm512_set1_epi64( 0x5DB62599DF6CA7B0 );
sc->h5 = _mm512_set1_epi64( 0xEABE394CA9D5C3F4 );
sc->h6 = _mm512_set1_epi64( 0x991112C71A75B523 );
sc->h7 = _mm512_set1_epi64( 0xAE18A40B660FCC33 );
sc->bcount = 0;
sc->ptr = 0;
}
@@ -660,14 +640,14 @@ void skein512_8way_full( skein512_8way_context *sc, void *out, const void *data,
// Init
h0 = m512_const1_64( 0x4903ADFF749C51CE );
h1 = m512_const1_64( 0x0D95DE399746DF03 );
h2 = m512_const1_64( 0x8FD1934127C79BCE );
h3 = m512_const1_64( 0x9A255629FF352CB1 );
h4 = m512_const1_64( 0x5DB62599DF6CA7B0 );
h5 = m512_const1_64( 0xEABE394CA9D5C3F4 );
h6 = m512_const1_64( 0x991112C71A75B523 );
h7 = m512_const1_64( 0xAE18A40B660FCC33 );
h0 = _mm512_set1_epi64( 0x4903ADFF749C51CE );
h1 = _mm512_set1_epi64( 0x0D95DE399746DF03 );
h2 = _mm512_set1_epi64( 0x8FD1934127C79BCE );
h3 = _mm512_set1_epi64( 0x9A255629FF352CB1 );
h4 = _mm512_set1_epi64( 0x5DB62599DF6CA7B0 );
h5 = _mm512_set1_epi64( 0xEABE394CA9D5C3F4 );
h6 = _mm512_set1_epi64( 0x991112C71A75B523 );
h7 = _mm512_set1_epi64( 0xAE18A40B660FCC33 );
// Update
@@ -734,14 +714,14 @@ skein512_8way_prehash64( skein512_8way_context *sc, const void *data )
buf[5] = vdata[5];
buf[6] = vdata[6];
buf[7] = vdata[7];
register __m512i h0 = m512_const1_64( 0x4903ADFF749C51CE );
register __m512i h1 = m512_const1_64( 0x0D95DE399746DF03 );
register __m512i h2 = m512_const1_64( 0x8FD1934127C79BCE );
register __m512i h3 = m512_const1_64( 0x9A255629FF352CB1 );
register __m512i h4 = m512_const1_64( 0x5DB62599DF6CA7B0 );
register __m512i h5 = m512_const1_64( 0xEABE394CA9D5C3F4 );
register __m512i h6 = m512_const1_64( 0x991112C71A75B523 );
register __m512i h7 = m512_const1_64( 0xAE18A40B660FCC33 );
register __m512i h0 = _mm512_set1_epi64( 0x4903ADFF749C51CE );
register __m512i h1 = _mm512_set1_epi64( 0x0D95DE399746DF03 );
register __m512i h2 = _mm512_set1_epi64( 0x8FD1934127C79BCE );
register __m512i h3 = _mm512_set1_epi64( 0x9A255629FF352CB1 );
register __m512i h4 = _mm512_set1_epi64( 0x5DB62599DF6CA7B0 );
register __m512i h5 = _mm512_set1_epi64( 0xEABE394CA9D5C3F4 );
register __m512i h6 = _mm512_set1_epi64( 0x991112C71A75B523 );
register __m512i h7 = _mm512_set1_epi64( 0xAE18A40B660FCC33 );
uint64_t bcount = 1;
UBI_BIG_8WAY( 224, 0 );
@@ -830,28 +810,28 @@ skein512_8way_close(void *cc, void *dst)
void skein256_4way_init( skein256_4way_context *sc )
{
sc->h0 = m256_const1_64( 0xCCD044A12FDB3E13 );
sc->h1 = m256_const1_64( 0xE83590301A79A9EB );
sc->h2 = m256_const1_64( 0x55AEA0614F816E6F );
sc->h3 = m256_const1_64( 0x2A2767A4AE9B94DB );
sc->h4 = m256_const1_64( 0xEC06025E74DD7683 );
sc->h5 = m256_const1_64( 0xE7A436CDC4746251 );
sc->h6 = m256_const1_64( 0xC36FBAF9393AD185 );
sc->h7 = m256_const1_64( 0x3EEDBA1833EDFC13 );
sc->h0 = _mm256_set1_epi64x( 0xCCD044A12FDB3E13 );
sc->h1 = _mm256_set1_epi64x( 0xE83590301A79A9EB );
sc->h2 = _mm256_set1_epi64x( 0x55AEA0614F816E6F );
sc->h3 = _mm256_set1_epi64x( 0x2A2767A4AE9B94DB );
sc->h4 = _mm256_set1_epi64x( 0xEC06025E74DD7683 );
sc->h5 = _mm256_set1_epi64x( 0xE7A436CDC4746251 );
sc->h6 = _mm256_set1_epi64x( 0xC36FBAF9393AD185 );
sc->h7 = _mm256_set1_epi64x( 0x3EEDBA1833EDFC13 );
sc->bcount = 0;
sc->ptr = 0;
}
void skein512_4way_init( skein512_4way_context *sc )
{
sc->h0 = m256_const1_64( 0x4903ADFF749C51CE );
sc->h1 = m256_const1_64( 0x0D95DE399746DF03 );
sc->h2 = m256_const1_64( 0x8FD1934127C79BCE );
sc->h3 = m256_const1_64( 0x9A255629FF352CB1 );
sc->h4 = m256_const1_64( 0x5DB62599DF6CA7B0 );
sc->h5 = m256_const1_64( 0xEABE394CA9D5C3F4 );
sc->h6 = m256_const1_64( 0x991112C71A75B523 );
sc->h7 = m256_const1_64( 0xAE18A40B660FCC33 );
sc->h0 = _mm256_set1_epi64x( 0x4903ADFF749C51CE );
sc->h1 = _mm256_set1_epi64x( 0x0D95DE399746DF03 );
sc->h2 = _mm256_set1_epi64x( 0x8FD1934127C79BCE );
sc->h3 = _mm256_set1_epi64x( 0x9A255629FF352CB1 );
sc->h4 = _mm256_set1_epi64x( 0x5DB62599DF6CA7B0 );
sc->h5 = _mm256_set1_epi64x( 0xEABE394CA9D5C3F4 );
sc->h6 = _mm256_set1_epi64x( 0x991112C71A75B523 );
sc->h7 = _mm256_set1_epi64x( 0xAE18A40B660FCC33 );
sc->bcount = 0;
sc->ptr = 0;
}
@@ -954,14 +934,14 @@ skein512_4way_full( skein512_4way_context *sc, void *out, const void *data,
const int buf_size = 64; // 64 * __m256i
uint64_t bcount = 0;
h0 = m256_const1_64( 0x4903ADFF749C51CE );
h1 = m256_const1_64( 0x0D95DE399746DF03 );
h2 = m256_const1_64( 0x8FD1934127C79BCE );
h3 = m256_const1_64( 0x9A255629FF352CB1 );
h4 = m256_const1_64( 0x5DB62599DF6CA7B0 );
h5 = m256_const1_64( 0xEABE394CA9D5C3F4 );
h6 = m256_const1_64( 0x991112C71A75B523 );
h7 = m256_const1_64( 0xAE18A40B660FCC33 );
h0 = _mm256_set1_epi64x( 0x4903ADFF749C51CE );
h1 = _mm256_set1_epi64x( 0x0D95DE399746DF03 );
h2 = _mm256_set1_epi64x( 0x8FD1934127C79BCE );
h3 = _mm256_set1_epi64x( 0x9A255629FF352CB1 );
h4 = _mm256_set1_epi64x( 0x5DB62599DF6CA7B0 );
h5 = _mm256_set1_epi64x( 0xEABE394CA9D5C3F4 );
h6 = _mm256_set1_epi64x( 0x991112C71A75B523 );
h7 = _mm256_set1_epi64x( 0xAE18A40B660FCC33 );
// Update
@@ -1028,14 +1008,14 @@ skein512_4way_prehash64( skein512_4way_context *sc, const void *data )
buf[5] = vdata[5];
buf[6] = vdata[6];
buf[7] = vdata[7];
register __m256i h0 = m256_const1_64( 0x4903ADFF749C51CE );
register __m256i h1 = m256_const1_64( 0x0D95DE399746DF03 );
register __m256i h2 = m256_const1_64( 0x8FD1934127C79BCE );
register __m256i h3 = m256_const1_64( 0x9A255629FF352CB1 );
register __m256i h4 = m256_const1_64( 0x5DB62599DF6CA7B0 );
register __m256i h5 = m256_const1_64( 0xEABE394CA9D5C3F4 );
register __m256i h6 = m256_const1_64( 0x991112C71A75B523 );
register __m256i h7 = m256_const1_64( 0xAE18A40B660FCC33 );
register __m256i h0 = _mm256_set1_epi64x( 0x4903ADFF749C51CE );
register __m256i h1 = _mm256_set1_epi64x( 0x0D95DE399746DF03 );
register __m256i h2 = _mm256_set1_epi64x( 0x8FD1934127C79BCE );
register __m256i h3 = _mm256_set1_epi64x( 0x9A255629FF352CB1 );
register __m256i h4 = _mm256_set1_epi64x( 0x5DB62599DF6CA7B0 );
register __m256i h5 = _mm256_set1_epi64x( 0xEABE394CA9D5C3F4 );
register __m256i h6 = _mm256_set1_epi64x( 0x991112C71A75B523 );
register __m256i h7 = _mm256_set1_epi64x( 0xAE18A40B660FCC33 );
uint64_t bcount = 1;
UBI_BIG_4WAY( 224, 0 );