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

v3.9.0

Browse Source
This commit is contained in:
Jay D Dee
2019-05-19 13:39:45 -04:00
parent bfd1c002f9
commit e1aead3c76
139 changed files with 10907 additions and 4218 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
algo/x17/hmq1725.c
View File

@@ -24,7 +24,7 @@
#include "algo/echo/aes_ni/hash_api.h"
#endif
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/nist.h"
#include "algo/jh/sse2/jh_sse2_opt64.h"

6
algo/x17/x16r-4way.c
View File

@@ -20,7 +20,7 @@
#include "algo/keccak/keccak-hash-4way.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa-hash-2way.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/simd-hash-2way.h"
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/hamsi/hamsi-hash-4way.h"
@@ -256,11 +256,11 @@ void x16r_4way_hash( void* output, const void* input )
sph_fugue512_close( &ctx.fugue, hash3 );
break;
case SHABAL:
mm_interleave_4x32( vhash, in0, in1, in2, in3, size<<3 );
mm128_interleave_4x32( vhash, in0, in1, in2, in3, size<<3 );
shabal512_4way_init( &ctx.shabal );
shabal512_4way( &ctx.shabal, vhash, size );
shabal512_4way_close( &ctx.shabal, vhash );
mm_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, 512 );
mm128_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case WHIRLPOOL:
sph_whirlpool_init( &ctx.whirlpool );

2
algo/x17/x16r.c
View File

@@ -17,7 +17,7 @@
#include "algo/skein/sph_skein.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/nist.h"
#include "algo/echo/sph_echo.h"
#include "algo/hamsi/sph_hamsi.h"

79
algo/x17/x17-4way.c
View File

@@ -13,7 +13,7 @@
#include "algo/jh/jh-hash-4way.h"
#include "algo/keccak/keccak-hash-4way.h"
#include "algo/luffa/luffa-hash-2way.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/cubehash/cube-hash-2way.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/simd/simd-hash-2way.h"
#include "algo/echo/aes_ni/hash_api.h"
@@ -32,7 +32,7 @@ typedef struct {
jh512_4way_context jh;
keccak512_4way_context keccak;
luffa_2way_context luffa;
cubehashParam cube;
cube_2way_context cube;
sph_shavite512_context shavite;
simd_2way_context simd;
hashState_echo echo;
@@ -55,7 +55,7 @@ void init_x17_4way_ctx()
jh512_4way_init( &x17_4way_ctx.jh );
keccak512_4way_init( &x17_4way_ctx.keccak );
luffa_2way_init( &x17_4way_ctx.luffa, 512 );
cubehashInit( &x17_4way_ctx.cube, 512, 16, 32 );
cube_2way_init( &x17_4way_ctx.cube, 512, 16, 32 );
sph_shavite512_init( &x17_4way_ctx.shavite );
simd_2way_init( &x17_4way_ctx.simd, 512 );
init_echo( &x17_4way_ctx.echo, 512 );
@@ -73,11 +73,11 @@ void x17_4way_hash( void *state, const void *input )
uint64_t hash2[8] __attribute__ ((aligned (64)));
uint64_t hash3[8] __attribute__ ((aligned (64)));
uint64_t vhash[8*4] __attribute__ ((aligned (64)));
uint64_t vhash32[8*4] __attribute__ ((aligned (64)));
uint64_t vhashB[8*4] __attribute__ ((aligned (64)));
x17_4way_ctx_holder ctx;
memcpy( &ctx, &x17_4way_ctx, sizeof(x17_4way_ctx) );
// 1 Blake
// 1 Blake 4 way 64 bit
blake512_4way( &ctx.blake, input, 80 );
blake512_4way_close( &ctx.blake, vhash );
@@ -114,25 +114,22 @@ void x17_4way_hash( void *state, const void *input )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// 7 Luffa
mm256_interleave_2x128( vhash, hash0, hash1, 512 );
// 7 Luffa parallel 2 way
mm256_interleave_2x128( vhash, hash0, hash1, 512 );
mm256_interleave_2x128( vhashB, hash2, hash3, 512 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
mm256_deinterleave_2x128( hash0, hash1, vhash, 512 );
mm256_interleave_2x128( vhash, hash2, hash3, 512 );
luffa_2way_init( &ctx.luffa, 512 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
mm256_deinterleave_2x128( hash2, hash3, vhash, 512 );
luffa_2way_update_close( &ctx.luffa, vhashB, vhashB, 64 );
// 8 Cubehash
cubehashUpdateDigest( &ctx.cube, (byte*)hash0, (const byte*) hash0, 64 );
memcpy( &ctx.cube, &x17_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash1, (const byte*) hash1, 64 );
memcpy( &ctx.cube, &x17_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash2, (const byte*) hash2, 64 );
memcpy( &ctx.cube, &x17_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash3, (const byte*) hash3, 64 );
// 8 Cubehash parallel 2 way
cube_2way_update_close( &ctx.cube, vhash, vhash, 64 );
cube_2way_reinit( &ctx.cube );
cube_2way_update_close( &ctx.cube, vhashB, vhashB, 64 );
// 9 Shavite
mm256_deinterleave_2x128( hash0, hash1, vhash, 512 );
mm256_deinterleave_2x128( hash2, hash3, vhashB, 512 );
// 9 Shavite serial
sph_shavite512( &ctx.shavite, hash0, 64 );
sph_shavite512_close( &ctx.shavite, hash0 );
memcpy( &ctx.shavite, &x17_4way_ctx.shavite,
@@ -148,7 +145,7 @@ void x17_4way_hash( void *state, const void *input )
sph_shavite512( &ctx.shavite, hash3, 64 );
sph_shavite512_close( &ctx.shavite, hash3 );
// 10 Simd
// 10 Simd parallel 2 way 128 bit
mm256_interleave_2x128( vhash, hash0, hash1, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
mm256_deinterleave_2x128( hash0, hash1, vhash, 512 );
@@ -157,7 +154,7 @@ void x17_4way_hash( void *state, const void *input )
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
mm256_deinterleave_2x128( hash2, hash3, vhash, 512 );
// 11 Echo
// 11 Echo serial
update_final_echo( &ctx.echo, (BitSequence *)hash0,
(const BitSequence *) hash0, 512 );
memcpy( &ctx.echo, &x17_4way_ctx.echo, sizeof(hashState_echo) );
@@ -170,13 +167,13 @@ void x17_4way_hash( void *state, const void *input )
update_final_echo( &ctx.echo, (BitSequence *)hash3,
(const BitSequence *) hash3, 512 );
// 12 Hamsi
// 12 Hamsi parallel 4 way 64 bit
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
hamsi512_4way( &ctx.hamsi, vhash, 64 );
hamsi512_4way_close( &ctx.hamsi, vhash );
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// 13 Fugue
// 13 Fugue serial
sph_fugue512( &ctx.fugue, hash0, 64 );
sph_fugue512_close( &ctx.fugue, hash0 );
memcpy( &ctx.fugue, &x17_4way_ctx.fugue, sizeof(sph_fugue512_context) );
@@ -189,11 +186,11 @@ void x17_4way_hash( void *state, const void *input )
sph_fugue512( &ctx.fugue, hash3, 64 );
sph_fugue512_close( &ctx.fugue, hash3 );
// 14 Shabal, parallel 32 bit
mm_interleave_4x32( vhash, hash0, hash1, hash2, hash3, 512 );
// 14 Shabal, parallel 4 way 32 bit SSE
mm128_interleave_4x32( vhash, hash0, hash1, hash2, hash3, 512 );
shabal512_4way( &ctx.shabal, vhash, 64 );
shabal512_4way_close( &ctx.shabal, vhash );
mm_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, 512 );
mm128_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, 512 );
// 15 Whirlpool
sph_whirlpool( &ctx.whirlpool, hash0, 64 );
@@ -217,17 +214,18 @@ void x17_4way_hash( void *state, const void *input )
sha512_4way_close( &ctx.sha512, vhash );
// 17 Haval parallel 32 bit
mm256_reinterleave_4x32( vhash32, vhash, 512 );
haval256_5_4way( &ctx.haval, vhash32, 64 );
haval256_5_4way_close( &ctx.haval, vhash );
mm256_reinterleave_4x32( vhashB, vhash, 512 );
haval256_5_4way( &ctx.haval, vhashB, 64 );
haval256_5_4way_close( &ctx.haval, state );
mm_deinterleave_4x32( state, state+32, state+64, state+96, vhash, 256 );
}
int scanhash_x17_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done )
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t *hash7 = &(hash[7<<2]);
uint32_t lane_hash[8];
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t endiandata[20] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
@@ -261,17 +259,20 @@ int scanhash_x17_4way( int thr_id, struct work *work, uint32_t max_nonce,
be32enc( noncep+6, n+3 );
x17_4way_hash( hash, vdata );
pdata[19] = n;
for ( int i = 0; i < 4; i++ )
if ( ( ( (hash+(i<<3))[7] & mask ) == 0 )
&& fulltest( hash+(i<<3), ptarget ) )
for ( int lane = 0; lane < 4; lane++ )
if ( ( ( hash7[ lane ] & mask ) == 0 ) )
{
pdata[19] = n+i;
nonces[ num_found++ ] = n+i;
work_set_target_ratio( work, hash+(i<<3) );
mm128_extract_lane_4x32( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) )
{
pdata[19] = n + lane;
nonces[ num_found++ ] = n + lane;
work_set_target_ratio( work, lane_hash );
}
}
n += 4;
n += 4;
} while ( ( num_found == 0 ) && ( n < max_nonce )
&& !work_restart[thr_id].restart );
break;

2
algo/x17/x17.c
View File

@@ -22,7 +22,7 @@
#include "algo/haval/sph-haval.h"
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/nist.h"
#include "algo/blake/sse2/blake.c"
#include "algo/bmw/sse2/bmw.c"

36
algo/x17/xevan-4way.c
View File

@@ -14,7 +14,7 @@
#include "algo/skein/skein-hash-4way.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa-hash-2way.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/simd-hash-2way.h"
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/hamsi/hamsi-hash-4way.h"
@@ -201,10 +201,10 @@ void xevan_4way_hash( void *output, const void *input )
sph_fugue512_close( &ctx.fugue, hash3 );
// Parallel 4way 32 bit
mm_interleave_4x32( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm128_interleave_4x32( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
shabal512_4way( &ctx.shabal, vhash, dataLen );
shabal512_4way_close( &ctx.shabal, vhash );
mm_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm128_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
// Serial
sph_whirlpool( &ctx.whirlpool, hash0, dataLen );
@@ -229,7 +229,7 @@ void xevan_4way_hash( void *output, const void *input )
mm256_reinterleave_4x32( vhash32, vhash, dataLen<<3 );
haval256_5_4way( &ctx.haval, vhash32, dataLen );
haval256_5_4way_close( &ctx.haval, vhash );
mm_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm128_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
memset( &vhash[ 4<<2 ], 0, (dataLen-32) << 2 );
@@ -339,10 +339,10 @@ void xevan_4way_hash( void *output, const void *input )
sph_fugue512( &ctx.fugue, hash3, dataLen );
sph_fugue512_close( &ctx.fugue, hash3 );
mm_interleave_4x32( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm128_interleave_4x32( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
shabal512_4way( &ctx.shabal, vhash, dataLen );
shabal512_4way_close( &ctx.shabal, vhash );
mm_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm128_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
sph_whirlpool( &ctx.whirlpool, hash0, dataLen );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
@@ -365,16 +365,15 @@ void xevan_4way_hash( void *output, const void *input )
mm256_reinterleave_4x32( vhash32, vhash, dataLen<<3 );
haval256_5_4way( &ctx.haval, vhash32, dataLen );
haval256_5_4way_close( &ctx.haval, vhash32 );
mm_deinterleave_4x32( output, output+32, output+64, output+96,
vhash32, 256 );
haval256_5_4way_close( &ctx.haval, output );
}
int scanhash_xevan_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done )
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t *hash7 = &(hash[7<<2]);
uint32_t lane_hash[8];
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t _ALIGN(64) endiandata[20];
uint32_t *pdata = work->data;
@@ -405,15 +404,16 @@ int scanhash_xevan_4way( int thr_id, struct work *work, uint32_t max_nonce,
be32enc( noncep+6, n+3 );
xevan_4way_hash( hash, vdata );
pdata[19] = n;
for ( int i = 0; i < 4; i++ )
if ( (hash+(i<<3))[7] <= Htarg && fulltest( hash+(i<<3), ptarget ) )
for ( int lane = 0; lane < 4; lane++ )
if ( hash7[ lane ] <= Htarg )
{
pdata[19] = n+i;
nonces[ num_found++ ] = n+i;
work_set_target_ratio( work, hash+(i<<3) );
mm128_extract_lane_4x32( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) )
{
pdata[19] = n + lane;
nonces[ num_found++ ] = n + lane;
work_set_target_ratio( work, lane_hash );
}
}
n += 4;
} while ( ( num_found == 0 ) && ( n < max_nonce )

2
algo/x17/xevan.c
View File

@@ -19,7 +19,7 @@
#include "algo/sha/sph_sha2.h"
#include "algo/haval/sph-haval.h"
#include "algo/simd/nist.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include <openssl/sha.h>
#ifdef NO_AES_NI
#include "algo/groestl/sph_groestl.h"