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

v3.12.6.1

Browse Source
This commit is contained in:
Jay D Dee
2020-03-07 14:11:06 -05:00
parent c0aadbcc99
commit 6e8b8ed34f
29 changed files with 294 additions and 168 deletions
Download Patch File Download Diff File Expand all files Collapse all files

33
algo/x22/x22i-4way.c
View File

@@ -62,7 +62,7 @@ union _x22i_8way_ctx_overlay
};
typedef union _x22i_8way_ctx_overlay x22i_8way_ctx_overlay;
void x22i_8way_hash( void *output, const void *input )
int x22i_8way_hash( void *output, const void *input, int thrid )
{
uint64_t vhash[8*8] __attribute__ ((aligned (128)));
uint64_t vhashA[8*8] __attribute__ ((aligned (64)));
@@ -129,6 +129,8 @@ void x22i_8way_hash( void *output, const void *input )
keccak512_8way_update( &ctx.keccak, vhash, 64 );
keccak512_8way_close( &ctx.keccak, vhash );
if ( work_restart[thrid].restart ) return 0;
rintrlv_8x64_4x128( vhashA, vhashB, vhash, 512 );
luffa512_4way_full( &ctx.luffa, vhashA, vhashA, 64 );
@@ -214,6 +216,8 @@ void x22i_8way_hash( void *output, const void *input )
#endif
if ( work_restart[thrid].restart ) return 0;
hamsi512_8way_init( &ctx.hamsi );
hamsi512_8way_update( &ctx.hamsi, vhash, 64 );
hamsi512_8way_close( &ctx.hamsi, vhash );
@@ -346,6 +350,8 @@ void x22i_8way_hash( void *output, const void *input )
sph_tiger (&ctx.tiger, (const void*) hash7, 64);
sph_tiger_close(&ctx.tiger, (void*) hashA7);
if ( work_restart[thrid].restart ) return 0;
memset( hash0, 0, 64 );
memset( hash1, 0, 64 );
memset( hash2, 0, 64 );
@@ -399,6 +405,8 @@ void x22i_8way_hash( void *output, const void *input )
sha256_8way_init( &ctx.sha256 );
sha256_8way_update( &ctx.sha256, vhash, 64 );
sha256_8way_close( &ctx.sha256, output );
return 1;
}
int scanhash_x22i_8way( struct work *work, uint32_t max_nonce,
@@ -428,8 +436,7 @@ int scanhash_x22i_8way( struct work *work, uint32_t max_nonce,
n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
do
{
x22i_8way_hash( hash, vdata );
if ( x22i_8way_hash( hash, vdata, thr_id ) )
for ( int lane = 0; lane < 8; lane++ )
if ( unlikely( ( hashd7[ lane ] <= targ32 ) && !bench ) )
{
@@ -437,7 +444,7 @@ int scanhash_x22i_8way( struct work *work, uint32_t max_nonce,
if ( likely( valid_hash( lane_hash, ptarget ) ) )
{
pdata[19] = bswap_32( n + lane );
submit_lane_solution( work, lane_hash, mythr, lane );
submit_solution( work, lane_hash, mythr );
}
}
*noncev = _mm512_add_epi32( *noncev,
@@ -524,7 +531,7 @@ union _x22i_4way_ctx_overlay
};
typedef union _x22i_4way_ctx_overlay x22i_ctx_overlay;
void x22i_4way_hash( void *output, const void *input )
int x22i_4way_hash( void *output, const void *input, int thrid )
{
uint64_t hash0[8*4] __attribute__ ((aligned (64)));
uint64_t hash1[8*4] __attribute__ ((aligned (64)));
@@ -563,6 +570,8 @@ void x22i_4way_hash( void *output, const void *input )
keccak512_4way_update( &ctx.keccak, vhash, 64 );
keccak512_4way_close( &ctx.keccak, vhash );
if ( work_restart[thrid].restart ) return false;
rintrlv_4x64_2x128( vhashA, vhashB, vhash, 512 );
luffa512_2way_full( &ctx.luffa, vhashA, vhashA, 64 );
@@ -591,6 +600,8 @@ void x22i_4way_hash( void *output, const void *input )
intrlv_4x64_512( vhash, hash0, hash1, hash2, hash3 );
if ( work_restart[thrid].restart ) return false;
hamsi512_4way_init( &ctx.hamsi );
hamsi512_4way_update( &ctx.hamsi, vhash, 64 );
hamsi512_4way_close( &ctx.hamsi, vhash );
@@ -636,6 +647,8 @@ void x22i_4way_hash( void *output, const void *input )
sha512_4way_close( &ctx.sha512, vhash );
dintrlv_4x64_512( &hash0[24], &hash1[24], &hash2[24], &hash3[24], vhash );
if ( work_restart[thrid].restart ) return false;
ComputeSingleSWIFFTX((unsigned char*)hash0, (unsigned char*)hashA0);
ComputeSingleSWIFFTX((unsigned char*)hash1, (unsigned char*)hashA1);
ComputeSingleSWIFFTX((unsigned char*)hash2, (unsigned char*)hashA2);
@@ -668,6 +681,8 @@ void x22i_4way_hash( void *output, const void *input )
sph_tiger (&ctx.tiger, (const void*) hash3, 64);
sph_tiger_close(&ctx.tiger, (void*) hashA3);
if ( work_restart[thrid].restart ) return false;
memset( hash0, 0, 64 );
memset( hash1, 0, 64 );
memset( hash2, 0, 64 );
@@ -700,8 +715,9 @@ void x22i_4way_hash( void *output, const void *input )
sha256_4way_init( &ctx.sha256 );
sha256_4way_update( &ctx.sha256, vhash, 64 );
sha256_4way_close( &ctx.sha256, output );
}
return 1;
}
int scanhash_x22i_4way( struct work* work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
@@ -729,8 +745,7 @@ int scanhash_x22i_4way( struct work* work, uint32_t max_nonce,
_mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
do
{
x22i_4way_hash( hash, vdata );
if ( x22i_4way_hash( hash, vdata, thr_id ) )
for ( int lane = 0; lane < 4; lane++ )
if ( unlikely( hashd7[ lane ] <= targ32 && !bench ) )
{
@@ -738,7 +753,7 @@ int scanhash_x22i_4way( struct work* work, uint32_t max_nonce,
if ( valid_hash( lane_hash, ptarget ) )
{
pdata[19] = bswap_32( n + lane );
submit_lane_solution( work, lane_hash, mythr, lane );
submit_solution( work, lane_hash, mythr );
}
}
*noncev = _mm256_add_epi32( *noncev,

12
algo/x22/x22i-gate.h
View File

@@ -16,19 +16,19 @@ bool register_x22i_algo( algo_gate_t* gate );
#if defined(X22I_8WAY)
void x22i_8way_hash( void *state, const void *input );
int x22i_8way_hash( void *state, const void *input, int thrid );
int scanhash_x22i_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#elif defined(X22I_4WAY)
void x22i_4way_hash( void *state, const void *input );
int x22i_4way_hash( void *state, const void *input, int thrid );
int scanhash_x22i_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#else
void x22i_hash( void *state, const void *input );
int x22i_hash( void *state, const void *input, int thrid );
int scanhash_x22i( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
@@ -44,19 +44,19 @@ bool register_x25i_algo( algo_gate_t* gate );
#if defined(X25X_8WAY)
void x25x_8way_hash( void *state, const void *input );
int x25x_8way_hash( void *state, const void *input, int thrid );
int scanhash_x25x_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#elif defined(X25X_4WAY)
void x25x_4way_hash( void *state, const void *input );
int x25x_4way_hash( void *state, const void *input, int thrid );
int scanhash_x25x_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#else
void x25x_hash( void *state, const void *input );
int x25x_hash( void *state, const void *input, int thrif );
int scanhash_x25x( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );

12
algo/x22/x22i.c
View File

@@ -59,7 +59,7 @@ union _x22i_context_overlay
};
typedef union _x22i_context_overlay x22i_context_overlay;
void x22i_hash( void *output, const void *input )
int x22i_hash( void *output, const void *input, int thrid )
{
unsigned char hash[64 * 4] __attribute__((aligned(64))) = {0};
unsigned char hash2[65] __attribute__((aligned(64))) = {0};
@@ -95,6 +95,8 @@ void x22i_hash( void *output, const void *input )
sph_keccak512(&ctx.keccak, (const void*) hash, 64);
sph_keccak512_close(&ctx.keccak, hash);
if ( work_restart[thrid].restart ) return 0;
init_luffa( &ctx.luffa, 512 );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
(const BitSequence*)hash, 64 );
@@ -121,6 +123,8 @@ void x22i_hash( void *output, const void *input )
sph_echo512_close( &ctx.echo, hash );
#endif
if ( work_restart[thrid].restart ) return 0;
sph_hamsi512_init(&ctx.hamsi);
sph_hamsi512(&ctx.hamsi, (const void*) hash, 64);
sph_hamsi512_close(&ctx.hamsi, hash);
@@ -143,6 +147,8 @@ void x22i_hash( void *output, const void *input )
ComputeSingleSWIFFTX((unsigned char*)hash, (unsigned char*)hash2);
if ( work_restart[thrid].restart ) return 0;
memset(hash, 0, 64);
sph_haval256_5_init(&ctx.haval);
sph_haval256_5(&ctx.haval,(const void*) hash2, 64);
@@ -165,6 +171,8 @@ void x22i_hash( void *output, const void *input )
SHA256_Final( (unsigned char*) hash, &ctx.sha256 );
memcpy(output, hash, 32);
return 1;
}
int scanhash_x22i( struct work *work, uint32_t max_nonce,
@@ -188,7 +196,7 @@ int scanhash_x22i( struct work *work, uint32_t max_nonce,
do
{
edata[19] = n;
x22i_hash( hash64, edata );
if ( x22i_hash( hash64, edata, thr_id ) );
if ( unlikely( valid_hash( hash64, ptarget ) && !bench ) )
{
pdata[19] = bswap_32( n );

36
algo/x22/x25x-4way.c
View File

@@ -94,7 +94,7 @@ union _x25x_8way_ctx_overlay
};
typedef union _x25x_8way_ctx_overlay x25x_8way_ctx_overlay;
void x25x_8way_hash( void *output, const void *input )
int x25x_8way_hash( void *output, const void *input, int thrid )
{
uint64_t vhash[8*8] __attribute__ ((aligned (128)));
unsigned char hash0[25][64] __attribute__((aligned(64))) = {0};
@@ -179,13 +179,15 @@ void x25x_8way_hash( void *output, const void *input )
jh512_8way_close( &ctx.jh, vhash );
dintrlv_8x64_512( hash0[4], hash1[4], hash2[4], hash3[4],
hash4[4], hash5[4], hash6[4], hash7[4], vhash );
keccak512_8way_init( &ctx.keccak );
keccak512_8way_update( &ctx.keccak, vhash, 64 );
keccak512_8way_close( &ctx.keccak, vhash );
dintrlv_8x64_512( hash0[5], hash1[5], hash2[5], hash3[5],
hash4[5], hash5[5], hash6[5], hash7[5], vhash );
if ( work_restart[thrid].restart ) return 0;
rintrlv_8x64_4x128( vhashA, vhashB, vhash, 512 );
luffa_4way_init( &ctx.luffa, 512 );
@@ -261,6 +263,7 @@ void x25x_8way_hash( void *output, const void *input )
intrlv_8x64_512( vhash, hash0[10], hash1[10], hash2[10], hash3[10],
hash4[10], hash5[10], hash6[10], hash7[10] );
#else
init_echo( &ctx.echo, 512 );
@@ -292,6 +295,8 @@ void x25x_8way_hash( void *output, const void *input )
#endif
if ( work_restart[thrid].restart ) return 0;
hamsi512_8way_init( &ctx.hamsi );
hamsi512_8way_update( &ctx.hamsi, vhash, 64 );
hamsi512_8way_close( &ctx.hamsi, vhash );
@@ -407,6 +412,8 @@ void x25x_8way_hash( void *output, const void *input )
sph_tiger (&ctx.tiger, (const void*) hash7[17], 64);
sph_tiger_close(&ctx.tiger, (void*) hash7[18]);
if ( work_restart[thrid].restart ) return 0;
intrlv_2x256( vhash, hash0[18], hash1[18], 256 );
LYRA2X_2WAY( vhash, 32, vhash, 32, 1, 4, 4 );
dintrlv_2x256( hash0[19], hash1[19], vhash, 256 );
@@ -468,6 +475,8 @@ void x25x_8way_hash( void *output, const void *input )
laneHash(512, (const BitSequence*)hash6[22], 512, (BitSequence*)hash6[23]);
laneHash(512, (const BitSequence*)hash7[22], 512, (BitSequence*)hash7[23]);
if ( work_restart[thrid].restart ) return 0;
x25x_shuffle( hash0 );
x25x_shuffle( hash1 );
x25x_shuffle( hash2 );
@@ -528,6 +537,8 @@ void x25x_8way_hash( void *output, const void *input )
blake2s_8way_init( &ctx.blake2s, 32 );
blake2s_8way_full_blocks( &ctx.blake2s, output, vhashX, 64*24 );
return 1;
}
int scanhash_x25x_8way( struct work *work, uint32_t max_nonce,
@@ -557,7 +568,7 @@ int scanhash_x25x_8way( struct work *work, uint32_t max_nonce,
n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
do
{
x25x_8way_hash( hash, vdata );
if ( x25x_8way_hash( hash, vdata, thr_id ) );
for ( int lane = 0; lane < 8; lane++ )
if ( unlikely( ( hashd7[ lane ] <= targ32 ) && !bench ) )
@@ -566,7 +577,7 @@ int scanhash_x25x_8way( struct work *work, uint32_t max_nonce,
if ( likely( valid_hash( lane_hash, ptarget ) ) )
{
pdata[19] = bswap_32( n + lane );
submit_lane_solution( work, lane_hash, mythr, lane );
submit_solution( work, lane_hash, mythr );
}
}
*noncev = _mm512_add_epi32( *noncev,
@@ -654,7 +665,7 @@ union _x25x_4way_ctx_overlay
};
typedef union _x25x_4way_ctx_overlay x25x_4way_ctx_overlay;
void x25x_4way_hash( void *output, const void *input )
int x25x_4way_hash( void *output, const void *input, int thrid )
{
uint64_t vhash[8*4] __attribute__ ((aligned (128)));
unsigned char hash0[25][64] __attribute__((aligned(64))) = {0};
@@ -686,6 +697,8 @@ void x25x_4way_hash( void *output, const void *input )
jh512_4way_close( &ctx.jh, vhash );
dintrlv_4x64_512( hash0[4], hash1[4], hash2[4], hash3[4], vhash );
if ( work_restart[thrid].restart ) return 0;
keccak512_4way_init( &ctx.keccak );
keccak512_4way_update( &ctx.keccak, vhash, 64 );
keccak512_4way_close( &ctx.keccak, vhash );
@@ -738,6 +751,8 @@ void x25x_4way_hash( void *output, const void *input )
intrlv_4x64_512( vhash, hash0[10], hash1[10], hash2[10], hash3[10] );
if ( work_restart[thrid].restart ) return 0;
hamsi512_4way_init( &ctx.hamsi );
hamsi512_4way_update( &ctx.hamsi, vhash, 64 );
hamsi512_4way_close( &ctx.hamsi, vhash );
@@ -819,6 +834,8 @@ void x25x_4way_hash( void *output, const void *input )
LYRA2RE( (void*)hash3[19], 32, (const void*)hash3[18], 32,
(const void*)hash3[18], 32, 1, 4, 4 );
if ( work_restart[thrid].restart ) return 0;
sph_gost512_init(&ctx.gost);
sph_gost512 (&ctx.gost, (const void*) hash0[19], 64);
sph_gost512_close(&ctx.gost, (void*) hash0[20]);
@@ -850,6 +867,8 @@ void x25x_4way_hash( void *output, const void *input )
laneHash(512, (const BitSequence*)hash2[22], 512, (BitSequence*)hash2[23]);
laneHash(512, (const BitSequence*)hash3[22], 512, (BitSequence*)hash3[23]);
if ( work_restart[thrid].restart ) return 0;
x25x_shuffle( hash0 );
x25x_shuffle( hash1 );
x25x_shuffle( hash2 );
@@ -882,6 +901,8 @@ void x25x_4way_hash( void *output, const void *input )
blake2s_4way_init( &ctx.blake2s, 32 );
blake2s_4way_full_blocks( &ctx.blake2s, output, vhashX, 64*24 );
return 1;
}
int scanhash_x25x_4way( struct work* work, uint32_t max_nonce,
@@ -910,8 +931,7 @@ int scanhash_x25x_4way( struct work* work, uint32_t max_nonce,
_mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
do
{
x25x_4way_hash( hash, vdata );
if ( x25x_4way_hash( hash, vdata, thr_id ) )
for ( int lane = 0; lane < 4; lane++ )
if ( unlikely( hashd7[ lane ] <= targ32 && !bench ) )
{
@@ -919,7 +939,7 @@ int scanhash_x25x_4way( struct work* work, uint32_t max_nonce,
if ( valid_hash( lane_hash, ptarget ) )
{
pdata[19] = bswap_32( n + lane );
submit_lane_solution( work, lane_hash, mythr, lane );
submit_solution( work, lane_hash, mythr );
}
}
*noncev = _mm256_add_epi32( *noncev,

14
algo/x22/x25x.c
View File

@@ -64,7 +64,7 @@ union _x25x_context_overlay
};
typedef union _x25x_context_overlay x25x_context_overlay;
void x25x_hash( void *output, const void *input )
int x25x_hash( void *output, const void *input, int thrid )
{
unsigned char hash[25][64] __attribute__((aligned(64))) = {0};
x25x_context_overlay ctx;
@@ -99,6 +99,8 @@ void x25x_hash( void *output, const void *input )
sph_keccak512(&ctx.keccak, (const void*) &hash[4], 64);
sph_keccak512_close(&ctx.keccak, &hash[5]);
if ( work_restart[thrid].restart ) return 0;
init_luffa( &ctx.luffa, 512 );
update_and_final_luffa( &ctx.luffa, (BitSequence*)&hash[6],
(const BitSequence*)&hash[5], 64 );
@@ -125,7 +127,9 @@ void x25x_hash( void *output, const void *input )
sph_echo512_close( &ctx.echo, &hash[10] );
#endif
sph_hamsi512_init(&ctx.hamsi);
if ( work_restart[thrid].restart ) return 0;
sph_hamsi512_init(&ctx.hamsi);
sph_hamsi512(&ctx.hamsi, (const void*) &hash[10], 64);
sph_hamsi512_close(&ctx.hamsi, &hash[11]);
@@ -151,6 +155,8 @@ void x25x_hash( void *output, const void *input )
sph_haval256_5(&ctx.haval,(const void*) &hash[16], 64);
sph_haval256_5_close(&ctx.haval,&hash[17]);
if ( work_restart[thrid].restart ) return 0;
sph_tiger_init(&ctx.tiger);
sph_tiger (&ctx.tiger, (const void*) &hash[17], 64);
sph_tiger_close(&ctx.tiger, (void*) &hash[18]);
@@ -199,6 +205,8 @@ void x25x_hash( void *output, const void *input )
blake2s_simple( (uint8_t*)&hash[24], (const void*)(&hash[0]), 64 * 24 );
memcpy(output, &hash[24], 32);
return 1;
}
int scanhash_x25x( struct work *work, uint32_t max_nonce,
@@ -222,7 +230,7 @@ int scanhash_x25x( struct work *work, uint32_t max_nonce,
do
{
edata[19] = n;
x25x_hash( hash64, edata );
if ( x25x_hash( hash64, edata, thr_id ) );
if ( unlikely( valid_hash( hash64, ptarget ) && !bench ) )
{
pdata[19] = bswap_32( n );