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

v3.10.6

Browse Source
This commit is contained in:
Jay D Dee
2019-12-25 01:26:26 -05:00
parent c65b0ff7a6
commit 241bc26767
35 changed files with 3036 additions and 643 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
algo/x16/x16r-gate.c
View File

@@ -52,10 +52,10 @@ bool register_x16r_algo( algo_gate_t* gate )
bool register_x16rv2_algo( algo_gate_t* gate )
{
#if defined (X16R_8WAY)
#if defined (X16RV2_8WAY)
gate->scanhash = (void*)&scanhash_x16rv2_8way;
gate->hash = (void*)&x16rv2_8way_hash;
#elif defined (X16R_4WAY)
#elif defined (X16RV2_4WAY)
gate->scanhash = (void*)&scanhash_x16rv2_4way;
gate->hash = (void*)&x16rv2_4way_hash;
#else
@@ -205,10 +205,10 @@ void veil_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
bool register_x16rt_algo( algo_gate_t* gate )
{
#if defined (X16R_8WAY)
#if defined (X16RT_8WAY)
gate->scanhash = (void*)&scanhash_x16rt_8way;
gate->hash = (void*)&x16rt_8way_hash;
#elif defined (X16R_4WAY)
#elif defined (X16RT_4WAY)
gate->scanhash = (void*)&scanhash_x16rt_4way;
gate->hash = (void*)&x16rt_4way_hash;
#else
@@ -222,10 +222,10 @@ bool register_x16rt_algo( algo_gate_t* gate )
bool register_x16rt_veil_algo( algo_gate_t* gate )
{
#if defined (X16R_8WAY)
#if defined (X16RT_8WAY)
gate->scanhash = (void*)&scanhash_x16rt_8way;
gate->hash = (void*)&x16rt_8way_hash;
#elif defined (X16R_4WAY)
#elif defined (X16RT_4WAY)
gate->scanhash = (void*)&scanhash_x16rt_4way;
gate->hash = (void*)&x16rt_4way_hash;
#else
@@ -258,16 +258,23 @@ bool register_hex_algo( algo_gate_t* gate )
bool register_x21s_algo( algo_gate_t* gate )
{
#if defined (X16R_4WAY)
#if defined (X21S_8WAY)
gate->scanhash = (void*)&scanhash_x21s_8way;
gate->hash = (void*)&x21s_8way_hash;
gate->miner_thread_init = (void*)&x21s_8way_thread_init;
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT;
#elif defined (X21S_4WAY)
gate->scanhash = (void*)&scanhash_x21s_4way;
gate->hash = (void*)&x21s_4way_hash;
gate->miner_thread_init = (void*)&x21s_4way_thread_init;
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | SHA_OPT | AVX512_OPT;
#else
gate->scanhash = (void*)&scanhash_x21s;
gate->hash = (void*)&x21s_hash;
gate->miner_thread_init = (void*)&x21s_thread_init;
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | SHA_OPT | AVX512_OPT;
#endif
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | SHA_OPT;
// gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | SHA_OPT | AVX512_OPT;
x16_r_s_getAlgoString = (void*)&x16s_getAlgoString;
opt_target_factor = 256.0;
return true;

77
algo/x16/x16r-gate.h
View File

@@ -12,6 +12,24 @@
#define X16R_4WAY 1
#endif
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define X16RV2_8WAY 1
#elif defined(__AVX2__) && defined(__AES__)
#define X16RV2_4WAY 1
#endif
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define X16RT_8WAY 1
#elif defined(__AVX2__) && defined(__AES__)
#define X16RT_4WAY 1
#endif
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define X21S_8WAY 1
#elif defined(__AVX2__) && defined(__AES__)
#define X21S_4WAY 1
#endif
enum x16r_Algo {
BLAKE = 0,
BMW,
@@ -46,18 +64,13 @@ bool register_x16rt_algo( algo_gate_t* gate );
bool register_hex__algo( algo_gate_t* gate );
bool register_x21s__algo( algo_gate_t* gate );
// x16r, x16s
#if defined(X16R_8WAY)
void x16r_8way_hash( void *state, const void *input );
int scanhash_x16r_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
void x16rv2_8way_hash( void *state, const void *input );
int scanhash_x16rv2_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
void x16rt_8way_hash( void *state, const void *input );
int scanhash_x16rt_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#elif defined(X16R_4WAY)
@@ -65,31 +78,65 @@ void x16r_4way_hash( void *state, const void *input );
int scanhash_x16r_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
void x16rv2_4way_hash( void *state, const void *input );
int scanhash_x16rv2_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
void x16rt_4way_hash( void *state, const void *input );
int scanhash_x16rt_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#else
void x16r_hash( void *state, const void *input );
int scanhash_x16r( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#endif
// x16Rv2
#if defined(X16RV2_8WAY)
void x16rv2_8way_hash( void *state, const void *input );
int scanhash_x16rv2_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#elif defined(X16RV2_4WAY)
void x16rv2_4way_hash( void *state, const void *input );
int scanhash_x16rv2_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#else
void x16rv2_hash( void *state, const void *input );
int scanhash_x16rv2( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#endif
// x16rt, veil
#if defined(X16RT_8WAY)
void x16rt_8way_hash( void *state, const void *input );
int scanhash_x16rt_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#elif defined(X16RT_4WAY)
void x16rt_4way_hash( void *state, const void *input );
int scanhash_x16rt_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#else
void x16rt_hash( void *state, const void *input );
int scanhash_x16rt( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#endif
#if defined(X16R_4WAY)
// x21s
#if defined(X21S_8WAY)
void x21s_8way_hash( void *state, const void *input );
int scanhash_x21s_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
bool x21s_8way_thread_init();
#elif defined(X21S_4WAY)
void x21s_4way_hash( void *state, const void *input );
int scanhash_x21s_4way( struct work *work, uint32_t max_nonce,

4
algo/x16/x16rt-4way.c
View File

@@ -24,7 +24,7 @@
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
#if defined (X16R_8WAY)
#if defined (X16RT_8WAY)
union _x16rt_8way_context_overlay
{
@@ -407,7 +407,7 @@ int scanhash_x16rt_8way( struct work *work, uint32_t max_nonce,
return 0;
}
#elif defined (X16R_4WAY)
#elif defined (X16RT_4WAY)
union _x16rt_4way_context_overlay
{

7
algo/x16/x16rv2-4way.c
View File

@@ -31,7 +31,7 @@
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
#if defined (X16R_8WAY)
#if defined (X16RV2_8WAY)
union _x16rv2_8way_context_overlay
{
@@ -497,10 +497,7 @@ int scanhash_x16rv2_8way( struct work *work, uint32_t max_nonce,
return 0;
}
#elif defined (X16R_4WAY)
#elif defined (X16RV2_4WAY)
union _x16rv2_4way_context_overlay
{

483
algo/x16/x21s-4way.c
View File

@@ -1,13 +1,10 @@
/**
* x16r algo implementation
* x21s algo implementation
*
* Implementation by tpruvot@github Jan 2018
* Optimized by JayDDee@github Jan 2018
*/
#include "x16r-gate.h"
#if defined (X16R_4WAY)
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
@@ -21,6 +18,7 @@
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa-hash-2way.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/cubehash/cube-hash-2way.h"
#include "algo/simd/simd-hash-2way.h"
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/hamsi/hamsi-hash-4way.h"
@@ -38,6 +36,483 @@
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
#if defined (X21S_8WAY)
static __thread uint64_t* x21s_8way_matrix;
union _x21s_8way_context_overlay
{
blake512_8way_context blake;
bmw512_8way_context bmw;
hashState_groestl groestl;
skein512_8way_context skein;
jh512_8way_context jh;
keccak512_8way_context keccak;
luffa_4way_context luffa;
cube_4way_context cube;
sph_shavite512_context shavite;
simd_4way_context simd;
hashState_echo echo;
hamsi512_8way_context hamsi;
sph_fugue512_context fugue;
shabal512_8way_context shabal;
sph_whirlpool_context whirlpool;
sha512_8way_context sha512;
haval256_5_8way_context haval;
sph_tiger_context tiger;
sph_gost512_context gost;
sha256_8way_context sha256;
} __attribute__ ((aligned (64)));
typedef union _x21s_8way_context_overlay x21s_8way_context_overlay;
void x21s_8way_hash( void* output, const void* input )
{
uint32_t vhash[24*8] __attribute__ ((aligned (128)));
uint32_t hash0[24] __attribute__ ((aligned (64)));
uint32_t hash1[24] __attribute__ ((aligned (64)));
uint32_t hash2[24] __attribute__ ((aligned (64)));
uint32_t hash3[24] __attribute__ ((aligned (64)));
uint32_t hash4[24] __attribute__ ((aligned (64)));
uint32_t hash5[24] __attribute__ ((aligned (64)));
uint32_t hash6[24] __attribute__ ((aligned (64)));
uint32_t hash7[24] __attribute__ ((aligned (64)));
x21s_8way_context_overlay ctx;
void *in0 = (void*) hash0;
void *in1 = (void*) hash1;
void *in2 = (void*) hash2;
void *in3 = (void*) hash3;
void *in4 = (void*) hash4;
void *in5 = (void*) hash5;
void *in6 = (void*) hash6;
void *in7 = (void*) hash7;
int size = 80;
dintrlv_8x64( hash0, hash1, hash2, hash3, hash4, hash5, hash6, hash7,
input, 640 );
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case BLAKE:
blake512_8way_init( &ctx.blake );
if ( i == 0 )
blake512_8way_update( &ctx.blake, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
blake512_8way_update( &ctx.blake, vhash, size );
}
blake512_8way_close( &ctx.blake, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case BMW:
bmw512_8way_init( &ctx.bmw );
if ( i == 0 )
bmw512_8way_update( &ctx.bmw, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
bmw512_8way_update( &ctx.bmw, vhash, size );
}
bmw512_8way_close( &ctx.bmw, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case GROESTL:
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash0,
(const char*)in0, size<<3 );
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash1,
(const char*)in1, size<<3 );
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash2,
(const char*)in2, size<<3 );
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash3,
(const char*)in3, size<<3 );
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash4,
(const char*)in4, size<<3 );
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash5,
(const char*)in5, size<<3 );
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash6,
(const char*)in6, size<<3 );
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash7,
(const char*)in7, size<<3 );
break;
case SKEIN:
skein512_8way_init( &ctx.skein );
if ( i == 0 )
skein512_8way_update( &ctx.skein, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
skein512_8way_update( &ctx.skein, vhash, size );
}
skein512_8way_close( &ctx.skein, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case JH:
jh512_8way_init( &ctx.jh );
if ( i == 0 )
jh512_8way_update( &ctx.jh, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
jh512_8way_update( &ctx.jh, vhash, size );
}
jh512_8way_close( &ctx.jh, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case KECCAK:
keccak512_8way_init( &ctx.keccak );
if ( i == 0 )
keccak512_8way_update( &ctx.keccak, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
keccak512_8way_update( &ctx.keccak, vhash, size );
}
keccak512_8way_close( &ctx.keccak, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case LUFFA:
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
luffa_4way_init( &ctx.luffa, 512 );
luffa_4way_update_close( &ctx.luffa, vhash, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
luffa_4way_init( &ctx.luffa, 512 );
luffa_4way_update_close( &ctx.luffa, vhash, vhash, size);
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
break;
case CUBEHASH:
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
cube_4way_init( &ctx.cube, 512, 16, 32 );
cube_4way_update_close( &ctx.cube, vhash, vhash, 64 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
cube_4way_init( &ctx.cube, 512, 16, 32 );
cube_4way_update_close( &ctx.cube, vhash, vhash, 64 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
break;
case SHAVITE:
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in0, size );
sph_shavite512_close( &ctx.shavite, hash0 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in1, size );
sph_shavite512_close( &ctx.shavite, hash1 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in2, size );
sph_shavite512_close( &ctx.shavite, hash2 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in3, size );
sph_shavite512_close( &ctx.shavite, hash3 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in4, size );
sph_shavite512_close( &ctx.shavite, hash4 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in5, size );
sph_shavite512_close( &ctx.shavite, hash5 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in6, size );
sph_shavite512_close( &ctx.shavite, hash6 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in7, size );
sph_shavite512_close( &ctx.shavite, hash7 );
break;
case SIMD:
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
simd_4way_init( &ctx.simd, 512 );
simd_4way_update_close( &ctx.simd, vhash, vhash, size<<3 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
simd_4way_init( &ctx.simd, 512 );
simd_4way_update_close( &ctx.simd, vhash, vhash, size<<3 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
break;
case ECHO:
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash0,
(const BitSequence*)in0, size<<3 );
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash1,
(const BitSequence*)in1, size<<3 );
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash2,
(const BitSequence*)in2, size<<3 );
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash3,
(const BitSequence*)in3, size<<3 );
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash4,
(const BitSequence*)in4, size<<3 );
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash5,
(const BitSequence*)in5, size<<3 );
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash6,
(const BitSequence*)in6, size<<3 );
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash7,
(const BitSequence*)in7, size<<3 );
break;
case HAMSI:
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
hamsi512_8way_init( &ctx.hamsi );
hamsi512_8way_update( &ctx.hamsi, vhash, size );
hamsi512_8way_close( &ctx.hamsi, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case FUGUE:
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in0, size );
sph_fugue512_close( &ctx.fugue, hash0 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in1, size );
sph_fugue512_close( &ctx.fugue, hash1 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in2, size );
sph_fugue512_close( &ctx.fugue, hash2 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in3, size );
sph_fugue512_close( &ctx.fugue, hash3 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in4, size );
sph_fugue512_close( &ctx.fugue, hash4 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in5, size );
sph_fugue512_close( &ctx.fugue, hash5 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in6, size );
sph_fugue512_close( &ctx.fugue, hash6 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in7, size );
sph_fugue512_close( &ctx.fugue, hash7 );
break;
case SHABAL:
intrlv_8x32( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
shabal512_8way_init( &ctx.shabal );
shabal512_8way_update( &ctx.shabal, vhash, size );
shabal512_8way_close( &ctx.shabal, vhash );
dintrlv_8x32_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case WHIRLPOOL:
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in0, size );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in1, size );
sph_whirlpool_close( &ctx.whirlpool, hash1 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in2, size );
sph_whirlpool_close( &ctx.whirlpool, hash2 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in3, size );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in4, size );
sph_whirlpool_close( &ctx.whirlpool, hash4 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in5, size );
sph_whirlpool_close( &ctx.whirlpool, hash5 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in6, size );
sph_whirlpool_close( &ctx.whirlpool, hash6 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in7, size );
sph_whirlpool_close( &ctx.whirlpool, hash7 );
break;
case SHA_512:
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
sha512_8way_init( &ctx.sha512 );
sha512_8way_update( &ctx.sha512, vhash, size );
sha512_8way_close( &ctx.sha512, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
}
size = 64;
}
intrlv_8x32_512( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7 );
haval256_5_8way_init( &ctx.haval );
haval256_5_8way_update( &ctx.haval, vhash, 64 );
haval256_5_8way_close( &ctx.haval, vhash );
dintrlv_8x32_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
sph_tiger_init( &ctx.tiger );
sph_tiger ( &ctx.tiger, (const void*) hash0, 64 );
sph_tiger_close( &ctx.tiger, (void*) hash0 );
sph_tiger_init( &ctx.tiger );
sph_tiger ( &ctx.tiger, (const void*) hash1, 64 );
sph_tiger_close( &ctx.tiger, (void*) hash1 );
sph_tiger_init( &ctx.tiger );
sph_tiger ( &ctx.tiger, (const void*) hash2, 64 );
sph_tiger_close( &ctx.tiger, (void*) hash2 );
sph_tiger_init( &ctx.tiger );
sph_tiger ( &ctx.tiger, (const void*) hash3, 64 );
sph_tiger_close( &ctx.tiger, (void*) hash3 );
sph_tiger_init( &ctx.tiger );
sph_tiger ( &ctx.tiger, (const void*) hash4, 64 );
sph_tiger_close( &ctx.tiger, (void*) hash4 );
sph_tiger_init( &ctx.tiger );
sph_tiger ( &ctx.tiger, (const void*) hash5, 64 );
sph_tiger_close( &ctx.tiger, (void*) hash5 );
sph_tiger_init( &ctx.tiger );
sph_tiger ( &ctx.tiger, (const void*) hash6, 64 );
sph_tiger_close( &ctx.tiger, (void*) hash6 );
sph_tiger_init( &ctx.tiger );
sph_tiger ( &ctx.tiger, (const void*) hash7, 64 );
sph_tiger_close( &ctx.tiger, (void*) hash7 );
intrlv_2x256( vhash, hash0, hash1, 256 );
LYRA2REV2_2WAY( x21s_8way_matrix, vhash, 32, vhash, 32, 1, 4, 4 );
dintrlv_2x256( hash0, hash1, vhash, 256 );
intrlv_2x256( vhash, hash2, hash3, 256 );
LYRA2REV2_2WAY( x21s_8way_matrix, vhash, 32, vhash, 32, 1, 4, 4 );
dintrlv_2x256( hash2, hash3, vhash, 256 );
intrlv_2x256( vhash, hash4, hash5, 256 );
LYRA2REV2_2WAY( x21s_8way_matrix, vhash, 32, vhash, 32, 1, 4, 4 );
dintrlv_2x256( hash4, hash5, vhash, 256 );
intrlv_2x256( vhash, hash6, hash7, 256 );
LYRA2REV2_2WAY( x21s_8way_matrix, vhash, 32, vhash, 32, 1, 4, 4 );
dintrlv_2x256( hash6, hash7, vhash, 256 );
sph_gost512_init( &ctx.gost );
sph_gost512 ( &ctx.gost, (const void*) hash0, 64 );
sph_gost512_close( &ctx.gost, (void*) hash0 );
sph_gost512_init( &ctx.gost );
sph_gost512 ( &ctx.gost, (const void*) hash1, 64 );
sph_gost512_close( &ctx.gost, (void*) hash1 );
sph_gost512_init( &ctx.gost );
sph_gost512 ( &ctx.gost, (const void*) hash2, 64 );
sph_gost512_close( &ctx.gost, (void*) hash2 );
sph_gost512_init( &ctx.gost );
sph_gost512 ( &ctx.gost, (const void*) hash3, 64 );
sph_gost512_close( &ctx.gost, (void*) hash3 );
sph_gost512_init( &ctx.gost );
sph_gost512 ( &ctx.gost, (const void*) hash4, 64 );
sph_gost512_close( &ctx.gost, (void*) hash4 );
sph_gost512_init( &ctx.gost );
sph_gost512 ( &ctx.gost, (const void*) hash5, 64 );
sph_gost512_close( &ctx.gost, (void*) hash5 );
sph_gost512_init( &ctx.gost );
sph_gost512 ( &ctx.gost, (const void*) hash6, 64 );
sph_gost512_close( &ctx.gost, (void*) hash6 );
sph_gost512_init( &ctx.gost );
sph_gost512 ( &ctx.gost, (const void*) hash7, 64 );
sph_gost512_close( &ctx.gost, (void*) hash7 );
intrlv_8x32_512( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7 );
sha256_8way_init( &ctx.sha256 );
sha256_8way_update( &ctx.sha256, vhash, 64 );
sha256_8way_close( &ctx.sha256, output );
}
int scanhash_x21s_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr)
{
uint32_t hash[8*16] __attribute__ ((aligned (128)));
uint32_t vdata[24*8] __attribute__ ((aligned (64)));
uint32_t *hash7 = &hash[7<<3];
uint32_t lane_hash[8] __attribute__ ((aligned (64)));
uint32_t bedata1[2] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t Htarg = ptarget[7];
const uint32_t first_nonce = pdata[19];
uint32_t n = first_nonce;
const uint32_t last_nonce = max_nonce - 16;
int thr_id = mythr->id;
__m512i *noncev = (__m512i*)vdata + 9; // aligned
volatile uint8_t *restart = &(work_restart[thr_id].restart);
if ( opt_benchmark )
ptarget[7] = 0x0cff;
mm512_bswap32_intrlv80_8x64( vdata, pdata );
bedata1[0] = bswap_32( pdata[1] );
bedata1[1] = bswap_32( pdata[2] );
uint32_t ntime = bswap_32( pdata[17] );
if ( s_ntime != ntime )
{
x16_r_s_getAlgoString( (const uint8_t*)bedata1, hashOrder );
s_ntime = ntime;
if ( opt_debug && !thr_id )
applog( LOG_DEBUG, "hash order %s (%08x)", hashOrder, ntime );
}
do
{
*noncev = mm512_intrlv_blend_32( mm512_bswap_32(
_mm512_set_epi32( n+7, 0, n+6, 0, n+5, 0, n+4, 0,
n+3, 0, n+2, 0, n+1, 0, n, 0 ) ), *noncev );
x21s_8way_hash( hash, vdata );
pdata[19] = n;
for ( int lane = 0; lane < 8; lane++ )
if ( unlikely( hash7[lane] <= Htarg ) )
{
extr_lane_8x32( lane_hash, hash, lane, 256 );
if ( likely( fulltest( lane_hash, ptarget ) && !opt_benchmark ) )
{
pdata[19] = n + lane;
submit_lane_solution( work, lane_hash, mythr, lane );
}
}
n += 8;
} while ( ( n < last_nonce ) && !(*restart) );
*hashes_done = n - first_nonce;
return 0;
}
bool x21s_8way_thread_init()
{
const int64_t ROW_LEN_INT64 = BLOCK_LEN_INT64 * 4; // nCols
const int64_t ROW_LEN_BYTES = ROW_LEN_INT64 * 8;
const int size = (int64_t)ROW_LEN_BYTES * 4; // nRows;
x21s_8way_matrix = _mm_malloc( 2 * size, 64 );
return x21s_8way_matrix;
}
#elif defined (X21S_4WAY)
static __thread uint64_t* x21s_4way_matrix;
union _x21s_4way_context_overlay