Files
cpuminer-opt-gpu/algo/blake/blake-4way.c
Jay D Dee e1aead3c76 v3.9.0
2019-05-19 13:39:45 -04:00

145 lines
4.1 KiB
C

#include "blake-gate.h"
#include "blake-hash-4way.h"
#include <string.h>
#include <stdint.h>
#include <memory.h>
#if defined (BLAKE_4WAY)
blake256r14_4way_context blake_4w_ctx;
void blakehash_4way(void *state, const void *input)
{
uint32_t vhash[8*4] __attribute__ ((aligned (64)));
blake256r14_4way_context ctx;
memcpy( &ctx, &blake_4w_ctx, sizeof ctx );
blake256r14_4way( &ctx, input + (64<<2), 16 );
blake256r14_4way_close( &ctx, vhash );
mm128_deinterleave_4x32( state, state+32, state+64, state+96, vhash, 256 );
}
int scanhash_blake_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done )
{
uint32_t vdata[20*4] __attribute__ ((aligned (64)));
uint32_t hash[8*4] __attribute__ ((aligned (32)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
uint32_t HTarget = ptarget[7];
uint32_t _ALIGN(32) edata[20];
uint32_t n = first_nonce;
uint32_t *nonces = work->nonces;
int num_found = 0;
if (opt_benchmark)
HTarget = 0x7f;
// we need big endian data...
swab32_array( edata, pdata, 20 );
mm128_interleave_4x32( vdata, edata, edata, edata, edata, 640 );
blake256r14_4way_init( &blake_4w_ctx );
blake256r14_4way( &blake_4w_ctx, vdata, 64 );
uint32_t *noncep = vdata + 76; // 19*4
do {
be32enc( noncep, n );
be32enc( noncep +1, n+1 );
be32enc( noncep +2, n+2 );
be32enc( noncep +3, n+3 );
blakehash_4way( hash, vdata );
for ( int i = 0; i < 4; i++ )
if ( (hash+(i<<3))[7] <= HTarget && fulltest( hash+(i<<3), ptarget ) )
{
pdata[19] = n+i;
nonces[ num_found++ ] = n+i;
work_set_target_ratio( work, hash+(i<<3) );
}
n += 4;
} while ( (num_found == 0) && (n < max_nonce)
&& !work_restart[thr_id].restart );
*hashes_done = n - first_nonce + 1;
return num_found;
}
#endif
#if defined(BLAKE_8WAY)
blake256r14_8way_context blake_8w_ctx;
void blakehash_8way( void *state, const void *input )
{
uint32_t vhash[8*8] __attribute__ ((aligned (64)));
blake256r14_8way_context ctx;
memcpy( &ctx, &blake_8w_ctx, sizeof ctx );
blake256r14_8way( &ctx, input + (64<<3), 16 );
blake256r14_8way_close( &ctx, vhash );
mm256_deinterleave_8x32( state, state+ 32, state+ 64, state+ 96,
state+128, state+160, state+192, state+224,
vhash, 256 );
}
int scanhash_blake_8way( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done )
{
uint32_t vdata[20*8] __attribute__ ((aligned (64)));
uint32_t hash[8*8] __attribute__ ((aligned (32)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
uint32_t HTarget = ptarget[7];
uint32_t _ALIGN(32) edata[20];
uint32_t n = first_nonce;
uint32_t *nonces = work->nonces;
int num_found = 0;
if (opt_benchmark)
HTarget = 0x7f;
// we need big endian data...
swab32_array( edata, pdata, 20 );
mm256_interleave_8x32( vdata, edata, edata, edata, edata,
edata, edata, edata, edata, 640 );
blake256r14_8way_init( &blake_8w_ctx );
blake256r14_8way( &blake_8w_ctx, vdata, 64 );
uint32_t *noncep = vdata + 152; // 19*8
do {
be32enc( noncep, n );
be32enc( noncep +1, n+1 );
be32enc( noncep +2, n+2 );
be32enc( noncep +3, n+3 );
be32enc( noncep +4, n+4 );
be32enc( noncep +5, n+5 );
be32enc( noncep +6, n+6 );
be32enc( noncep +7, n+7 );
pdata[19] = n;
blakehash_8way( hash, vdata );
for ( int i = 0; i < 8; i++ )
if ( (hash+i)[7] <= HTarget && fulltest( hash+i, ptarget ) )
{
pdata[19] = n+i;
num_found++;
nonces[i] = n+i;
work_set_target_ratio( work, hash+1 );
}
n += 8;
} while ( (num_found == 0) && (n < max_nonce)
&& !work_restart[thr_id].restart );
*hashes_done = n - first_nonce + 1;
return num_found;
}
#endif