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

v3.12.2

Browse Source
This commit is contained in:
Jay D Dee
2020-02-09 13:30:40 -05:00
parent dc2f8d81d3
commit 3da2b958cf
39 changed files with 1496 additions and 1518 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
algo/lyra2/lyra2-gate.c
View File

@@ -119,7 +119,7 @@ bool register_lyra2rev2_algo( algo_gate_t* gate )
gate->scanhash = (void*)&scanhash_lyra2rev2;
gate->hash = (void*)&lyra2rev2_hash;
#endif
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT;
gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT;
gate->miner_thread_init = (void*)&lyra2rev2_thread_init;
opt_target_factor = 256.0;
return true;

220
algo/lyra2/lyra2rev2-4way.c
View File

@@ -7,27 +7,8 @@
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/cubehash/cube-hash-2way.h"
#if defined (LYRA2REV2_8WAY)
typedef struct {
blake256_8way_context blake;
keccak256_8way_context keccak;
cube_4way_context cube;
skein256_8way_context skein;
bmw256_8way_context bmw;
} lyra2v2_8way_ctx_holder __attribute__ ((aligned (64)));
static lyra2v2_8way_ctx_holder l2v2_8way_ctx;
bool init_lyra2rev2_8way_ctx()
{
keccak256_8way_init( &l2v2_8way_ctx.keccak );
cube_4way_init( &l2v2_8way_ctx.cube, 256, 16, 32 );
skein256_8way_init( &l2v2_8way_ctx.skein );
bmw256_8way_init( &l2v2_8way_ctx.bmw );
return true;
}
#if 0
void lyra2rev2_8way_hash( void *state, const void *input )
{
uint32_t vhash[8*8] __attribute__ ((aligned (128)));
@@ -52,14 +33,24 @@ void lyra2rev2_8way_hash( void *state, const void *input )
keccak256_8way_update( &ctx.keccak, vhashA, 32 );
keccak256_8way_close( &ctx.keccak, vhash );
rintrlv_8x64_4x128( vhashA, vhashB, vhash, 256 );
dintrlv_8x64( hash0, hash1, hash2, hash3,
hash4, hash5, hash6, hash7, vhash, 256 );
cube_4way_update_close( &ctx.cube, vhashA, vhashA, 32 );
cube_4way_init( &ctx.cube, 256, 16, 32 );
cube_4way_update_close( &ctx.cube, vhashB, vhashB, 32 );
cubehash_full( &ctx.cube, (byte*) hash0, 256, (const byte*) hash0, 32 );
cubehash_full( &ctx.cube, (byte*) hash1, 256, (const byte*) hash1, 32 );
cubehash_full( &ctx.cube, (byte*) hash2, 256, (const byte*) hash2, 32 );
cubehash_full( &ctx.cube, (byte*) hash3, 256, (const byte*) hash3, 32 );
cubehash_full( &ctx.cube, (byte*) hash4, 256, (const byte*) hash4, 32 );
cubehash_full( &ctx.cube, (byte*) hash5, 256, (const byte*) hash5, 32 );
cubehash_full( &ctx.cube, (byte*) hash6, 256, (const byte*) hash6, 32 );
cubehash_full( &ctx.cube, (byte*) hash7, 256, (const byte*) hash7, 32 );
dintrlv_4x128( hash0, hash1, hash2, hash3, vhashA, 256 );
dintrlv_4x128( hash4, hash5, hash6, hash7, vhashB, 256 );
// cube_4way_update_close( &ctx.cube, vhashA, vhashA, 32 );
// cube_4way_init( &ctx.cube, 256, 16, 32 );
// cube_4way_update_close( &ctx.cube, vhashB, vhashB, 32 );
//
// dintrlv_4x128( hash0, hash1, hash2, hash3, vhashA, 256 );
// dintrlv_4x128( hash4, hash5, hash6, hash7, vhashB, 256 );
intrlv_2x256( vhash, hash0, hash1, 256 );
LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
@@ -80,15 +71,123 @@ void lyra2rev2_8way_hash( void *state, const void *input )
skein256_8way_update( &ctx.skein, vhash, 32 );
skein256_8way_close( &ctx.skein, vhash );
rintrlv_8x64_4x128( vhashA, vhashB, vhash, 256 );
dintrlv_8x64( hash0, hash1, hash2, hash3,
hash4, hash5, hash6, hash7, vhash, 256 );
cube_4way_init( &ctx.cube, 256, 16, 32 );
cube_4way_update_close( &ctx.cube, vhashA, vhashA, 32 );
cube_4way_init( &ctx.cube, 256, 16, 32 );
cube_4way_update_close( &ctx.cube, vhashB, vhashB, 32 );
dintrlv_4x128( hash0, hash1, hash2, hash3, vhashA, 256 );
dintrlv_4x128( hash4, hash5, hash6, hash7, vhashB, 256 );
cubehash_full( &ctx.cube, (byte*) hash0, 256, (const byte*) hash0, 32 );
cubehash_full( &ctx.cube, (byte*) hash1, 256, (const byte*) hash1, 32 );
cubehash_full( &ctx.cube, (byte*) hash2, 256, (const byte*) hash2, 32 );
cubehash_full( &ctx.cube, (byte*) hash3, 256, (const byte*) hash3, 32 );
cubehash_full( &ctx.cube, (byte*) hash4, 256, (const byte*) hash4, 32 );
cubehash_full( &ctx.cube, (byte*) hash5, 256, (const byte*) hash5, 32 );
cubehash_full( &ctx.cube, (byte*) hash6, 256, (const byte*) hash6, 32 );
cubehash_full( &ctx.cube, (byte*) hash7, 256, (const byte*) hash7, 32 );
// cube_4way_init( &ctx.cube, 256, 16, 32 );
// cube_4way_update_close( &ctx.cube, vhashA, vhashA, 32 );
// cube_4way_init( &ctx.cube, 256, 16, 32 );
// cube_4way_update_close( &ctx.cube, vhashB, vhashB, 32 );
//
// dintrlv_4x128( hash0, hash1, hash2, hash3, vhashA, 256 );
// dintrlv_4x128( hash4, hash5, hash6, hash7, vhashB, 256 );
intrlv_8x32( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, 256 );
bmw256_8way_update( &ctx.bmw, vhash, 32 );
bmw256_8way_close( &ctx.bmw, state );
}
#endif
#if defined (LYRA2REV2_8WAY)
typedef struct {
blake256_8way_context blake;
keccak256_8way_context keccak;
cubehashParam cube;
skein256_8way_context skein;
bmw256_8way_context bmw;
} lyra2v2_8way_ctx_holder __attribute__ ((aligned (64)));
static lyra2v2_8way_ctx_holder l2v2_8way_ctx;
bool init_lyra2rev2_8way_ctx()
{
keccak256_8way_init( &l2v2_8way_ctx.keccak );
cubehashInit( &l2v2_8way_ctx.cube, 256, 16, 32 );
skein256_8way_init( &l2v2_8way_ctx.skein );
bmw256_8way_init( &l2v2_8way_ctx.bmw );
return true;
}
void lyra2rev2_8way_hash( void *state, const void *input )
{
uint32_t vhash[8*8] __attribute__ ((aligned (128)));
uint32_t vhashA[8*8] __attribute__ ((aligned (64)));
uint32_t hash0[8] __attribute__ ((aligned (64)));
uint32_t hash1[8] __attribute__ ((aligned (64)));
uint32_t hash2[8] __attribute__ ((aligned (64)));
uint32_t hash3[8] __attribute__ ((aligned (64)));
uint32_t hash4[8] __attribute__ ((aligned (64)));
uint32_t hash5[8] __attribute__ ((aligned (64)));
uint32_t hash6[8] __attribute__ ((aligned (64)));
uint32_t hash7[8] __attribute__ ((aligned (64)));
lyra2v2_8way_ctx_holder ctx __attribute__ ((aligned (64)));
memcpy( &ctx, &l2v2_8way_ctx, sizeof(l2v2_8way_ctx) );
blake256_8way_update( &ctx.blake, input + (64<<3), 16 );
blake256_8way_close( &ctx.blake, vhash );
rintrlv_8x32_8x64( vhashA, vhash, 256 );
keccak256_8way_update( &ctx.keccak, vhashA, 32 );
keccak256_8way_close( &ctx.keccak, vhash );
dintrlv_8x64( hash0, hash1, hash2, hash3,
hash4, hash5, hash6, hash7, vhash, 256 );
cubehash_full( &ctx.cube, (byte*) hash0, 256, (const byte*) hash0, 32 );
cubehash_full( &ctx.cube, (byte*) hash1, 256, (const byte*) hash1, 32 );
cubehash_full( &ctx.cube, (byte*) hash2, 256, (const byte*) hash2, 32 );
cubehash_full( &ctx.cube, (byte*) hash3, 256, (const byte*) hash3, 32 );
cubehash_full( &ctx.cube, (byte*) hash4, 256, (const byte*) hash4, 32 );
cubehash_full( &ctx.cube, (byte*) hash5, 256, (const byte*) hash5, 32 );
cubehash_full( &ctx.cube, (byte*) hash6, 256, (const byte*) hash6, 32 );
cubehash_full( &ctx.cube, (byte*) hash7, 256, (const byte*) hash7, 32 );
intrlv_2x256( vhash, hash0, hash1, 256 );
LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
dintrlv_2x256( hash0, hash1, vhash, 256 );
intrlv_2x256( vhash, hash2, hash3, 256 );
LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
dintrlv_2x256( hash2, hash3, vhash, 256 );
intrlv_2x256( vhash, hash4, hash5, 256 );
LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
dintrlv_2x256( hash4, hash5, vhash, 256 );
intrlv_2x256( vhash, hash6, hash7, 256 );
LYRA2REV2_2WAY( l2v2_wholeMatrix, vhash, 32, vhash, 32, 1, 4, 4 );
dintrlv_2x256( hash6, hash7, vhash, 256 );
intrlv_8x64( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, 256 );
skein256_8way_update( &ctx.skein, vhash, 32 );
skein256_8way_close( &ctx.skein, vhash );
dintrlv_8x64( hash0, hash1, hash2, hash3,
hash4, hash5, hash6, hash7, vhash, 256 );
cubehash_full( &ctx.cube, (byte*) hash0, 256, (const byte*) hash0, 32 );
cubehash_full( &ctx.cube, (byte*) hash1, 256, (const byte*) hash1, 32 );
cubehash_full( &ctx.cube, (byte*) hash2, 256, (const byte*) hash2, 32 );
cubehash_full( &ctx.cube, (byte*) hash3, 256, (const byte*) hash3, 32 );
cubehash_full( &ctx.cube, (byte*) hash4, 256, (const byte*) hash4, 32 );
cubehash_full( &ctx.cube, (byte*) hash5, 256, (const byte*) hash5, 32 );
cubehash_full( &ctx.cube, (byte*) hash6, 256, (const byte*) hash6, 32 );
cubehash_full( &ctx.cube, (byte*) hash7, 256, (const byte*) hash7, 32 );
intrlv_8x32( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, 256 );
@@ -97,49 +196,49 @@ void lyra2rev2_8way_hash( void *state, const void *input )
bmw256_8way_close( &ctx.bmw, state );
}
int scanhash_lyra2rev2_8way( struct work *work, uint32_t max_nonce,
int scanhash_lyra2rev2_8way( struct work *work, const uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t hash[8*8] __attribute__ ((aligned (128)));
uint32_t vdata[20*8] __attribute__ ((aligned (64)));
uint32_t *hash7 = &(hash[7<<3]);
uint32_t lane_hash[8] __attribute__ ((aligned (64)));
uint32_t *hashd7 = &hash[7*8];
uint32_t lane_hash[8] __attribute__ ((aligned (32)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
const uint32_t last_nonce = max_nonce - 8;
uint32_t n = first_nonce;
const uint32_t Htarg = ptarget[7];
__m256i *noncev = (__m256i*)vdata + 19; // aligned
int thr_id = mythr->id;
const uint32_t targ32 = ptarget[7];
__m256i *noncev = (__m256i*)vdata + 19;
const int thr_id = mythr->id;
const bool bench = opt_benchmark;
if ( opt_benchmark )
( (uint32_t*)ptarget )[7] = 0x0000ff;
if ( bench ) ptarget[7] = 0x0000ff;
mm256_bswap32_intrlv80_8x32( vdata, pdata );
*noncev = _mm256_set_epi32( n+7, n+6, n+5, n+4, n+3, n+2, n+1, n );
blake256_8way_init( &l2v2_8way_ctx.blake );
blake256_8way_update( &l2v2_8way_ctx.blake, vdata, 64 );
do
{
*noncev = mm256_bswap_32( _mm256_set_epi32( n+7, n+6, n+5, n+4,
n+3, n+2, n+1, n ) );
lyra2rev2_8way_hash( hash, vdata );
pdata[19] = n;
for ( int lane = 0; lane < 8; lane++ ) if ( hash7[lane] <= Htarg )
for ( int lane = 0; lane < 8; lane++ )
if ( unlikely( hashd7[lane] <= targ32 ) )
{
extr_lane_8x32( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
if ( likely( valid_hash( lane_hash, ptarget ) && !bench ) )
{
pdata[19] = n + lane;
submit_lane_solution( work, lane_hash, mythr, lane );
pdata[19] = bswap_32( n + lane );
submit_lane_solution( work, lane_hash, mythr, lane );
}
}
*noncev = _mm256_add_epi32( *noncev, m256_const1_32( 8 ) );
n += 8;
} while ( (n < last_nonce) && !work_restart[thr_id].restart);
} while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );
pdata[19] = n;
*hashes_done = n - first_nonce;
return 0;
}
@@ -226,15 +325,16 @@ int scanhash_lyra2rev2_4way( struct work *work, uint32_t max_nonce,
{
uint32_t hash[8*4] __attribute__ ((aligned (64)));
uint32_t vdata[20*4] __attribute__ ((aligned (64)));
uint32_t *hash7 = &(hash[7<<2]);
uint32_t *hashd7 = &(hash[7<<2]);
uint32_t lane_hash[8] __attribute__ ((aligned (32)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
const uint32_t last_nonce = max_nonce - 4;
uint32_t n = first_nonce;
const uint32_t Htarg = ptarget[7];
__m128i *noncev = (__m128i*)vdata + 19; // aligned
int thr_id = mythr->id; // thr_id arg is deprecated
const uint32_t targ32 = ptarget[7];
__m128i *noncev = (__m128i*)vdata + 19;
int thr_id = mythr->id;
if ( opt_benchmark )
( (uint32_t*)ptarget )[7] = 0x0000ff;
@@ -249,20 +349,20 @@ int scanhash_lyra2rev2_4way( struct work *work, uint32_t max_nonce,
*noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
lyra2rev2_4way_hash( hash, vdata );
pdata[19] = n;
for ( int lane = 0; lane < 4; lane++ ) if ( hash7[lane] <= Htarg )
for ( int lane = 0; lane < 4; lane++ ) if ( hashd7[lane] <= targ32 )
{
extr_lane_4x32( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
if ( valid_hash( lane_hash, ptarget ) && !opt_benchmark )
{
pdata[19] = n + lane;
submit_lane_solution( work, lane_hash, mythr, lane );
}
}
n += 4;
} while ( (n < max_nonce-4) && !work_restart[thr_id].restart);
*hashes_done = n - first_nonce + 1;
} while ( (n < last_nonce) && !work_restart[thr_id].restart);
pdata[19] = n;
*hashes_done = n - first_nonce;
return 0;
}

2
algo/lyra2/lyra2rev2.c
View File

@@ -99,7 +99,7 @@ int scanhash_lyra2rev2( struct work *work,
lyra2rev2_hash(hash, endiandata);
if (hash[7] <= Htarg )
if( fulltest( hash, ptarget ) && !opt_benchmark )
if( valid_hash( hash, ptarget ) && !opt_benchmark )
{
pdata[19] = nonce;
submit_solution( work, hash, mythr );

20
algo/lyra2/lyra2rev3-4way.c
View File

@@ -130,7 +130,7 @@ int scanhash_lyra2rev3_16way( struct work *work, const uint32_t max_nonce,
{
uint32_t hash[8*16] __attribute__ ((aligned (128)));
uint32_t vdata[20*16] __attribute__ ((aligned (64)));
uint32_t *hash32 = &hash[7*16];
uint32_t *hashd7 = &hash[7*16];
uint32_t lane_hash[8] __attribute__ ((aligned (64)));
uint32_t *pdata = work->data;
const uint32_t *ptarget = work->target;
@@ -159,10 +159,10 @@ int scanhash_lyra2rev3_16way( struct work *work, const uint32_t max_nonce,
pdata[19] = n;
for ( int lane = 0; lane < 16; lane++ )
if ( unlikely( hash32[lane] <= targ32 ) )
if ( unlikely( hashd7[lane] <= targ32 ) )
{
extr_lane_16x32( lane_hash, hash, lane, 256 );
if ( likely( fulltest( lane_hash, ptarget ) && !opt_benchmark ) )
if ( likely( valid_hash( lane_hash, ptarget ) && !opt_benchmark ) )
{
pdata[19] = n + lane;
submit_lane_solution( work, lane_hash, mythr, lane );
@@ -170,6 +170,7 @@ int scanhash_lyra2rev3_16way( struct work *work, const uint32_t max_nonce,
}
n += 16;
} while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );
pdata[19] = n;
*hashes_done = n - first_nonce;
return 0;
}
@@ -194,7 +195,7 @@ bool init_lyra2rev3_8way_ctx()
void lyra2rev3_8way_hash( void *state, const void *input )
{
uint32_t vhash[8*8] __attribute__ ((aligned (64)));
uint32_t vhash[8*8] __attribute__ ((aligned (128)));
uint32_t hash0[8] __attribute__ ((aligned (64)));
uint32_t hash1[8] __attribute__ ((aligned (32)));
uint32_t hash2[8] __attribute__ ((aligned (32)));
@@ -250,9 +251,9 @@ void lyra2rev3_8way_hash( void *state, const void *input )
int scanhash_lyra2rev3_8way( struct work *work, const uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t hash[8*8] __attribute__ ((aligned (64)));
uint32_t hash[8*8] __attribute__ ((aligned (128)));
uint32_t vdata[20*8] __attribute__ ((aligned (64)));
uint32_t *hash32 = &hash[7*8];
uint32_t *hashd7 = &hash[7*8];
uint32_t lane_hash[8] __attribute__ ((aligned (32)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
@@ -277,7 +278,7 @@ int scanhash_lyra2rev3_8way( struct work *work, const uint32_t max_nonce,
pdata[19] = n;
for ( int lane = 0; lane < 8; lane++ )
if ( unlikely( hash32[lane] <= targ32 ) )
if ( unlikely( hashd7[lane] <= targ32 ) )
{
extr_lane_8x32( lane_hash, hash, lane, 256 );
if ( likely( valid_hash( lane_hash, ptarget ) && !bench ) )
@@ -357,7 +358,7 @@ int scanhash_lyra2rev3_4way( struct work *work, const uint32_t max_nonce,
{
uint32_t hash[8*4] __attribute__ ((aligned (64)));
uint32_t vdata[20*4] __attribute__ ((aligned (64)));
uint32_t *hash32 = &(hash[7*4]);
uint32_t *hashd7 = &(hash[7*4]);
uint32_t lane_hash[8] __attribute__ ((aligned (32)));
uint32_t *pdata = work->data;
const uint32_t *ptarget = work->target;
@@ -379,7 +380,7 @@ int scanhash_lyra2rev3_4way( struct work *work, const uint32_t max_nonce,
do
{
lyra2rev3_4way_hash( hash, vdata );
for ( int lane = 0; lane < 4; lane++ ) if ( hash32[lane] <= targ32 )
for ( int lane = 0; lane < 4; lane++ ) if ( hashd7[lane] <= targ32 )
{
extr_lane_4x32( lane_hash, hash, lane, 256 );
if ( valid_hash( lane_hash, ptarget ) && !opt_benchmark )
@@ -391,6 +392,7 @@ int scanhash_lyra2rev3_4way( struct work *work, const uint32_t max_nonce,
*noncev = _mm_add_epi32( *noncev, m128_const1_32( 4 ) );
n += 4;
} while ( (n < max_nonce-4) && !work_restart[thr_id].restart);
pdata[19] = n;
*hashes_done = n - first_nonce + 1;
return 0;
}

2
algo/lyra2/lyra2rev3.c
View File

@@ -88,7 +88,7 @@ int scanhash_lyra2rev3( struct work *work,
lyra2rev3_hash(hash, endiandata);
if (hash[7] <= Htarg )
if( fulltest( hash, ptarget ) && !opt_benchmark )
if( valid_hash( hash, ptarget ) && !opt_benchmark )
{
pdata[19] = nonce;
submit_solution( work, hash, mythr );

7
algo/lyra2/lyra2z.c
View File

@@ -56,7 +56,7 @@ int scanhash_lyra2z( struct work *work, uint32_t max_nonce,
const uint32_t Htarg = ptarget[7];
const uint32_t first_nonce = pdata[19];
uint32_t nonce = first_nonce;
int thr_id = mythr->id; // thr_id arg is deprecated
int thr_id = mythr->id;
if (opt_benchmark)
ptarget[7] = 0x0000ff;
@@ -65,14 +65,13 @@ int scanhash_lyra2z( struct work *work, uint32_t max_nonce,
be32enc(&endiandata[i], pdata[i]);
}
lyra2z_midstate( endiandata );
lyra2z_midstate( endiandata );
do {
be32enc(&endiandata[19], nonce);
lyra2z_hash( hash, endiandata );
if ( hash[7] <= Htarg )
if ( fulltest( hash, ptarget ) && !opt_benchmark )
if ( valid_hash( hash, ptarget ) && !opt_benchmark )
{
pdata[19] = nonce;
submit_solution( work, hash, mythr );

34
algo/lyra2/lyra2z330.c
View File

@@ -9,7 +9,7 @@ void lyra2z330_hash(void *state, const void *input, uint32_t height)
{
uint32_t _ALIGN(256) hash[16];
LYRA2Z( lyra2z330_wholeMatrix, hash, 32, input, 80, input, 80,
LYRA2Z( lyra2z330_wholeMatrix, hash, 32, input, 80, input, 80,
2, 330, 256 );
memcpy(state, hash, 32);
@@ -18,38 +18,40 @@ void lyra2z330_hash(void *state, const void *input, uint32_t height)
int scanhash_lyra2z330( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t hash[8] __attribute__ ((aligned (64)));
uint32_t endiandata[20] __attribute__ ((aligned (64)));
uint32_t hash[8] __attribute__ ((aligned (128)));
uint32_t edata[20] __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 nonce = first_nonce;
int thr_id = mythr->id; // thr_id arg is deprecated
const int thr_id = mythr->id;
if (opt_benchmark)
ptarget[7] = 0x0000ff;
casti_m128i( endiandata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
casti_m128i( endiandata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
casti_m128i( endiandata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
casti_m128i( endiandata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
casti_m128i( endiandata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
casti_m128i( edata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
casti_m128i( edata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
casti_m128i( edata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
casti_m128i( edata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
casti_m128i( edata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
do
{
be32enc( &endiandata[19], nonce );
lyra2z330_hash( hash, endiandata, work->height );
if ( hash[7] <= Htarg )
if ( fulltest( hash, ptarget ) && !opt_benchmark )
edata[19] = nonce;
LYRA2Z( lyra2z330_wholeMatrix, hash, 32, edata, 80, edata, 80,
2, 330, 256 );
// lyra2z330_hash( hash, edata, work->height );
if ( valid_hash( hash, ptarget ) && !opt_benchmark )
{
pdata[19] = nonce;
be32enc( pdata + 19, nonce );
submit_solution( work, hash, mythr );
}
nonce++;
} while ( nonce < max_nonce && !work_restart[thr_id].restart );
pdata[19] = nonce;
*hashes_done = pdata[19] - first_nonce + 1;
*hashes_done = nonce - first_nonce;
return 0;
}