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

v3.9.5.1

Browse Source
This commit is contained in:
Jay D Dee
2019-07-02 15:10:38 -04:00
parent 0d48d573ce
commit 0d769ee0fe
53 changed files with 1755 additions and 1170 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
algo/lyra2/lyra2h-4way.c
View File

@@ -36,7 +36,7 @@ void lyra2h_4way_hash( void *state, const void *input )
blake256_4way( &ctx_blake, input + (64*4), 16 );
blake256_4way_close( &ctx_blake, vhash );
mm128_dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 256 );
dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 256 );
LYRA2Z( lyra2h_4way_matrix, state, 32, hash0, 32, hash0, 32,
16, 16, 16 );

6
algo/lyra2/lyra2rev2-4way.c
View File

@@ -78,7 +78,7 @@ void lyra2rev2_4way_hash( void *state, const void *input )
cubehashInit( &ctx.cube, 256, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash3, (const byte*) hash3, 32 );
mm128_intrlv_4x32( vhash, hash0, hash1, hash2, hash3, 256 );
intrlv_4x32( vhash, hash0, hash1, hash2, hash3, 256 );
bmw256_4way( &ctx.bmw, vhash, 32 );
bmw256_4way_close( &ctx.bmw, state );
@@ -90,7 +90,7 @@ 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 lane_hash[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];
@@ -116,7 +116,7 @@ int scanhash_lyra2rev2_4way( struct work *work, uint32_t max_nonce,
for ( int lane = 0; lane < 4; lane++ ) if ( hash7[lane] <= Htarg )
{
mm128_extract_lane_4x32( lane_hash, hash, lane, 256 );
extr_lane_4x32( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
{
pdata[19] = n + lane;

12
algo/lyra2/lyra2rev3-4way.c
View File

@@ -92,7 +92,7 @@ int scanhash_lyra2rev3_8way( struct work *work, uint32_t max_nonce,
uint32_t hash[8*8] __attribute__ ((aligned (64)));
uint32_t vdata[20*8] __attribute__ ((aligned (64)));
uint32_t *hash7 = &(hash[7<<3]);
uint32_t lane_hash[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];
@@ -115,7 +115,7 @@ int scanhash_lyra2rev3_8way( struct work *work, uint32_t max_nonce,
for ( int lane = 0; lane < 8; lane++ ) if ( hash7[lane] <= Htarg )
{
mm256_extract_lane_8x32( lane_hash, hash, lane, 256 );
mm256_extr_lane_8x32( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
{
pdata[19] = n + lane;
@@ -161,7 +161,7 @@ void lyra2rev3_4way_hash( void *state, const void *input )
blake256_4way( &ctx.blake, input, 80 );
blake256_4way_close( &ctx.blake, vhash );
mm128_dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 256 );
dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 256 );
LYRA2REV3( l2v3_wholeMatrix, hash0, 32, hash0, 32, hash0, 32, 1, 4, 4 );
LYRA2REV3( l2v3_wholeMatrix, hash1, 32, hash1, 32, hash1, 32, 1, 4, 4 );
@@ -181,7 +181,7 @@ void lyra2rev3_4way_hash( void *state, const void *input )
LYRA2REV3( l2v3_wholeMatrix, hash2, 32, hash2, 32, hash2, 32, 1, 4, 4 );
LYRA2REV3( l2v3_wholeMatrix, hash3, 32, hash3, 32, hash3, 32, 1, 4, 4 );
mm128_intrlv_4x32( vhash, hash0, hash1, hash2, hash3, 256 );
intrlv_4x32( vhash, hash0, hash1, hash2, hash3, 256 );
bmw256_4way( &ctx.bmw, vhash, 32 );
bmw256_4way_close( &ctx.bmw, state );
}
@@ -192,7 +192,7 @@ int scanhash_lyra2rev3_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 lane_hash[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];
@@ -214,7 +214,7 @@ int scanhash_lyra2rev3_4way( struct work *work, uint32_t max_nonce,
for ( int lane = 0; lane < 4; lane++ ) if ( hash7[lane] <= Htarg )
{
mm128_extract_lane_4x32( lane_hash, hash, lane, 256 );
extr_lane_4x32( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
{
pdata[19] = n + lane;

2
algo/lyra2/lyra2z-4way.c
View File

@@ -36,7 +36,7 @@ void lyra2z_4way_hash( void *state, const void *input )
blake256_4way( &ctx_blake, input + (64*4), 16 );
blake256_4way_close( &ctx_blake, vhash );
mm128_dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 256 );
dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 256 );
LYRA2Z( lyra2z_4way_matrix, state , 32, hash0, 32, hash0, 32, 8, 8, 8 );
LYRA2Z( lyra2z_4way_matrix, state+32, 32, hash1, 32, hash1, 32, 8, 8, 8 );

4
algo/lyra2/phi2-4way.c
View File

@@ -168,7 +168,7 @@ int scanhash_phi2_4way( struct work *work, uint32_t max_nonce,
uint32_t _ALIGN(128) edata[36];
uint32_t vdata[4][36] __attribute__ ((aligned (64)));
uint32_t *hash7 = &(hash[25]);
uint32_t lane_hash[8];
uint32_t lane_hash[8] __attribute__ ((aligned (32)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t Htarg = ptarget[7];
@@ -217,7 +217,7 @@ int scanhash_phi2_4way( struct work *work, uint32_t max_nonce,
for ( int lane = 0; lane < 4; lane++ ) if ( hash7[ lane<<1 ] < Htarg )
{
mm256_extract_lane_4x64( lane_hash, hash, lane, 256 );
mm256_extr_lane_4x64( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
{
pdata[19] = n + lane;