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

v3.9.5.4

Browse Source
This commit is contained in:
Jay D Dee
2019-07-15 17:00:26 -04:00
parent e625ed5420
commit e2d5762ef2
63 changed files with 1973 additions and 2980 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
algo/x13/phi1612-4way.c
View File

@@ -53,7 +53,7 @@ void phi1612_4way_hash( void *state, const void *input )
jh512_4way_close( &ctx.jh, vhash );
// Serial to the end
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// Cubehash
cubehashUpdateDigest( &ctx.cube, (byte*)hash0, (const byte*) hash0, 64 );
@@ -114,7 +114,6 @@ int scanhash_phi1612_4way( struct work *work, uint32_t max_nonce,
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
@@ -125,9 +124,7 @@ int scanhash_phi1612_4way( struct work *work, uint32_t max_nonce,
if ( opt_benchmark )
( (uint32_t*)ptarget )[7] = 0x0cff;
swab32_array( edata, pdata, 20 );
mm256_intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
// mm256_bswap_intrlv80_4x64( vdata, pdata );
mm256_bswap32_intrlv80_4x64( vdata, pdata );
do {
*noncev = mm256_intrlv_blend_32( mm256_bswap_32(

7
algo/x13/skunk-4way.c
View File

@@ -33,7 +33,7 @@ void skunk_4way_hash( void *output, const void *input )
skein512_4way( &ctx.skein, input, 80 );
skein512_4way_close( &ctx.skein, vhash );
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash0, (const byte*)hash0, 64 );
memcpy( &ctx.cube, &skunk_4way_ctx.cube, sizeof(cubehashParam) );
@@ -78,7 +78,6 @@ int scanhash_skunk_4way( struct work *work, uint32_t max_nonce,
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
@@ -91,9 +90,7 @@ int scanhash_skunk_4way( struct work *work, uint32_t max_nonce,
if ( opt_benchmark )
((uint32_t*)ptarget)[7] = 0x0cff;
swab32_array( edata, pdata, 20 );
mm256_intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
// mm256_bswap_intrlv80_4x64( vdata, pdata );
mm256_bswap32_intrlv80_4x64( vdata, pdata );
do
{
*noncev = mm256_intrlv_blend_32( mm256_bswap_32(

31
algo/x13/x13-4way.c
View File

@@ -74,7 +74,7 @@ void x13_4way_hash( void *state, const void *input )
bmw512_4way_close( &ctx.bmw, vhash );
// Serial
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// 3 Groestl
update_and_final_groestl( &ctx.groestl, (char*)hash0, (char*)hash0, 512 );
@@ -86,7 +86,7 @@ void x13_4way_hash( void *state, const void *input )
update_and_final_groestl( &ctx.groestl, (char*)hash3, (char*)hash3, 512 );
// Parallel 4way 64 bit
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
// 4 Skein
skein512_4way( &ctx.skein, vhash, 64 );
@@ -101,16 +101,16 @@ void x13_4way_hash( void *state, const void *input )
keccak512_4way_close( &ctx.keccak, vhash );
// Serial
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// 7 Luffa
mm256_intrlv_2x128( vhash, hash0, hash1, 512 );
intrlv_2x128( vhash, hash0, hash1, 512 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
mm256_dintrlv_2x128( hash0, hash1, vhash, 512 );
mm256_intrlv_2x128( vhash, hash2, hash3, 512 );
dintrlv_2x128( hash0, hash1, vhash, 512 );
intrlv_2x128( vhash, hash2, hash3, 512 );
luffa_2way_init( &ctx.luffa, 512 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
mm256_dintrlv_2x128( hash2, hash3, vhash, 512 );
dintrlv_2x128( hash2, hash3, vhash, 512 );
// 8 Cubehash
cubehashUpdateDigest( &ctx.cube, (byte*)hash0, (const byte*) hash0, 64 );
@@ -138,13 +138,13 @@ void x13_4way_hash( void *state, const void *input )
sph_shavite512_close( &ctx.shavite, hash3 );
// 10 Simd
mm256_intrlv_2x128( vhash, hash0, hash1, 512 );
intrlv_2x128( vhash, hash0, hash1, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
mm256_dintrlv_2x128( hash0, hash1, vhash, 512 );
mm256_intrlv_2x128( vhash, hash2, hash3, 512 );
dintrlv_2x128( hash0, hash1, vhash, 512 );
intrlv_2x128( vhash, hash2, hash3, 512 );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
mm256_dintrlv_2x128( hash2, hash3, vhash, 512 );
dintrlv_2x128( hash2, hash3, vhash, 512 );
// 11 Echo
update_final_echo( &ctx.echo, (BitSequence *)hash0,
@@ -160,10 +160,10 @@ void x13_4way_hash( void *state, const void *input )
(const BitSequence *) hash3, 512 );
// 12 Hamsi parallel 4way 32 bit
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
hamsi512_4way( &ctx.hamsi, vhash, 64 );
hamsi512_4way_close( &ctx.hamsi, vhash );
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// 13 Fugue serial
sph_fugue512( &ctx.fugue, hash0, 64 );
@@ -189,7 +189,6 @@ int scanhash_x13_4way( struct work *work, uint32_t max_nonce,
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
uint32_t n = pdata[19];
@@ -202,9 +201,7 @@ int scanhash_x13_4way( struct work *work, uint32_t max_nonce,
uint32_t masks[] = { 0xFFFFFFFF, 0xFFFFFFF0, 0xFFFFFF00,
0xFFFFF000, 0xFFFF0000, 0 };
swab32_array( edata, pdata, 20 );
mm256_intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
// mm256_bswap_intrlv80_4x64( vdata, pdata );
mm256_bswap32_intrlv80_4x64( vdata, pdata );
for ( int m=0; m < 6; m++ )
if ( Htarg <= htmax[m] )

31
algo/x13/x13sm3-4way.c
View File

@@ -81,7 +81,7 @@ void x13sm3_4way_hash( void *state, const void *input )
bmw512_4way_close( &ctx.bmw, vhash );
// Serial
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// Groestl
update_and_final_groestl( &ctx.groestl, (char*)hash0, (char*)hash0, 512 );
@@ -93,7 +93,7 @@ void x13sm3_4way_hash( void *state, const void *input )
update_and_final_groestl( &ctx.groestl, (char*)hash3, (char*)hash3, 512 );
// Parallel 4way
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
// Skein
skein512_4way( &ctx.skein, vhash, 64 );
@@ -108,16 +108,16 @@ void x13sm3_4way_hash( void *state, const void *input )
keccak512_4way_close( &ctx.keccak, vhash );
// Serial to the end
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// Luffa
mm256_intrlv_2x128( vhash, hash0, hash1, 512 );
intrlv_2x128( vhash, hash0, hash1, 512 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
mm256_dintrlv_2x128( hash0, hash1, vhash, 512 );
mm256_intrlv_2x128( vhash, hash2, hash3, 512 );
dintrlv_2x128( hash0, hash1, vhash, 512 );
intrlv_2x128( vhash, hash2, hash3, 512 );
luffa_2way_init( &ctx.luffa, 512 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
mm256_dintrlv_2x128( hash2, hash3, vhash, 512 );
dintrlv_2x128( hash2, hash3, vhash, 512 );
// Cubehash
cubehashUpdateDigest( &ctx.cube, (byte*)hash0, (const byte*) hash0, 64 );
@@ -145,13 +145,13 @@ void x13sm3_4way_hash( void *state, const void *input )
sph_shavite512_close( &ctx.shavite, hash3 );
// Simd
mm256_intrlv_2x128( vhash, hash0, hash1, 512 );
intrlv_2x128( vhash, hash0, hash1, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
mm256_dintrlv_2x128( hash0, hash1, vhash, 512 );
mm256_intrlv_2x128( vhash, hash2, hash3, 512 );
dintrlv_2x128( hash0, hash1, vhash, 512 );
intrlv_2x128( vhash, hash2, hash3, 512 );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
mm256_dintrlv_2x128( hash2, hash3, vhash, 512 );
dintrlv_2x128( hash2, hash3, vhash, 512 );
// Echo
update_final_echo( &ctx.echo, (BitSequence *)hash0,
@@ -185,10 +185,10 @@ void x13sm3_4way_hash( void *state, const void *input )
dintrlv_4x32( hash0, hash1, hash2, hash3, sm3_vhash, 512 );
// Hamsi parallel 4x32x2
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
hamsi512_4way( &ctx.hamsi, vhash, 64 );
hamsi512_4way_close( &ctx.hamsi, vhash );
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// Fugue serial
sph_fugue512( &ctx.fugue, hash0, 64 );
@@ -214,7 +214,6 @@ int scanhash_x13sm3_4way( struct work *work, uint32_t max_nonce,
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
uint32_t n = pdata[19];
@@ -227,9 +226,7 @@ int scanhash_x13sm3_4way( struct work *work, uint32_t max_nonce,
uint32_t masks[] = { 0xFFFFFFFF, 0xFFFFFFF0, 0xFFFFFF00,
0xFFFFF000, 0xFFFF0000, 0 };
swab32_array( edata, pdata, 20 );
mm256_intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
// mm256_bswap_intrlv80_4x64( vdata, pdata );
mm256_bswap32_intrlv80_4x64( vdata, pdata );
blake512_4way_init( &x13sm3_ctx_mid );
blake512_4way( &x13sm3_ctx_mid, vdata, 64 );