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

v3.9.2.5

Browse Source
This commit is contained in:
Jay D Dee
2019-06-13 11:20:27 -04:00
parent 7fec680835
commit b2331375a3
70 changed files with 4413 additions and 4360 deletions
Download Patch File Download Diff File Expand all files Collapse all files

68
algo/x17/xevan-4way.c
View File

@@ -71,7 +71,7 @@ void xevan_4way_hash( void *output, const void *input )
bmw512_4way_close( &ctx.bmw, vhash );
// Serial
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash0, (char*)hash0,
@@ -87,7 +87,7 @@ void xevan_4way_hash( void *output, const void *input )
dataLen<<3 );
// Parallel 4way
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
skein512_4way_init( &ctx.skein );
skein512_4way( &ctx.skein, vhash, dataLen );
@@ -101,7 +101,7 @@ void xevan_4way_hash( void *output, const void *input )
keccak512_4way( &ctx.keccak, vhash, dataLen );
keccak512_4way_close( &ctx.keccak, vhash );
mm256_reinterleave_4x64_2x128( vhashA, vhashB, vhash, dataLen<<3 );
mm256_rintrlv_4x64_2x128( vhashA, vhashB, vhash, dataLen<<3 );
luffa_2way_init( &ctx.luffa, 512 );
luffa_2way_update_close( &ctx.luffa, vhashA, vhashA, dataLen );
@@ -123,8 +123,8 @@ void xevan_4way_hash( void *output, const void *input )
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhashB, vhashB, dataLen<<3 );
mm256_deinterleave_1x128( hash0, hash1, vhashA, dataLen<<3 );
mm256_deinterleave_1x128( hash2, hash3, vhashB, dataLen<<3 );
mm256_dintrlv_2x128( hash0, hash1, vhashA, dataLen<<3 );
mm256_dintrlv_2x128( hash2, hash3, vhashB, dataLen<<3 );
init_echo( &ctx.echo, 512 );
update_final_echo( &ctx.echo, (BitSequence *)hash0,
@@ -139,13 +139,13 @@ void xevan_4way_hash( void *output, const void *input )
update_final_echo( &ctx.echo, (BitSequence *)hash3,
(const BitSequence *) hash3, dataLen<<3 );
// Parallel
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
hamsi512_4way_init( &ctx.hamsi );
hamsi512_4way( &ctx.hamsi, vhash, dataLen );
hamsi512_4way_close( &ctx.hamsi, vhash );
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, hash0, dataLen );
@@ -161,13 +161,13 @@ void xevan_4way_hash( void *output, const void *input )
sph_fugue512_close( &ctx.fugue, hash3 );
// Parallel 4way 32 bit
mm128_interleave_4x32( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm128_intrlv_4x32( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
shabal512_4way_init( &ctx.shabal );
shabal512_4way( &ctx.shabal, vhash, dataLen );
shabal512_4way_close( &ctx.shabal, vhash );
mm128_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm128_dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
// Serial
sph_whirlpool_init( &ctx.whirlpool );
@@ -183,19 +183,19 @@ void xevan_4way_hash( void *output, const void *input )
sph_whirlpool( &ctx.whirlpool, hash3, dataLen );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
sha512_4way_init( &ctx.sha512 );
sha512_4way( &ctx.sha512, vhash, dataLen );
sha512_4way_close( &ctx.sha512, vhash );
mm256_reinterleave_4x64_4x32( vhashA, vhash, dataLen<<3 );
mm256_rintrlv_4x64_4x32( vhashA, vhash, dataLen<<3 );
haval256_5_4way_init( &ctx.haval );
haval256_5_4way( &ctx.haval, vhashA, dataLen );
haval256_5_4way_close( &ctx.haval, vhashA );
mm256_reinterleave_4x32_4x64( vhash, vhashA, dataLen<<3 );
mm256_rintrlv_4x32_4x64( vhash, vhashA, dataLen<<3 );
memset( &vhash[ 4<<2 ], 0, (dataLen-32) << 2 );
@@ -207,7 +207,7 @@ void xevan_4way_hash( void *output, const void *input )
bmw512_4way( &ctx.bmw, vhash, dataLen );
bmw512_4way_close( &ctx.bmw, vhash );
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash0, (char*)hash0,
@@ -222,7 +222,7 @@ void xevan_4way_hash( void *output, const void *input )
update_and_final_groestl( &ctx.groestl, (char*)hash3, (char*)hash3,
dataLen<<3 );
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
skein512_4way_init( &ctx.skein );
skein512_4way( &ctx.skein, vhash, dataLen );
@@ -236,7 +236,7 @@ void xevan_4way_hash( void *output, const void *input )
keccak512_4way( &ctx.keccak, vhash, dataLen );
keccak512_4way_close( &ctx.keccak, vhash );
mm256_reinterleave_4x64_2x128( vhashA, vhashB, vhash, dataLen<<3 );
mm256_rintrlv_4x64_2x128( vhashA, vhashB, vhash, dataLen<<3 );
luffa_2way_init( &ctx.luffa, 512 );
luffa_2way_update_close( &ctx.luffa, vhashA, vhashA, dataLen );
@@ -258,8 +258,8 @@ void xevan_4way_hash( void *output, const void *input )
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhashB, vhashB, dataLen<<3 );
mm256_deinterleave_1x128( hash0, hash1, vhashA, dataLen<<3 );
mm256_deinterleave_1x128( hash2, hash3, vhashB, dataLen<<3 );
mm256_dintrlv_2x128( hash0, hash1, vhashA, dataLen<<3 );
mm256_dintrlv_2x128( hash2, hash3, vhashB, dataLen<<3 );
init_echo( &ctx.echo, 512 );
update_final_echo( &ctx.echo, (BitSequence *)hash0,
@@ -274,13 +274,13 @@ void xevan_4way_hash( void *output, const void *input )
update_final_echo( &ctx.echo, (BitSequence *)hash3,
(const BitSequence *) hash3, dataLen<<3 );
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
hamsi512_4way_init( &ctx.hamsi );
hamsi512_4way( &ctx.hamsi, vhash, dataLen );
hamsi512_4way_close( &ctx.hamsi, vhash );
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, hash0, dataLen );
@@ -295,13 +295,13 @@ void xevan_4way_hash( void *output, const void *input )
sph_fugue512( &ctx.fugue, hash3, dataLen );
sph_fugue512_close( &ctx.fugue, hash3 );
mm128_interleave_4x32( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm128_intrlv_4x32( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
shabal512_4way_init( &ctx.shabal );
shabal512_4way( &ctx.shabal, vhash, dataLen );
shabal512_4way_close( &ctx.shabal, vhash );
mm128_deinterleave_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
mm128_dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, dataLen<<3 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, hash0, dataLen );
@@ -316,13 +316,13 @@ void xevan_4way_hash( void *output, const void *input )
sph_whirlpool( &ctx.whirlpool, hash3, dataLen );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, dataLen<<3 );
sha512_4way_init( &ctx.sha512 );
sha512_4way( &ctx.sha512, vhash, dataLen );
sha512_4way_close( &ctx.sha512, vhash );
mm256_reinterleave_4x64_4x32( vhashA, vhash, dataLen<<3 );
mm256_rintrlv_4x64_4x32( vhashA, vhash, dataLen<<3 );
haval256_5_4way_init( &ctx.haval );
haval256_5_4way( &ctx.haval, vhashA, dataLen );
@@ -336,7 +336,6 @@ int scanhash_xevan_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint32_t *hash7 = &(hash[7<<2]);
uint32_t lane_hash[8];
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t _ALIGN(64) endiandata[20];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
/* int */ thr_id = mythr->id; // thr_id arg is deprecated
@@ -349,15 +348,9 @@ int scanhash_xevan_4way( int thr_id, struct work *work, uint32_t max_nonce,
if ( opt_benchmark )
ptarget[7] = 0x0cff;
uint64_t *edata = (uint64_t*)endiandata;
casti_m256i( edata, 0 ) = mm256_bswap_32( casti_m256i( pdata, 0 ) );
casti_m256i( edata, 1 ) = mm256_bswap_32( casti_m256i( pdata, 1 ) );
casti_m128i( edata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
mm256_interleave_4x64( (uint64_t*)vdata, edata, edata, edata, edata, 640 );
mm256_bswap_intrlv80_4x64( vdata, pdata );
do {
*noncev = mm256_interleave_blend_32( mm256_bswap_32(
*noncev = mm256_intrlv_blend_32( mm256_bswap_32(
_mm256_set_epi32( n+3, 0,n+2, 0,n+1, 0, n, 0 ) ), *noncev );
xevan_4way_hash( hash, vdata );
@@ -365,17 +358,10 @@ int scanhash_xevan_4way( int thr_id, struct work *work, uint32_t max_nonce,
if ( hash7[ lane ] <= Htarg )
{
mm128_extract_lane_4x32( lane_hash, hash, lane, 256 );
if ( fulltest( lane_hash, ptarget ) )
if ( fulltest( lane_hash, ptarget ) && !opt_benchmark )
{
pdata[19] = n + lane;
work_set_target_ratio( work, lane_hash );
if ( submit_work( mythr, work ) )
applog( LOG_NOTICE,
"Share %d submitted by thread %d, lane %d.",
accepted_share_count + rejected_share_count + 1,
thr_id, lane );
else
applog( LOG_WARNING, "Failed to submit share." );
submit_solution( work, lane_hash, mythr, lane );
}
}
n += 4;