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

Browse Source
This commit is contained in:
Jay D Dee
2019-06-26 14:16:01 -04:00
parent d6e8d7a46e
commit 0d48d573ce
174 changed files with 1352 additions and 1556 deletions
Download Patch File Download Diff File Expand all files Collapse all files

68
algo/x12/x12-4way.c
View File

@@ -74,7 +74,7 @@ void x12_4way_hash( void *state, const void *input )
bmw512_4way_close( &ctx.bmw, vhash );
// Serial
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
mm256_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 x12_4way_hash( void *state, const void *input )
update_and_final_groestl( &ctx.groestl, (char*)hash3, (char*)hash3, 512 );
// Parallel 4way 64 bit
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
// 4 Skein
skein512_4way( &ctx.skein, vhash, 64 );
@@ -101,16 +101,16 @@ void x12_4way_hash( void *state, const void *input )
keccak512_4way_close( &ctx.keccak, vhash );
// Serial
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
mm256_dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
// 7 Luffa
mm256_interleave_2x128( vhash, hash0, hash1, 512 );
mm256_intrlv_2x128( vhash, hash0, hash1, 512 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
mm256_deinterleave_2x128( hash0, hash1, vhash, 512 );
mm256_interleave_2x128( vhash, hash2, hash3, 512 );
mm256_dintrlv_2x128( hash0, hash1, vhash, 512 );
mm256_intrlv_2x128( vhash, hash2, hash3, 512 );
luffa_2way_init( &ctx.luffa, 512 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash, 64 );
mm256_deinterleave_2x128( hash2, hash3, vhash, 512 );
mm256_dintrlv_2x128( hash2, hash3, vhash, 512 );
// 8 Cubehash
cubehashUpdateDigest( &ctx.cube, (byte*)hash0, (const byte*) hash0, 64 );
@@ -138,13 +138,13 @@ void x12_4way_hash( void *state, const void *input )
sph_shavite512_close( &ctx.shavite, hash3 );
// 10 Simd
mm256_interleave_2x128( vhash, hash0, hash1, 512 );
mm256_intrlv_2x128( vhash, hash0, hash1, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
mm256_deinterleave_2x128( hash0, hash1, vhash, 512 );
mm256_interleave_2x128( vhash, hash2, hash3, 512 );
mm256_dintrlv_2x128( hash0, hash1, vhash, 512 );
mm256_intrlv_2x128( vhash, hash2, hash3, 512 );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, 512 );
mm256_deinterleave_2x128( hash2, hash3, vhash, 512 );
mm256_dintrlv_2x128( hash2, hash3, vhash, 512 );
// 11 Echo
update_final_echo( &ctx.echo, (BitSequence *)hash0,
@@ -160,36 +160,15 @@ void x12_4way_hash( void *state, const void *input )
(const BitSequence *) hash3, 512 );
// 12 Hamsi parallel 4way 32 bit
mm256_interleave_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
mm256_intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
hamsi512_4way( &ctx.hamsi, vhash, 64 );
hamsi512_4way_close( &ctx.hamsi, vhash );
mm256_deinterleave_4x64( state, state+32, state+64, state+96, vhash, 256 );
/*
// 13 Fugue serial
sph_fugue512( &ctx.fugue, hash0, 64 );
sph_fugue512_close( &ctx.fugue, hash0 );
memcpy( &ctx.fugue, &x13_4way_ctx.fugue, sizeof(sph_fugue512_context) );
sph_fugue512( &ctx.fugue, hash1, 64 );
sph_fugue512_close( &ctx.fugue, hash1 );
memcpy( &ctx.fugue, &x13_4way_ctx.fugue, sizeof(sph_fugue512_context) );
sph_fugue512( &ctx.fugue, hash2, 64 );
sph_fugue512_close( &ctx.fugue, hash2 );
memcpy( &ctx.fugue, &x13_4way_ctx.fugue, sizeof(sph_fugue512_context) );
sph_fugue512( &ctx.fugue, hash3, 64 );
sph_fugue512_close( &ctx.fugue, hash3 );
memcpy( state, hash0, 32 );
memcpy( state+32, hash1, 32 );
memcpy( state+64, hash2, 32 );
memcpy( state+96, hash3, 32 );
*/
mm256_dintrlv_4x64( state, state+32, state+64, state+96, vhash, 256 );
}
int scanhash_x12_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done )
int scanhash_x12_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
@@ -198,9 +177,8 @@ int scanhash_x12_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint32_t *ptarget = work->target;
uint32_t n = pdata[19];
const uint32_t first_nonce = pdata[19];
uint32_t *nonces = work->nonces;
int num_found = 0;
uint32_t *noncep = vdata + 73; // 9*8 + 1
int thr_id = mythr->id; // thr_id arg is deprecated
const uint32_t Htarg = ptarget[7];
uint64_t htmax[] = { 0, 0xF, 0xFF,
0xFFF, 0xFFFF, 0x10000000 };
@@ -211,7 +189,7 @@ int scanhash_x12_4way( int thr_id, struct work *work, uint32_t max_nonce,
swab32_array( endiandata, pdata, 20 );
uint64_t *edata = (uint64_t*)endiandata;
mm256_interleave_4x64( (uint64_t*)vdata, edata, edata, edata, edata, 640 );
mm256_intrlv_4x64( (uint64_t*)vdata, edata, edata, edata, edata, 640 );
for ( int m=0; m < 6; m++ )
if ( Htarg <= htmax[m] )
@@ -228,21 +206,19 @@ int scanhash_x12_4way( int thr_id, struct work *work, uint32_t max_nonce,
pdata[19] = n;
for ( int i = 0; i < 4; i++ )
if ( ( ( (hash+(i<<3))[7] & mask ) == 0 )
&& fulltest( hash+(i<<3), ptarget ) )
if ( ( (hash+(i<<3))[7] & mask ) == 0 )
if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
{
pdata[19] = n+i;
nonces[ num_found++ ] = n+i;
work_set_target_ratio( work, hash+(i<<3) );
submit_lane_solution( work, hash+(i<<3), mythr, i );
}
n += 4;
} while ( ( num_found == 0 ) && ( n < max_nonce )
&& !work_restart[thr_id].restart );
} while ( ( n < max_nonce ) && !work_restart[thr_id].restart );
break;
}
*hashes_done = n - first_nonce + 1;
return num_found;
return 0;
}
#endif

8
algo/x12/x12-gate.h
View File

@@ -14,8 +14,8 @@ bool register_x12_algo( algo_gate_t* gate );
void x12_4way_hash( void *state, const void *input );
int scanhash_x12_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done );
int scanhash_x12_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
void init_x12_4way_ctx();
@@ -23,8 +23,8 @@ void init_x12_4way_ctx();
void x12hash( void *state, const void *input );
int scanhash_x12( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done );
int scanhash_x12( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
void init_x12_ctx();

5
algo/x12/x12.c
View File

@@ -164,8 +164,8 @@ void x12hash(void *output, const void *input)
memcpy(output, hashB, 32);
}
int scanhash_x12(int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done)
int scanhash_x12( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t endiandata[20] __attribute__((aligned(64)));
uint32_t hash64[8] __attribute__((aligned(64)));
@@ -173,6 +173,7 @@ int scanhash_x12(int thr_id, struct work *work, uint32_t max_nonce,
uint32_t *ptarget = work->target;
uint32_t n = pdata[19] - 1;
const uint32_t first_nonce = pdata[19];
int thr_id = mythr->id; // thr_id arg is deprecated
const uint32_t Htarg = ptarget[7];
uint64_t htmax[] = {