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

v3.10.2

Browse Source
This commit is contained in:
Jay D Dee
2019-12-09 15:59:02 -05:00
parent 73430b13b1
commit a17ff6f189
48 changed files with 3561 additions and 1367 deletions
Download Patch File Download Diff File Expand all files Collapse all files

186
algo/x13/phi1612-4way.c
View File

@@ -1,7 +1,4 @@
#include "phi1612-gate.h"
#if defined(PHI1612_4WAY)
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
@@ -9,10 +6,193 @@
#include "algo/skein/skein-hash-4way.h"
#include "algo/jh/jh-hash-4way.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/cubehash/cube-hash-2way.h"
#include "algo/fugue/sph_fugue.h"
#include "algo/gost/sph_gost.h"
#include "algo/echo/aes_ni/hash_api.h"
#if defined(PHI1612_8WAY)
typedef struct {
skein512_8way_context skein;
jh512_8way_context jh;
cube_4way_context cube;
sph_fugue512_context fugue;
sph_gost512_context gost;
hashState_echo echo;
} phi1612_8way_ctx_holder;
phi1612_8way_ctx_holder phi1612_8way_ctx __attribute__ ((aligned (64)));
void init_phi1612_8way_ctx()
{
skein512_8way_init( &phi1612_8way_ctx.skein );
jh512_8way_init( &phi1612_8way_ctx.jh );
cube_4way_init( &phi1612_8way_ctx.cube, 512, 16, 32 );
sph_fugue512_init( &phi1612_8way_ctx.fugue );
sph_gost512_init( &phi1612_8way_ctx.gost );
init_echo( &phi1612_8way_ctx.echo, 512 );
};
void phi1612_8way_hash( void *state, const void *input )
{
uint64_t vhash[8*8] __attribute__ ((aligned (128)));
uint64_t hash0[8] __attribute__ ((aligned (64)));
uint64_t hash1[8] __attribute__ ((aligned (64)));
uint64_t hash2[8] __attribute__ ((aligned (64)));
uint64_t hash3[8] __attribute__ ((aligned (64)));
uint64_t hash4[8] __attribute__ ((aligned (64)));
uint64_t hash5[8] __attribute__ ((aligned (64)));
uint64_t hash6[8] __attribute__ ((aligned (64)));
uint64_t hash7[8] __attribute__ ((aligned (64)));
phi1612_8way_ctx_holder ctx;
memcpy( &ctx, &phi1612_8way_ctx, sizeof(phi1612_8way_ctx) );
// Skein parallel 4way
skein512_8way_update( &ctx.skein, input, 80 );
skein512_8way_close( &ctx.skein, vhash );
// JH
jh512_8way_update( &ctx.jh, vhash, 64 );
jh512_8way_close( &ctx.jh, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6, hash7,
vhash );
// Cubehash
intrlv_4x128_512( vhash, hash0, hash1, hash2, hash3 );
cube_4way_update_close( &ctx.cube, vhash, vhash, 64 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128_512( vhash, hash4, hash5, hash6, hash7 );
cube_4way_init( &ctx.cube, 512, 16, 32 );
cube_4way_update_close( &ctx.cube, vhash, vhash, 64 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
// Fugue
sph_fugue512( &ctx.fugue, hash0, 64 );
sph_fugue512_close( &ctx.fugue, hash0 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, hash1, 64 );
sph_fugue512_close( &ctx.fugue, hash1 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, hash2, 64 );
sph_fugue512_close( &ctx.fugue, hash2 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, hash3, 64 );
sph_fugue512_close( &ctx.fugue, hash3 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, hash4, 64 );
sph_fugue512_close( &ctx.fugue, hash4 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, hash5, 64 );
sph_fugue512_close( &ctx.fugue, hash5 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, hash6, 64 );
sph_fugue512_close( &ctx.fugue, hash6 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, hash7, 64 );
sph_fugue512_close( &ctx.fugue, hash7 );
// Gost
sph_gost512( &ctx.gost, hash0, 64 );
sph_gost512_close( &ctx.gost, hash0 );
sph_gost512_init( &ctx.gost );
sph_gost512( &ctx.gost, hash1, 64 );
sph_gost512_close( &ctx.gost, hash1 );
sph_gost512_init( &ctx.gost );
sph_gost512( &ctx.gost, hash2, 64 );
sph_gost512_close( &ctx.gost, hash2 );
sph_gost512_init( &ctx.gost );
sph_gost512( &ctx.gost, hash3, 64 );
sph_gost512_close( &ctx.gost, hash3 );
sph_gost512_init( &ctx.gost );
sph_gost512( &ctx.gost, hash4, 64 );
sph_gost512_close( &ctx.gost, hash4 );
sph_gost512_init( &ctx.gost );
sph_gost512( &ctx.gost, hash5, 64 );
sph_gost512_close( &ctx.gost, hash5 );
sph_gost512_init( &ctx.gost );
sph_gost512( &ctx.gost, hash6, 64 );
sph_gost512_close( &ctx.gost, hash6 );
sph_gost512_init( &ctx.gost );
sph_gost512( &ctx.gost, hash7, 64 );
sph_gost512_close( &ctx.gost, hash7 );
// Echo
update_final_echo( &ctx.echo, (BitSequence *)hash0,
(const BitSequence *) hash0, 512 );
init_echo( &ctx.echo, 512 );
update_final_echo( &ctx.echo, (BitSequence *)hash1,
(const BitSequence *) hash1, 512 );
init_echo( &ctx.echo, 512 );
update_final_echo( &ctx.echo, (BitSequence *)hash2,
(const BitSequence *) hash2, 512 );
init_echo( &ctx.echo, 512 );
update_final_echo( &ctx.echo, (BitSequence *)hash3,
(const BitSequence *) hash3, 512 );
init_echo( &ctx.echo, 512 );
update_final_echo( &ctx.echo, (BitSequence *)hash4,
(const BitSequence *) hash4, 512 );
init_echo( &ctx.echo, 512 );
update_final_echo( &ctx.echo, (BitSequence *)hash5,
(const BitSequence *) hash5, 512 );
init_echo( &ctx.echo, 512 );
update_final_echo( &ctx.echo, (BitSequence *)hash6,
(const BitSequence *) hash6, 512 );
init_echo( &ctx.echo, 512 );
update_final_echo( &ctx.echo, (BitSequence *)hash7,
(const BitSequence *) hash7, 512 );
memcpy( state, hash0, 32 );
memcpy( state+ 32, hash1, 32 );
memcpy( state+ 64, hash2, 32 );
memcpy( state+ 96, hash3, 32 );
memcpy( state+128, hash4, 32 );
memcpy( state+160, hash5, 32 );
memcpy( state+192, hash6, 32 );
memcpy( state+224, hash7, 32 );
}
int scanhash_phi1612_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t hash[8*8] __attribute__ ((aligned (128)));
uint32_t vdata[24*8] __attribute__ ((aligned (64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
uint32_t n = first_nonce;
__m512i *noncev = (__m512i*)vdata + 9; // aligned
int thr_id = mythr->id;
const uint32_t Htarg = ptarget[7];
if ( opt_benchmark )
( (uint32_t*)ptarget )[7] = 0x0cff;
mm512_bswap32_intrlv80_8x64( vdata, pdata );
do {
*noncev = mm512_intrlv_blend_32( mm512_bswap_32(
_mm512_set_epi32( n+7, 0, n+6, 0, n+5, 0, n+4, 0,
n+3, 0, n+2, 0, n+1, 0, n, 0 ) ), *noncev );
phi1612_8way_hash( hash, vdata );
pdata[19] = n;
for ( int i = 0; i < 8; i++ )
if ( (hash+(i<<3))[7] <= Htarg )
if ( fulltest( hash+(i<<3), ptarget ) && !opt_benchmark )
{
pdata[19] = n+i;
submit_lane_solution( work, hash+(i<<3), mythr, i );
}
n += 8;
} while ( ( n < max_nonce-8 ) && !work_restart[thr_id].restart );
*hashes_done = n - first_nonce;
return 0;
}
#elif defined(PHI1612_4WAY)
typedef struct {
skein512_4way_context skein;
jh512_4way_context jh;