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

v3.11.0

Browse Source
This commit is contained in:
Jay D Dee
2020-01-02 23:54:08 -05:00
parent 241bc26767
commit 3572cb53c4
118 changed files with 7030 additions and 1575 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
algo/x14/polytimos-4way.c
View File

@@ -34,14 +34,14 @@ void polytimos_4way_hash( void *output, const void *input )
poly_4way_context_overlay ctx;
skein512_4way_init( &ctx.skein );
skein512_4way( &ctx.skein, input, 80 );
skein512_4way_update( &ctx.skein, input, 80 );
skein512_4way_close( &ctx.skein, vhash );
// Need to convert from 64 bit interleaved to 32 bit interleaved.
uint32_t vhash32[16*4];
rintrlv_4x64_4x32( vhash32, vhash, 512 );
shabal512_4way_init( &ctx.shabal );
shabal512_4way( &ctx.shabal, vhash32, 64 );
shabal512_4way_update( &ctx.shabal, vhash32, 64 );
shabal512_4way_close( &ctx.shabal, vhash32 );
dintrlv_4x32( hash0, hash1, hash2, hash3, vhash32, 512 );

4
algo/x14/veltor-4way.c
View File

@@ -38,7 +38,7 @@ void veltor_4way_hash( void *output, const void *input )
veltor_4way_ctx_holder ctx __attribute__ ((aligned (64)));
memcpy( &ctx, &veltor_4way_ctx, sizeof(veltor_4way_ctx) );
skein512_4way( &ctx.skein, input, 80 );
skein512_4way_update( &ctx.skein, input, 80 );
skein512_4way_close( &ctx.skein, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
@@ -55,7 +55,7 @@ void veltor_4way_hash( void *output, const void *input )
sph_shavite512_close( &ctx.shavite, hash3 );
intrlv_4x32( vhash, hash0, hash1, hash2, hash3, 512 );
shabal512_4way( &ctx.shabal, vhash, 64 );
shabal512_4way_update( &ctx.shabal, vhash, 64 );
shabal512_4way_close( &ctx.shabal, vhash );
dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 512 );

116
algo/x14/x14-4way.c
View File

@@ -19,24 +19,35 @@
#include "algo/hamsi/hamsi-hash-4way.h"
#include "algo/fugue/sph_fugue.h"
#include "algo/shabal/shabal-hash-4way.h"
#if defined(__VAES__)
#include "algo/groestl/groestl512-hash-4way.h"
#include "algo/shavite/shavite-hash-4way.h"
#include "algo/echo/echo-hash-4way.h"
#endif
#if defined(X14_8WAY)
typedef struct {
blake512_8way_context blake;
bmw512_8way_context bmw;
hashState_groestl groestl;
skein512_8way_context skein;
jh512_8way_context jh;
keccak512_8way_context keccak;
luffa_4way_context luffa;
cube_4way_context cube;
sph_shavite512_context shavite;
simd_4way_context simd;
hashState_echo echo;
hamsi512_8way_context hamsi;
sph_fugue512_context fugue;
shabal512_8way_context shabal;
#if defined(__VAES__)
groestl512_4way_context groestl;
shavite512_4way_context shavite;
echo_4way_context echo;
#else
hashState_groestl groestl;
sph_shavite512_context shavite;
hashState_echo echo;
#endif
} x14_8way_ctx_holder;
x14_8way_ctx_holder x14_8way_ctx __attribute__ ((aligned (64)));
@@ -45,25 +56,31 @@ void init_x14_8way_ctx()
{
blake512_8way_init( &x14_8way_ctx.blake );
bmw512_8way_init( &x14_8way_ctx.bmw );
init_groestl( &x14_8way_ctx.groestl, 64 );
skein512_8way_init( &x14_8way_ctx.skein );
jh512_8way_init( &x14_8way_ctx.jh );
keccak512_8way_init( &x14_8way_ctx.keccak );
luffa_4way_init( &x14_8way_ctx.luffa, 512 );
cube_4way_init( &x14_8way_ctx.cube, 512, 16, 32 );
sph_shavite512_init( &x14_8way_ctx.shavite );
simd_4way_init( &x14_8way_ctx.simd, 512 );
init_echo( &x14_8way_ctx.echo, 512 );
hamsi512_8way_init( &x14_8way_ctx.hamsi );
sph_fugue512_init( &x14_8way_ctx.fugue );
shabal512_8way_init( &x14_8way_ctx.shabal );
#if defined(__VAES__)
groestl512_4way_init( &x14_8way_ctx.groestl, 64 );
shavite512_4way_init( &x14_8way_ctx.shavite );
echo_4way_init( &x14_8way_ctx.echo, 512 );
#else
init_groestl( &x14_8way_ctx.groestl, 64 );
sph_shavite512_init( &x14_8way_ctx.shavite );
init_echo( &x14_8way_ctx.echo, 512 );
#endif
};
void x14_8way_hash( void *state, const void *input )
{
uint64_t vhash[8*8] __attribute__ ((aligned (128)));
uint64_t vhash0[4*8] __attribute__ ((aligned (64)));
uint64_t vhash1[4*8] __attribute__ ((aligned (64)));
uint64_t vhashA[4*8] __attribute__ ((aligned (64)));
uint64_t vhashB[4*8] __attribute__ ((aligned (64)));
uint64_t hash0[8] __attribute__ ((aligned (64)));
uint64_t hash1[8] __attribute__ ((aligned (64)));
uint64_t hash2[8] __attribute__ ((aligned (64)));
@@ -80,6 +97,19 @@ void x14_8way_hash( void *state, const void *input )
bmw512_8way_update( &ctx.bmw, vhash, 64 );
bmw512_8way_close( &ctx.bmw, vhash );
#if defined(__VAES__)
rintrlv_8x64_4x128( vhashA, vhashB, vhash, 512 );
groestl512_4way_update_close( &ctx.groestl, vhashA, vhashA, 512 );
groestl512_4way_init( &ctx.groestl, 64 );
groestl512_4way_update_close( &ctx.groestl, vhashB, vhashB, 512 );
rintrlv_4x128_8x64( vhash, vhashA, vhashB, 512 );
#else
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6, hash7,
vhash );
@@ -108,6 +138,9 @@ void x14_8way_hash( void *state, const void *input )
intrlv_8x64_512( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7 );
#endif
skein512_8way_update( &ctx.skein, vhash, 64 );
skein512_8way_close( &ctx.skein, vhash );
@@ -117,18 +150,26 @@ void x14_8way_hash( void *state, const void *input )
keccak512_8way_update( &ctx.keccak, vhash, 64 );
keccak512_8way_close( &ctx.keccak, vhash );
rintrlv_8x64_4x128( vhash0, vhash1, vhash, 512 );
rintrlv_8x64_4x128( vhashA, vhashB, vhash, 512 );
luffa_4way_update_close( &ctx.luffa, vhash0, vhash0, 64 );
luffa_4way_update_close( &ctx.luffa, vhashA, vhashA, 64 );
luffa_4way_init( &ctx.luffa, 512 );
luffa_4way_update_close( &ctx.luffa, vhash1, vhash1, 64 );
luffa_4way_update_close( &ctx.luffa, vhashB, vhashB, 64 );
cube_4way_update_close( &ctx.cube, vhash0, vhash0, 64 );
cube_4way_update_close( &ctx.cube, vhashA, vhashA, 64 );
cube_4way_init( &ctx.cube, 512, 16, 32 );
cube_4way_update_close( &ctx.cube, vhash1, vhash1, 64 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash0 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash1 );
cube_4way_update_close( &ctx.cube, vhashB, vhashB, 64 );
#if defined(__VAES__)
shavite512_4way_update_close( &ctx.shavite, vhashA, vhashA, 64 );
shavite512_4way_init( &ctx.shavite );
shavite512_4way_update_close( &ctx.shavite, vhashB, vhashB, 64 );
#else
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhashA );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhashB );
sph_shavite512( &ctx.shavite, hash0, 64 );
sph_shavite512_close( &ctx.shavite, hash0 );
@@ -161,14 +202,28 @@ void x14_8way_hash( void *state, const void *input )
sph_shavite512( &ctx.shavite, hash7, 64 );
sph_shavite512_close( &ctx.shavite, hash7 );
intrlv_4x128_512( vhash, hash0, hash1, hash2, hash3 );
simd_4way_update_close( &ctx.simd, vhash, vhash, 512 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128_512( vhash, hash4, hash5, hash6, hash7 );
simd_4way_init( &ctx.simd, 512 );
simd_4way_update_close( &ctx.simd, vhash, vhash, 512 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
intrlv_4x128_512( vhashA, hash0, hash1, hash2, hash3 );
intrlv_4x128_512( vhashB, hash4, hash5, hash6, hash7 );
#endif
simd_4way_update_close( &ctx.simd, vhashA, vhashA, 512 );
simd_4way_init( &ctx.simd, 512 );
simd_4way_update_close( &ctx.simd, vhashB, vhashB, 512 );
#if defined(__VAES__)
echo_4way_update_close( &ctx.echo, vhashA, vhashA, 512 );
echo_4way_init( &ctx.echo, 512 );
echo_4way_update_close( &ctx.echo, vhashB, vhashB, 512 );
rintrlv_4x128_8x64( vhash, vhashA, vhashB, 512 );
#else
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhashA );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhashB );
update_final_echo( &ctx.echo, (BitSequence *)hash0,
(const BitSequence *) hash0, 512 );
memcpy( &ctx.echo, &x14_8way_ctx.echo, sizeof(hashState_echo) );
@@ -195,6 +250,9 @@ void x14_8way_hash( void *state, const void *input )
intrlv_8x64_512( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7 );
#endif
hamsi512_8way_update( &ctx.hamsi, vhash, 64 );
hamsi512_8way_close( &ctx.hamsi, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6, hash7,
@@ -325,11 +383,11 @@ void x14_4way_hash( void *state, const void *input )
memcpy( &ctx, &x14_4way_ctx, sizeof(x14_4way_ctx) );
// 1 Blake
blake512_4way( &ctx.blake, input, 80 );
blake512_4way_update( &ctx.blake, input, 80 );
blake512_4way_close( &ctx.blake, vhash );
// 2 Bmw
bmw512_4way( &ctx.bmw, vhash, 64 );
bmw512_4way_update( &ctx.bmw, vhash, 64 );
bmw512_4way_close( &ctx.bmw, vhash );
// Serial
@@ -348,15 +406,15 @@ void x14_4way_hash( void *state, const void *input )
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
// 4 Skein
skein512_4way( &ctx.skein, vhash, 64 );
skein512_4way_update( &ctx.skein, vhash, 64 );
skein512_4way_close( &ctx.skein, vhash );
// 5 JH
jh512_4way( &ctx.jh, vhash, 64 );
jh512_4way_update( &ctx.jh, vhash, 64 );
jh512_4way_close( &ctx.jh, vhash );
// 6 Keccak
keccak512_4way( &ctx.keccak, vhash, 64 );
keccak512_4way_update( &ctx.keccak, vhash, 64 );
keccak512_4way_close( &ctx.keccak, vhash );
// Serial
@@ -420,7 +478,7 @@ void x14_4way_hash( void *state, const void *input )
// 12 Hamsi parallel 4way 32 bit
intrlv_4x64( vhash, hash0, hash1, hash2, hash3, 512 );
hamsi512_4way( &ctx.hamsi, vhash, 64 );
hamsi512_4way_update( &ctx.hamsi, vhash, 64 );
hamsi512_4way_close( &ctx.hamsi, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );

2
algo/x14/x14-gate.c
View File

@@ -15,7 +15,7 @@ bool register_x14_algo( algo_gate_t* gate )
gate->scanhash = (void*)&scanhash_x14;
gate->hash = (void*)&x14hash;
#endif
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT;
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT | VAES_OPT;
return true;
};