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

v3.11.9

Browse Source
This commit is contained in:
Jay D Dee
2020-02-04 01:31:59 -05:00
parent 0681ca996d
commit 1b76cee239
106 changed files with 1695 additions and 4481 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
algo/x16/hex.c
View File

@@ -85,13 +85,6 @@ void hex_hash( void* output, const void* input )
memcpy( &ctx, &hex_ctx, sizeof(ctx) );
void *in = (void*) input;
int size = 80;
/*
if ( s_ntime == UINT32_MAX )
{
const uint8_t* in8 = (uint8_t*) input;
x16_r_s_getAlgoString( &in8[4], hashOrder );
}
*/
char elem = hashOrder[0];
uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
@@ -249,12 +242,8 @@ int scanhash_hex( struct work *work, uint32_t max_nonce,
const bool bench = opt_benchmark;
if ( bench ) ptarget[7] = 0x0cff;
casti_m128i( edata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
casti_m128i( edata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
casti_m128i( edata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
casti_m128i( edata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
casti_m128i( edata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
mm128_bswap32_80( edata, pdata );
uint32_t ntime = swab32(pdata[17]);
if ( s_ntime != ntime )
{
@@ -277,6 +266,10 @@ int scanhash_hex( struct work *work, uint32_t max_nonce,
sph_skein512_init( &hex_ctx.skein );
sph_skein512( &hex_ctx.skein, edata, 64 );
break;
case LUFFA:
init_luffa( &hex_ctx.luffa, 512 );
update_luffa( &hex_ctx.luffa, (const BitSequence*)edata, 64 );
break;
case CUBEHASH:
cubehashInit( &hex_ctx.cube, 512, 16, 32 );
cubehashUpdate( &hex_ctx.cube, (const byte*)edata, 64 );

384
algo/x16/x16r-4way.c
View File

@@ -2,74 +2,85 @@
* x16r algo implementation
*
* Implementation by tpruvot@github Jan 2018
* Optimized by JayDDee@github Jan 2018
* Optimized by https://github.com/JayDDee/ Jan 2018
*/
#include "x16r-gate.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "algo/blake/blake-hash-4way.h"
#include "algo/bmw/bmw-hash-4way.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#include "algo/skein/skein-hash-4way.h"
#include "algo/jh/jh-hash-4way.h"
#include "algo/keccak/keccak-hash-4way.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa-hash-2way.h"
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/cube-hash-2way.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/simd-hash-2way.h"
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/hamsi/hamsi-hash-4way.h"
#include "algo/fugue/sph_fugue.h"
#include "algo/shabal/shabal-hash-4way.h"
#include "algo/whirlpool/sph_whirlpool.h"
#include "algo/sha/sha-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
// The hash and prehash code is shared among x16r, x16s, x16rt, and x21s.
// The generic function performs the x16 hash as per the hash order
// and produces a 512 bit intermediate hash which needs to be converted
// to 256 bit final hash by a wrapper function.
#if defined (X16R_8WAY)
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
// Perform midstate prehash of hash functions with block size <= 64 bytes
// and interleave 4x64 before nonce insertion for final hash.
union _x16r_8way_context_overlay
void x16r_8way_prehash( void *vdata, void *pdata )
{
blake512_8way_context blake;
bmw512_8way_context bmw;
skein512_8way_context skein;
jh512_8way_context jh;
keccak512_8way_context keccak;
luffa_4way_context luffa;
cubehashParam cube;
// cube_4way_context cube;
simd_4way_context simd;
hamsi512_8way_context hamsi;
sph_fugue512_context fugue;
shabal512_8way_context shabal;
sph_whirlpool_context whirlpool;
sha512_8way_context sha512;
#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
} __attribute__ ((aligned (64)));
uint32_t vdata2[20*8] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
typedef union _x16r_8way_context_overlay x16r_8way_context_overlay;
const char elem = x16r_hash_order[0];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
static __thread x16r_8way_context_overlay x16r_ctx;
switch ( algo )
{
case JH:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
jh512_8way_init( &x16r_ctx.jh );
jh512_8way_update( &x16r_ctx.jh, vdata, 64 );
break;
case SKEIN:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
skein512_8way_init( &x16r_ctx.skein );
skein512_8way_update( &x16r_ctx.skein, vdata, 64 );
break;
case LUFFA:
mm128_bswap32_80( edata, pdata );
intrlv_4x128( vdata2, edata, edata, edata, edata, 640 );
luffa_4way_init( &x16r_ctx.luffa, 512 );
luffa_4way_update( &x16r_ctx.luffa, vdata2, 64 );
rintrlv_4x128_8x64( vdata, vdata2, vdata2, 640 );
break;
case CUBEHASH:
mm128_bswap32_80( edata, pdata );
cubehashInit( &x16r_ctx.cube, 512, 16, 32 );
cubehashUpdate( &x16r_ctx.cube, (const byte*)edata, 64 );
intrlv_8x64( vdata, edata, edata, edata, edata,
edata, edata, edata, edata, 640 );
break;
case HAMSI:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
hamsi512_8way_init( &x16r_ctx.hamsi );
hamsi512_8way_update( &x16r_ctx.hamsi, vdata, 64 );
break;
case SHABAL:
mm256_bswap32_intrlv80_8x32( vdata2, pdata );
shabal512_8way_init( &x16r_ctx.shabal );
shabal512_8way_update( &x16r_ctx.shabal, vdata2, 64 );
rintrlv_8x32_8x64( vdata, vdata2, 640 );
break;
case WHIRLPOOL:
mm128_bswap32_80( edata, pdata );
sph_whirlpool_init( &x16r_ctx.whirlpool );
sph_whirlpool( &x16r_ctx.whirlpool, edata, 64 );
intrlv_8x64( vdata, edata, edata, edata, edata,
edata, edata, edata, edata, 640 );
break;
default:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
}
}
void x16r_8way_hash( void* output, const void* input )
// Perform the full x16r hash and returns 512 bit intermediate hash.
// Called by wrapper hash function to optionally continue hashing and
// convert to final hash.
void x16r_8way_hash_generic( void* output, const void* input )
{
uint32_t vhash[20*8] __attribute__ ((aligned (128)));
uint32_t hash0[20] __attribute__ ((aligned (64)));
@@ -97,7 +108,7 @@ void x16r_8way_hash( void* output, const void* input )
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const char elem = x16r_hash_order[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
@@ -464,23 +475,39 @@ void x16r_8way_hash( void* output, const void* input )
size = 64;
}
memcpy( output, hash0, 32 );
memcpy( output+32, hash1, 32 );
memcpy( output+64, hash2, 32 );
memcpy( output+96, hash3, 32 );
memcpy( output+128, hash4, 32 );
memcpy( output+160, hash5, 32 );
memcpy( output+192, hash6, 32 );
memcpy( output+224, hash7, 32 );
memcpy( output, hash0, 64 );
memcpy( output+64, hash1, 64 );
memcpy( output+128, hash2, 64 );
memcpy( output+192, hash3, 64 );
memcpy( output+256, hash4, 64 );
memcpy( output+320, hash5, 64 );
memcpy( output+384, hash6, 64 );
memcpy( output+448, hash7, 64 );
}
// x16-r,-s,-rt wrapper called directly by scanhash to repackage 512 bit
// hash to 256 bit final hash.
void x16r_8way_hash( void* output, const void* input )
{
uint8_t hash[64*8] __attribute__ ((aligned (128)));
x16r_8way_hash_generic( hash, input );
memcpy( output, hash, 32 );
memcpy( output+32, hash+64, 32 );
memcpy( output+64, hash+128, 32 );
memcpy( output+96, hash+192, 32 );
memcpy( output+128, hash+256, 32 );
memcpy( output+160, hash+320, 32 );
memcpy( output+192, hash+384, 32 );
memcpy( output+224, hash+448, 32 );
}
// x16r only
int scanhash_x16r_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr)
{
uint32_t hash[16*8] __attribute__ ((aligned (128)));
uint32_t vdata[20*8] __attribute__ ((aligned (64)));
uint32_t vdata2[20*8] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t bedata1[2] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
@@ -496,66 +523,18 @@ int scanhash_x16r_8way( struct work *work, uint32_t max_nonce,
bedata1[0] = bswap_32( pdata[1] );
bedata1[1] = bswap_32( pdata[2] );
static __thread uint32_t s_ntime = UINT32_MAX;
const uint32_t ntime = bswap_32( pdata[17] );
if ( s_ntime != ntime )
{
x16_r_s_getAlgoString( (const uint8_t*)bedata1, hashOrder );
x16_r_s_getAlgoString( (const uint8_t*)bedata1, x16r_hash_order );
s_ntime = ntime;
if ( opt_debug && !thr_id )
applog( LOG_INFO, "hash order %s (%08x)", hashOrder, ntime );
applog( LOG_INFO, "hash order %s (%08x)", x16r_hash_order, ntime );
}
// Do midstate prehash on hash functions with block size <= 64 bytes.
const char elem = hashOrder[0];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case JH:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
jh512_8way_init( &x16r_ctx.jh );
jh512_8way_update( &x16r_ctx.jh, vdata, 64 );
break;
case SKEIN:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
skein512_8way_init( &x16r_ctx.skein );
skein512_8way_update( &x16r_ctx.skein, vdata, 64 );
break;
case LUFFA:
mm128_bswap32_80( edata, pdata );
intrlv_4x128( vdata2, edata, edata, edata, edata, 640 );
luffa_4way_init( &x16r_ctx.luffa, 512 );
luffa_4way_update( &x16r_ctx.luffa, vdata2, 64 );
rintrlv_4x128_8x64( vdata, vdata2, vdata2, 640 );
break;
case CUBEHASH:
mm128_bswap32_80( edata, pdata );
cubehashInit( &x16r_ctx.cube, 512, 16, 32 );
cubehashUpdate( &x16r_ctx.cube, (const byte*)edata, 64 );
intrlv_8x64( vdata, edata, edata, edata, edata,
edata, edata, edata, edata, 640 );
break;
case HAMSI:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
hamsi512_8way_init( &x16r_ctx.hamsi );
hamsi512_8way_update( &x16r_ctx.hamsi, vdata, 64 );
break;
case SHABAL:
mm256_bswap32_intrlv80_8x32( vdata2, pdata );
shabal512_8way_init( &x16r_ctx.shabal );
shabal512_8way_update( &x16r_ctx.shabal, vdata2, 64 );
rintrlv_8x32_8x64( vdata, vdata2, 640 );
break;
case WHIRLPOOL:
mm128_bswap32_80( edata, pdata );
sph_whirlpool_init( &x16r_ctx.whirlpool );
sph_whirlpool( &x16r_ctx.whirlpool, edata, 64 );
intrlv_8x64( vdata, edata, edata, edata, edata,
edata, edata, edata, edata, 640 );
break;
default:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
}
x16r_8way_prehash( vdata, pdata );
*noncev = mm512_intrlv_blend_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 );
@@ -580,34 +559,62 @@ int scanhash_x16r_8way( struct work *work, uint32_t max_nonce,
#elif defined (X16R_4WAY)
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
union _x16r_4way_context_overlay
void x16r_4way_prehash( void *vdata, void *pdata )
{
blake512_4way_context blake;
bmw512_4way_context bmw;
hashState_echo echo;
hashState_groestl groestl;
skein512_4way_context skein;
jh512_4way_context jh;
keccak512_4way_context keccak;
luffa_2way_context luffa;
hashState_luffa luffa1;
cubehashParam cube;
sph_shavite512_context shavite;
simd_2way_context simd;
hamsi512_4way_context hamsi;
sph_fugue512_context fugue;
shabal512_4way_context shabal;
sph_whirlpool_context whirlpool;
sha512_4way_context sha512;
} __attribute__ ((aligned (64)));
typedef union _x16r_4way_context_overlay x16r_4way_context_overlay;
uint32_t vdata2[20*4] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
static __thread x16r_4way_context_overlay x16r_ctx;
const char elem = x16r_hash_order[0];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
void x16r_4way_hash( void* output, const void* input )
switch ( algo )
{
case JH:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
jh512_4way_init( &x16r_ctx.jh );
jh512_4way_update( &x16r_ctx.jh, vdata, 64 );
break;
case SKEIN:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
skein512_4way_init( &x16r_ctx.skein );
skein512_4way_update( &x16r_ctx.skein, vdata, 64 );
break;
case LUFFA:
mm128_bswap32_80( edata, pdata );
intrlv_2x128( vdata2, edata, edata, 640 );
luffa_2way_init( &x16r_ctx.luffa, 512 );
luffa_2way_update( &x16r_ctx.luffa, vdata2, 64 );
rintrlv_2x128_4x64( vdata, vdata2, vdata2, 640 );
break;
case CUBEHASH:
mm128_bswap32_80( edata, pdata );
cubehashInit( &x16r_ctx.cube, 512, 16, 32 );
cubehashUpdate( &x16r_ctx.cube, (const byte*)edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
case HAMSI:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
hamsi512_4way_init( &x16r_ctx.hamsi );
hamsi512_4way_update( &x16r_ctx.hamsi, vdata, 64 );
break;
case SHABAL:
mm128_bswap32_intrlv80_4x32( vdata2, pdata );
shabal512_4way_init( &x16r_ctx.shabal );
shabal512_4way_update( &x16r_ctx.shabal, vdata2, 64 );
rintrlv_4x32_4x64( vdata, vdata2, 640 );
break;
case WHIRLPOOL:
mm128_bswap32_80( edata, pdata );
sph_whirlpool_init( &x16r_ctx.whirlpool );
sph_whirlpool( &x16r_ctx.whirlpool, edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
default:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
}
}
void x16r_4way_hash_generic( void* output, const void* input )
{
uint32_t vhash[20*4] __attribute__ ((aligned (128)));
uint32_t hash0[20] __attribute__ ((aligned (64)));
@@ -626,7 +633,7 @@ void x16r_4way_hash( void* output, const void* input )
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const char elem = x16r_hash_order[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
@@ -698,11 +705,12 @@ void x16r_4way_hash( void* output, const void* input )
case LUFFA:
if ( i == 0 )
{
intrlv_2x128( vhash, in0, in1, size<<3 );
luffa512_2way_full( &ctx.luffa, vhash, vhash + (16<<1), 16 );
intrlv_2x128( vhash, hash0, hash1, 640 );
luffa_2way_update_close( &ctx.luffa, vhash, vhash + (16<<1), 16 );
dintrlv_2x128_512( hash0, hash1, vhash );
intrlv_2x128( vhash, in2, in3, size<<3 );
luffa512_2way_full( &ctx.luffa, vhash, vhash + (16<<1), 16 );
intrlv_2x128( vhash, hash2, hash3, 640 );
memcpy( &ctx, &x16r_ctx, sizeof(ctx) );
luffa_2way_update_close( &ctx.luffa, vhash, vhash + (16<<1), 16 );
dintrlv_2x128_512( hash2, hash3, vhash );
}
else
@@ -863,10 +871,21 @@ void x16r_4way_hash( void* output, const void* input )
}
size = 64;
}
memcpy( output, hash0, 32 );
memcpy( output+32, hash1, 32 );
memcpy( output+64, hash2, 32 );
memcpy( output+96, hash3, 32 );
memcpy( output, hash0, 64 );
memcpy( output+64, hash1, 64 );
memcpy( output+128, hash2, 64 );
memcpy( output+192, hash3, 64 );
}
void x16r_4way_hash( void* output, const void* input )
{
uint8_t hash[64*4] __attribute__ ((aligned (64)));
x16r_4way_hash_generic( hash, input );
memcpy( output, hash, 32 );
memcpy( output+32, hash+64, 32 );
memcpy( output+64, hash+128, 32 );
memcpy( output+96, hash+192, 32 );
}
int scanhash_x16r_4way( struct work *work, uint32_t max_nonce,
@@ -874,8 +893,6 @@ int scanhash_x16r_4way( struct work *work, uint32_t max_nonce,
{
uint32_t hash[16*4] __attribute__ ((aligned (64)));
uint32_t vdata[20*4] __attribute__ ((aligned (64)));
uint32_t vdata2[20*4] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t bedata1[2] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
@@ -891,67 +908,20 @@ int scanhash_x16r_4way( struct work *work, uint32_t max_nonce,
bedata1[0] = bswap_32( pdata[1] );
bedata1[1] = bswap_32( pdata[2] );
static __thread uint32_t s_ntime = UINT32_MAX;
const uint32_t ntime = bswap_32( pdata[17] );
if ( s_ntime != ntime )
{
x16_r_s_getAlgoString( (const uint8_t*)bedata1, hashOrder );
x16_r_s_getAlgoString( (const uint8_t*)bedata1, x16r_hash_order );
s_ntime = ntime;
if ( opt_debug && !thr_id )
applog( LOG_INFO, "hash order %s (%08x)", hashOrder, ntime );
}
// Do midstate prehash on hash functions with block size <= 64 bytes.
const char elem = hashOrder[0];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case JH:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
jh512_4way_init( &x16r_ctx.jh );
jh512_4way_update( &x16r_ctx.jh, vdata, 64 );
break;
case SKEIN:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
skein512_4way_init( &x16r_ctx.skein );
skein512_4way_update( &x16r_ctx.skein, vdata, 64 );
break;
case LUFFA:
mm128_bswap32_80( edata, pdata );
intrlv_2x128( vdata2, edata, edata, 640 );
luffa_2way_init( &x16r_ctx.luffa, 512 );
luffa_2way_update( &x16r_ctx.luffa, vdata2, 64 );
rintrlv_2x128_4x64( vdata, vdata2, vdata2, 512 );
break;
case CUBEHASH:
mm128_bswap32_80( edata, pdata );
cubehashInit( &x16r_ctx.cube, 512, 16, 32 );
cubehashUpdate( &x16r_ctx.cube, (const byte*)edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
case HAMSI:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
hamsi512_4way_init( &x16r_ctx.hamsi );
hamsi512_4way_update( &x16r_ctx.hamsi, vdata, 64 );
break;
case SHABAL:
mm128_bswap32_intrlv80_4x32( vdata2, pdata );
shabal512_4way_init( &x16r_ctx.shabal );
shabal512_4way_update( &x16r_ctx.shabal, vdata2, 64 );
rintrlv_4x32_4x64( vdata, vdata2, 640 );
break;
case WHIRLPOOL:
mm128_bswap32_80( edata, pdata );
sph_whirlpool_init( &x16r_ctx.whirlpool );
sph_whirlpool( &x16r_ctx.whirlpool, edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
default:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
applog( LOG_INFO, "hash order %s (%08x)", x16r_hash_order, ntime );
}
x16r_4way_prehash( vdata, pdata );
*noncev = mm256_intrlv_blend_32(
_mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
do
{
x16r_4way_hash( hash, vdata );

41
algo/x16/x16r-gate.c
View File

@@ -1,7 +1,22 @@
#include "x16r-gate.h"
__thread char x16r_hash_order[ X16R_HASH_FUNC_COUNT + 1 ] = { 0 };
void (*x16_r_s_getAlgoString) ( const uint8_t*, char* ) = NULL;
#if defined (X16R_8WAY)
__thread x16r_8way_context_overlay x16r_ctx;
#elif defined (X16R_4WAY)
__thread x16r_4way_context_overlay x16r_ctx;
#endif
__thread x16r_context_overlay x16_ctx;
void x16r_getAlgoString( const uint8_t* prevblock, char *output )
{
char *sptr = output;
@@ -207,15 +222,15 @@ void veil_build_extraheader( struct work* g_work, struct stratum_ctx* sctx )
bool register_x16rt_algo( algo_gate_t* gate )
{
#if defined (X16RT_8WAY)
#if defined (X16R_8WAY)
gate->scanhash = (void*)&scanhash_x16rt_8way;
gate->hash = (void*)&x16rt_8way_hash;
#elif defined (X16RT_4WAY)
gate->hash = (void*)&x16r_8way_hash;
#elif defined (X16R_4WAY)
gate->scanhash = (void*)&scanhash_x16rt_4way;
gate->hash = (void*)&x16rt_4way_hash;
gate->hash = (void*)&x16r_4way_hash;
#else
gate->scanhash = (void*)&scanhash_x16rt;
gate->hash = (void*)&x16rt_hash;
gate->hash = (void*)&x16r_hash;
#endif
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT | VAES_OPT;
opt_target_factor = 256.0;
@@ -224,15 +239,15 @@ bool register_x16rt_algo( algo_gate_t* gate )
bool register_x16rt_veil_algo( algo_gate_t* gate )
{
#if defined (X16RT_8WAY)
#if defined (X16R_8WAY)
gate->scanhash = (void*)&scanhash_x16rt_8way;
gate->hash = (void*)&x16rt_8way_hash;
#elif defined (X16RT_4WAY)
gate->hash = (void*)&x16r_8way_hash;
#elif defined (X16R_4WAY)
gate->scanhash = (void*)&scanhash_x16rt_4way;
gate->hash = (void*)&x16rt_4way_hash;
gate->hash = (void*)&x16r_4way_hash;
#else
gate->scanhash = (void*)&scanhash_x16rt;
gate->hash = (void*)&x16rt_hash;
gate->hash = (void*)&x16r_hash;
#endif
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT | VAES_OPT;
gate->build_extraheader = (void*)&veil_build_extraheader;
@@ -247,7 +262,7 @@ bool register_x16rt_veil_algo( algo_gate_t* gate )
bool register_hex_algo( algo_gate_t* gate )
{
gate->scanhash = (void*)&scanhash_hex;
gate->hash = (void*)&hex_hash;
gate->hash = (void*)&x16r_hash;
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT;
gate->gen_merkle_root = (void*)&SHA256_gen_merkle_root;
opt_target_factor = 128.0;
@@ -260,13 +275,13 @@ bool register_hex_algo( algo_gate_t* gate )
bool register_x21s_algo( algo_gate_t* gate )
{
#if defined (X21S_8WAY)
#if defined (X16R_8WAY)
gate->scanhash = (void*)&scanhash_x21s_8way;
gate->hash = (void*)&x21s_8way_hash;
gate->miner_thread_init = (void*)&x21s_8way_thread_init;
gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT
| VAES_OPT;
#elif defined (X21S_4WAY)
#elif defined (X16R_4WAY)
gate->scanhash = (void*)&scanhash_x21s_4way;
gate->hash = (void*)&x21s_4way_hash;
gate->miner_thread_init = (void*)&x21s_4way_thread_init;

191
algo/x16/x16r-gate.h
View File

@@ -5,29 +5,60 @@
#include "simd-utils.h"
#include <stdint.h>
#include <unistd.h>
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define X16R_8WAY 1
#elif defined(__AVX2__) && defined(__AES__)
#define X16R_4WAY 1
#include "algo/blake/sph_blake.h"
#include "algo/bmw/sph_bmw.h"
#include "algo/groestl/sph_groestl.h"
#include "algo/jh/sph_jh.h"
#include "algo/keccak/sph_keccak.h"
#include "algo/skein/sph_skein.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/nist.h"
#include "algo/echo/sph_echo.h"
#include "algo/hamsi/sph_hamsi.h"
#include "algo/fugue/sph_fugue.h"
#include "algo/shabal/sph_shabal.h"
#include "algo/whirlpool/sph_whirlpool.h"
#include <openssl/sha.h>
#if defined(__AES__)
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#endif
#if defined (__AVX2__)
#include "algo/blake/blake-hash-4way.h"
#include "algo/bmw/bmw-hash-4way.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#include "algo/skein/skein-hash-4way.h"
#include "algo/jh/jh-hash-4way.h"
#include "algo/keccak/keccak-hash-4way.h"
#include "algo/luffa/luffa-hash-2way.h"
#include "algo/simd/simd-hash-2way.h"
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/hamsi/hamsi-hash-4way.h"
#include "algo/shabal/shabal-hash-4way.h"
#include "algo/sha/sha-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
#endif // AVX2
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define X16R_8WAY 1
#define X16RV2_8WAY 1
#define X16RT_8WAY 1
#define X21S_8WAY 1
#elif defined(__AVX2__) && defined(__AES__)
#define X16RV2_4WAY 1
#endif
#define X16RT_4WAY 1
#define X21S_4WAY 1
#define X16R_4WAY 1
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define X16RT_8WAY 1
#elif defined(__AVX2__) && defined(__AES__)
#define X16RT_4WAY 1
#endif
#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
#define X21S_8WAY 1
#elif defined(__AVX2__) && defined(__AES__)
#define X21S_4WAY 1
#endif
enum x16r_Algo {
@@ -50,6 +81,8 @@ enum x16r_Algo {
X16R_HASH_FUNC_COUNT
};
extern __thread char x16r_hash_order[ X16R_HASH_FUNC_COUNT + 1 ];
extern void (*x16_r_s_getAlgoString) ( const uint8_t*, char* );
void x16r_getAlgoString( const uint8_t *prevblock, char *output );
void x16s_getAlgoString( const uint8_t *prevblock, char *output );
@@ -67,25 +100,115 @@ bool register_x21s__algo( algo_gate_t* gate );
// x16r, x16s
#if defined(X16R_8WAY)
void x16r_8way_hash( void *state, const void *input );
int scanhash_x16r_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
union _x16r_8way_context_overlay
{
blake512_8way_context blake;
bmw512_8way_context bmw;
skein512_8way_context skein;
jh512_8way_context jh;
keccak512_8way_context keccak;
luffa_4way_context luffa;
cubehashParam cube;
simd_4way_context simd;
hamsi512_8way_context hamsi;
sph_fugue512_context fugue;
shabal512_8way_context shabal;
sph_whirlpool_context whirlpool;
sha512_8way_context sha512;
#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
} __attribute__ ((aligned (64)));
typedef union _x16r_8way_context_overlay x16r_8way_context_overlay;
extern __thread x16r_8way_context_overlay x16r_ctx;
void x16r_8way_prehash( void *, void * );
void x16r_8way_hash_generic( void *, const void * );
void x16r_8way_hash( void *, const void * );
int scanhash_x16r_8way( struct work *, uint32_t ,
uint64_t *, struct thr_info * );
extern __thread x16r_8way_context_overlay x16r_ctx;
#elif defined(X16R_4WAY)
void x16r_4way_hash( void *state, const void *input );
int scanhash_x16r_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
union _x16r_4way_context_overlay
{
blake512_4way_context blake;
bmw512_4way_context bmw;
hashState_echo echo;
hashState_groestl groestl;
skein512_4way_context skein;
jh512_4way_context jh;
keccak512_4way_context keccak;
luffa_2way_context luffa;
hashState_luffa luffa1;
cubehashParam cube;
sph_shavite512_context shavite;
simd_2way_context simd;
hamsi512_4way_context hamsi;
sph_fugue512_context fugue;
shabal512_4way_context shabal;
sph_whirlpool_context whirlpool;
sha512_4way_context sha512;
} __attribute__ ((aligned (64)));
#else
typedef union _x16r_4way_context_overlay x16r_4way_context_overlay;
void x16r_hash( void *state, const void *input );
int scanhash_x16r( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
extern __thread x16r_4way_context_overlay x16r_ctx;
void x16r_4way_prehash( void *, void * );
void x16r_4way_hash_generic( void *, const void * );
void x16r_4way_hash( void *, const void * );
int scanhash_x16r_4way( struct work *, uint32_t,
uint64_t *, struct thr_info * );
extern __thread x16r_4way_context_overlay x16r_ctx;
#endif
// needed for hex
union _x16r_context_overlay
{
#if defined(__AES__)
hashState_echo echo;
hashState_groestl groestl;
#else
sph_groestl512_context groestl;
sph_echo512_context echo;
#endif
sph_blake512_context blake;
sph_bmw512_context bmw;
sph_skein512_context skein;
sph_jh512_context jh;
sph_keccak512_context keccak;
hashState_luffa luffa;
cubehashParam cube;
sph_shavite512_context shavite;
hashState_sd simd;
sph_hamsi512_context hamsi;
sph_fugue512_context fugue;
sph_shabal512_context shabal;
sph_whirlpool_context whirlpool;
SHA512_CTX sha512;
} __attribute__ ((aligned (64)));
typedef union _x16r_context_overlay x16r_context_overlay;
extern __thread x16r_context_overlay x16_ctx;
void x16r_prehash( void *, void * );
void x16r_hash_generic( void *, const void * );
void x16r_hash( void *, const void * );
int scanhash_x16r( struct work *, uint32_t, uint64_t *, struct thr_info * );
// x16Rv2
#if defined(X16RV2_8WAY)
@@ -108,35 +231,35 @@ int scanhash_x16rv2( struct work *work, uint32_t max_nonce,
#endif
// x16rt, veil
#if defined(X16RT_8WAY)
#if defined(X16R_8WAY)
void x16rt_8way_hash( void *state, const void *input );
//void x16rt_8way_hash( void *state, const void *input );
int scanhash_x16rt_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#elif defined(X16RT_4WAY)
#elif defined(X16R_4WAY)
void x16rt_4way_hash( void *state, const void *input );
//void x16rt_4way_hash( void *state, const void *input );
int scanhash_x16rt_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#else
void x16rt_hash( void *state, const void *input );
//void x16rt_hash( void *state, const void *input );
int scanhash_x16rt( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
#endif
// x21s
#if defined(X21S_8WAY)
#if defined(X16R_8WAY)
void x21s_8way_hash( void *state, const void *input );
int scanhash_x21s_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );
bool x21s_8way_thread_init();
#elif defined(X21S_4WAY)
#elif defined(X16R_4WAY)
void x21s_4way_hash( void *state, const void *input );
int scanhash_x21s_4way( struct work *work, uint32_t max_nonce,
@@ -152,7 +275,7 @@ bool x21s_thread_init();
#endif
void hex_hash( void *state, const void *input );
//void hex_hash( void *state, const void *input );
int scanhash_hex( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr );

245
algo/x16/x16r.c
View File

@@ -9,72 +9,56 @@
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "algo/blake/sph_blake.h"
#include "algo/bmw/sph_bmw.h"
#include "algo/groestl/sph_groestl.h"
#include "algo/jh/sph_jh.h"
#include "algo/keccak/sph_keccak.h"
#include "algo/skein/sph_skein.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/nist.h"
#include "algo/echo/sph_echo.h"
#include "algo/hamsi/sph_hamsi.h"
#include "algo/fugue/sph_fugue.h"
#include "algo/shabal/sph_shabal.h"
#include "algo/whirlpool/sph_whirlpool.h"
#include <openssl/sha.h>
#if defined(__AES__)
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#endif
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
union _x16r_context_overlay
void x16r_prehash( void *edata, void *pdata )
{
#if defined(__AES__)
hashState_echo echo;
hashState_groestl groestl;
#else
sph_groestl512_context groestl;
sph_echo512_context echo;
#endif
sph_blake512_context blake;
sph_bmw512_context bmw;
sph_skein512_context skein;
sph_jh512_context jh;
sph_keccak512_context keccak;
hashState_luffa luffa;
cubehashParam cube;
sph_shavite512_context shavite;
hashState_sd simd;
sph_hamsi512_context hamsi;
sph_fugue512_context fugue;
sph_shabal512_context shabal;
sph_whirlpool_context whirlpool;
SHA512_CTX sha512;
};
typedef union _x16r_context_overlay x16r_context_overlay;
const char elem = x16r_hash_order[0];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
void x16r_hash( void* output, const void* input )
switch ( algo )
{
case JH:
sph_jh512_init( &x16_ctx.jh );
sph_jh512( &x16_ctx.jh, edata, 64 );
break;
case SKEIN:
sph_skein512_init( &x16_ctx.skein );
sph_skein512( &x16_ctx.skein, edata, 64 );
break;
case LUFFA:
init_luffa( &x16_ctx.luffa, 512 );
update_luffa( &x16_ctx.luffa, (const BitSequence*)edata, 64 );
break;
case CUBEHASH:
cubehashInit( &x16_ctx.cube, 512, 16, 32 );
cubehashUpdate( &x16_ctx.cube, (const byte*)edata, 64 );
break;
case HAMSI:
sph_hamsi512_init( &x16_ctx.hamsi );
sph_hamsi512( &x16_ctx.hamsi, edata, 64 );
break;
case SHABAL:
sph_shabal512_init( &x16_ctx.shabal );
sph_shabal512( &x16_ctx.shabal, edata, 64 );
break;
case WHIRLPOOL:
sph_whirlpool_init( &x16_ctx.whirlpool );
sph_whirlpool( &x16_ctx.whirlpool, edata, 64 );
break;
}
}
void x16r_hash_generic( void* output, const void* input )
{
uint32_t _ALIGN(128) hash[16];
x16r_context_overlay ctx;
memcpy( &ctx, &x16_ctx, sizeof(ctx) );
void *in = (void*) input;
int size = 80;
/*
if ( s_ntime == UINT32_MAX )
{
const uint8_t* in8 = (uint8_t*) input;
x16_r_s_getAlgoString( &in8[4], hashOrder );
}
*/
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const char elem = x16r_hash_order[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
@@ -91,23 +75,21 @@ void x16r_hash( void* output, const void* input )
break;
case GROESTL:
#if defined(__AES__)
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash,
(const char*)in, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash, (char*)in, size<<3 );
#else
sph_groestl512_init( &ctx.groestl );
sph_groestl512( &ctx.groestl, in, size );
sph_groestl512_close(&ctx.groestl, hash);
#endif
break;
case SKEIN:
sph_skein512_init( &ctx.skein );
sph_skein512( &ctx.skein, in, size );
sph_skein512_close( &ctx.skein, hash );
break;
case JH:
sph_jh512_init( &ctx.jh );
sph_jh512(&ctx.jh, in, size );
if ( i == 0 )
sph_jh512(&ctx.jh, in+64, 16 );
else
{
sph_jh512_init( &ctx.jh );
sph_jh512(&ctx.jh, in, size );
}
sph_jh512_close(&ctx.jh, hash );
break;
case KECCAK:
@@ -115,15 +97,31 @@ void x16r_hash( void* output, const void* input )
sph_keccak512( &ctx.keccak, in, size );
sph_keccak512_close( &ctx.keccak, hash );
break;
case SKEIN:
if ( i == 0 )
sph_skein512(&ctx.skein, in+64, 16 );
else
{
sph_skein512_init( &ctx.skein );
sph_skein512( &ctx.skein, in, size );
}
sph_skein512_close( &ctx.skein, hash );
break;
case LUFFA:
init_luffa( &ctx.luffa, 512 );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
(const BitSequence*)in, size );
if ( i == 0 )
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
(const BitSequence*)in+64, 16 );
else
luffa_full( &ctx.luffa, (BitSequence*)hash, 512,
(const BitSequence*)in, size );
break;
case CUBEHASH:
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash,
(const byte*)in, size );
if ( i == 0 )
cubehashUpdateDigest( &ctx.cube, (byte*)hash,
(const byte*)in+64, 16 );
else
cubehash_full( &ctx.cube, (byte*)hash, 512,
(byte*)in, size );
break;
case SHAVITE:
sph_shavite512_init( &ctx.shavite );
@@ -131,93 +129,109 @@ void x16r_hash( void* output, const void* input )
sph_shavite512_close( &ctx.shavite, hash );
break;
case SIMD:
init_sd( &ctx.simd, 512 );
update_final_sd( &ctx.simd, (BitSequence *)hash,
(const BitSequence*)in, size<<3 );
simd_full( &ctx.simd, (BitSequence *)hash,
(const BitSequence*)in, size<<3 );
break;
case ECHO:
#if defined(__AES__)
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash,
(const BitSequence*)in, size<<3 );
echo_full( &ctx.echo, hash, 512, in, size );
#else
sph_echo512_init( &ctx.echo );
sph_echo512( &ctx.echo, in, size );
sph_echo512_close( &ctx.echo, hash );
sph_echo512_init( &ctx.echo );
sph_echo512( &ctx.echo, in, size );
sph_echo512_close( &ctx.echo, hash );
#endif
break;
case HAMSI:
sph_hamsi512_init( &ctx.hamsi );
sph_hamsi512( &ctx.hamsi, in, size );
sph_hamsi512_close( &ctx.hamsi, hash );
if ( i == 0 )
sph_hamsi512( &ctx.hamsi, in+64, 16 );
else
{
sph_hamsi512_init( &ctx.hamsi );
sph_hamsi512( &ctx.hamsi, in, size );
}
sph_hamsi512_close( &ctx.hamsi, hash );
break;
case FUGUE:
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in, size );
sph_fugue512_close( &ctx.fugue, hash );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in, size );
sph_fugue512_close( &ctx.fugue, hash );
break;
case SHABAL:
sph_shabal512_init( &ctx.shabal );
sph_shabal512( &ctx.shabal, in, size );
sph_shabal512_close( &ctx.shabal, hash );
if ( i == 0 )
sph_shabal512( &ctx.shabal, in+64, 16 );
else
{
sph_shabal512_init( &ctx.shabal );
sph_shabal512( &ctx.shabal, in, size );
}
sph_shabal512_close( &ctx.shabal, hash );
break;
case WHIRLPOOL:
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in, size );
sph_whirlpool_close( &ctx.whirlpool, hash );
if ( i == 0 )
sph_whirlpool( &ctx.whirlpool, in+64, 16 );
else
{
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in, size );
}
sph_whirlpool_close( &ctx.whirlpool, hash );
break;
case SHA_512:
SHA512_Init( &ctx.sha512 );
SHA512_Update( &ctx.sha512, in, size );
SHA512_Final( (unsigned char*) hash, &ctx.sha512 );
SHA512_Init( &ctx.sha512 );
SHA512_Update( &ctx.sha512, in, size );
SHA512_Final( (unsigned char*) hash, &ctx.sha512 );
break;
}
in = (void*) hash;
size = 64;
}
memcpy(output, hash, 32);
memcpy( output, hash, 64 );
}
void x16r_hash( void* output, const void* input )
{
uint8_t hash[64] __attribute__ ((aligned (64)));
x16r_hash_generic( hash, input );
memcpy( output, hash, 32 );
}
int scanhash_x16r( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t _ALIGN(128) hash32[8];
uint32_t _ALIGN(128) endiandata[20];
uint32_t _ALIGN(128) edata[20];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t Htarg = ptarget[7];
const uint32_t first_nonce = pdata[19];
int thr_id = mythr->id; // thr_id arg is deprecated
const int thr_id = mythr->id;
uint32_t nonce = first_nonce;
volatile uint8_t *restart = &(work_restart[thr_id].restart);
volatile uint8_t *restart = &( work_restart[thr_id].restart );
const bool bench = opt_benchmark;
if ( bench ) ptarget[7] = 0x0cff;
casti_m128i( endiandata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
casti_m128i( endiandata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
casti_m128i( endiandata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
casti_m128i( endiandata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
casti_m128i( endiandata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
mm128_bswap32_80( edata, pdata );
static __thread uint32_t s_ntime = UINT32_MAX;
if ( s_ntime != pdata[17] )
{
uint32_t ntime = swab32(pdata[17]);
x16_r_s_getAlgoString( (const uint8_t*) (&endiandata[1]), hashOrder );
x16_r_s_getAlgoString( (const uint8_t*)(&edata[1]), x16r_hash_order );
s_ntime = ntime;
if ( opt_debug && !thr_id )
applog( LOG_DEBUG, "hash order %s (%08x)", hashOrder, ntime );
applog( LOG_DEBUG, "hash order %s (%08x)", x16r_hash_order, ntime );
}
if ( opt_benchmark )
ptarget[7] = 0x0cff;
x16r_prehash( edata, pdata );
do
{
be32enc( &endiandata[19], nonce );
x16r_hash( hash32, endiandata );
edata[19] = nonce;
x16r_hash( hash32, edata );
if ( hash32[7] <= Htarg )
if (fulltest( hash32, ptarget ) && !opt_benchmark )
if ( unlikely( valid_hash( hash32, ptarget ) && !bench ) )
{
pdata[19] = nonce;
pdata[19] = bswap_32( nonce );
submit_solution( work, hash32, mythr );
}
nonce++;
@@ -226,3 +240,4 @@ int scanhash_x16r( struct work *work, uint32_t max_nonce,
*hashes_done = pdata[19] - first_nonce + 1;
return 0;
}

886
algo/x16/x16rt-4way.c
View File

@@ -2,481 +2,14 @@
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "algo/blake/blake-hash-4way.h"
#include "algo/bmw/bmw-hash-4way.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#include "algo/skein/skein-hash-4way.h"
#include "algo/jh/jh-hash-4way.h"
#include "algo/keccak/keccak-hash-4way.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa-hash-2way.h"
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/cubehash/cube-hash-2way.h"
#include "algo/simd/simd-hash-2way.h"
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/hamsi/hamsi-hash-4way.h"
#include "algo/fugue/sph_fugue.h"
#include "algo/shabal/shabal-hash-4way.h"
#include "algo/whirlpool/sph_whirlpool.h"
#include "algo/sha/sha-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 (X16RT_8WAY)
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
union _x16rt_8way_context_overlay
{
blake512_8way_context blake;
bmw512_8way_context bmw;
skein512_8way_context skein;
jh512_8way_context jh;
keccak512_8way_context keccak;
luffa_4way_context luffa;
cubehashParam cube;
// cube_4way_context cube;
simd_4way_context simd;
hamsi512_8way_context hamsi;
sph_fugue512_context fugue;
shabal512_8way_context shabal;
sph_whirlpool_context whirlpool;
sha512_8way_context sha512;
#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
} __attribute__ ((aligned (64)));
typedef union _x16rt_8way_context_overlay x16rt_8way_context_overlay;
static __thread x16rt_8way_context_overlay x16rt_ctx;
void x16rt_8way_hash( void* output, const void* input )
{
uint32_t vhash[20*8] __attribute__ ((aligned (128)));
uint32_t hash0[20] __attribute__ ((aligned (64)));
uint32_t hash1[20] __attribute__ ((aligned (64)));
uint32_t hash2[20] __attribute__ ((aligned (64)));
uint32_t hash3[20] __attribute__ ((aligned (64)));
uint32_t hash4[20] __attribute__ ((aligned (64)));
uint32_t hash5[20] __attribute__ ((aligned (64)));
uint32_t hash6[20] __attribute__ ((aligned (64)));
uint32_t hash7[20] __attribute__ ((aligned (64)));
x16rt_8way_context_overlay ctx;
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
void *in0 = (void*) hash0;
void *in1 = (void*) hash1;
void *in2 = (void*) hash2;
void *in3 = (void*) hash3;
void *in4 = (void*) hash4;
void *in5 = (void*) hash5;
void *in6 = (void*) hash6;
void *in7 = (void*) hash7;
int size = 80;
dintrlv_8x64( hash0, hash1, hash2, hash3, hash4, hash5, hash6, hash7,
input, 640 );
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case BLAKE:
if ( i == 0 )
blake512_8way_full( &ctx.blake, vhash, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
blake512_8way_full( &ctx.blake, vhash, vhash, size );
}
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5,
hash6, hash7, vhash );
break;
case BMW:
bmw512_8way_init( &ctx.bmw );
if ( i == 0 )
bmw512_8way_update( &ctx.bmw, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
bmw512_8way_update( &ctx.bmw, vhash, size );
}
bmw512_8way_close( &ctx.bmw, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case GROESTL:
#if defined(__VAES__)
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
groestl512_4way_init( &ctx.groestl, 64 );
groestl512_4way_update_close( &ctx.groestl, vhash, vhash, size<<3 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
groestl512_4way_init( &ctx.groestl, 64 );
groestl512_4way_update_close( &ctx.groestl, vhash, vhash, size<<3 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
#else
groestl512_full( &ctx.groestl, (char*)hash0, (char*)in0, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash1, (char*)in1, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash2, (char*)in2, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash3, (char*)in3, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash4, (char*)in4, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash5, (char*)in5, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash6, (char*)in6, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash7, (char*)in7, size<<3 );
#endif
break;
case JH:
if ( i == 0 )
jh512_8way_update( &ctx.jh, input + (64<<3), 16 );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
jh512_8way_init( &ctx.jh );
jh512_8way_update( &ctx.jh, vhash, size );
}
jh512_8way_close( &ctx.jh, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case KECCAK:
keccak512_8way_init( &ctx.keccak );
if ( i == 0 )
keccak512_8way_update( &ctx.keccak, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
keccak512_8way_update( &ctx.keccak, vhash, size );
}
keccak512_8way_close( &ctx.keccak, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case SKEIN:
if ( i == 0 )
skein512_8way_update( &ctx.skein, input + (64<<3), 16 );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
skein512_8way_init( &ctx.skein );
skein512_8way_update( &ctx.skein, vhash, size );
}
skein512_8way_close( &ctx.skein, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case LUFFA:
if ( i == 0 )
{
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
luffa_4way_update_close( &ctx.luffa, vhash,
vhash + (16<<2), 16 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
luffa_4way_update_close( &ctx.luffa, vhash,
vhash + (16<<2), 16 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
}
else
{
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
luffa512_4way_full( &ctx.luffa, vhash, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
luffa512_4way_full( &ctx.luffa, vhash, vhash, size );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
}
break;
case CUBEHASH:
if ( i == 0 )
{
cubehashUpdateDigest( &ctx.cube, (byte*)hash0,
(const byte*)in0 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash1,
(const byte*)in1 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash2,
(const byte*)in2 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash3,
(const byte*)in3 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash4,
(const byte*)in4 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash5,
(const byte*)in5 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash6,
(const byte*)in6 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash7,
(const byte*)in7 + 64, 16 );
}
else
{
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash0,
(const byte*)in0, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash1,
(const byte*)in1, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash2,
(const byte*)in2, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash3,
(const byte*)in3, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash4,
(const byte*)in4, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash5,
(const byte*)in5, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash6,
(const byte*)in6, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash7,
(const byte*)in7, size );
}
break;
case SHAVITE:
#if defined(__VAES__)
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
shavite512_4way_full( &ctx.shavite, vhash, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
shavite512_4way_full( &ctx.shavite, vhash, vhash, size );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
#else
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in0, size );
sph_shavite512_close( &ctx.shavite, hash0 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in1, size );
sph_shavite512_close( &ctx.shavite, hash1 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in2, size );
sph_shavite512_close( &ctx.shavite, hash2 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in3, size );
sph_shavite512_close( &ctx.shavite, hash3 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in4, size );
sph_shavite512_close( &ctx.shavite, hash4 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in5, size );
sph_shavite512_close( &ctx.shavite, hash5 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in6, size );
sph_shavite512_close( &ctx.shavite, hash6 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in7, size );
sph_shavite512_close( &ctx.shavite, hash7 );
#endif
break;
case SIMD:
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
simd512_4way_full( &ctx.simd, vhash, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
simd512_4way_full( &ctx.simd, vhash, vhash, size );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
break;
case ECHO:
#if defined(__VAES__)
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
echo_4way_full( &ctx.echo, vhash, 512, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
echo_4way_full( &ctx.echo, vhash, 512, vhash, size );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
#else
echo_full( &ctx.echo, (BitSequence *)hash0, 512,
(const BitSequence *)in0, size );
echo_full( &ctx.echo, (BitSequence *)hash1, 512,
(const BitSequence *)in1, size );
echo_full( &ctx.echo, (BitSequence *)hash2, 512,
(const BitSequence *)in2, size );
echo_full( &ctx.echo, (BitSequence *)hash3, 512,
(const BitSequence *)in3, size );
echo_full( &ctx.echo, (BitSequence *)hash4, 512,
(const BitSequence *)in4, size );
echo_full( &ctx.echo, (BitSequence *)hash5, 512,
(const BitSequence *)in5, size );
echo_full( &ctx.echo, (BitSequence *)hash6, 512,
(const BitSequence *)in6, size );
echo_full( &ctx.echo, (BitSequence *)hash7, 512,
(const BitSequence *)in7, size );
#endif
break;
case HAMSI:
if ( i == 0 )
hamsi512_8way_update( &ctx.hamsi, input + (64<<3), 16 );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
hamsi512_8way_init( &ctx.hamsi );
hamsi512_8way_update( &ctx.hamsi, vhash, size );
}
hamsi512_8way_close( &ctx.hamsi, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case FUGUE:
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in0, size );
sph_fugue512_close( &ctx.fugue, hash0 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in1, size );
sph_fugue512_close( &ctx.fugue, hash1 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in2, size );
sph_fugue512_close( &ctx.fugue, hash2 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in3, size );
sph_fugue512_close( &ctx.fugue, hash3 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in4, size );
sph_fugue512_close( &ctx.fugue, hash4 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in5, size );
sph_fugue512_close( &ctx.fugue, hash5 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in6, size );
sph_fugue512_close( &ctx.fugue, hash6 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in7, size );
sph_fugue512_close( &ctx.fugue, hash7 );
break;
case SHABAL:
intrlv_8x32( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
if ( i == 0 )
shabal512_8way_update( &ctx.shabal, vhash + (16<<3), 16 );
else
{
shabal512_8way_init( &ctx.shabal );
shabal512_8way_update( &ctx.shabal, vhash, size );
}
shabal512_8way_close( &ctx.shabal, vhash );
dintrlv_8x32_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case WHIRLPOOL:
if ( i == 0 )
{
sph_whirlpool( &ctx.whirlpool, in0 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in1 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash1 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in2 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash2 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in3 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in4 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash4 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in5 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash5 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in6 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash6 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in7 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash7 );
}
else
{
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in0, size );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in1, size );
sph_whirlpool_close( &ctx.whirlpool, hash1 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in2, size );
sph_whirlpool_close( &ctx.whirlpool, hash2 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in3, size );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in4, size );
sph_whirlpool_close( &ctx.whirlpool, hash4 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in5, size );
sph_whirlpool_close( &ctx.whirlpool, hash5 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in6, size );
sph_whirlpool_close( &ctx.whirlpool, hash6 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in7, size );
sph_whirlpool_close( &ctx.whirlpool, hash7 );
}
break;
case SHA_512:
sha512_8way_init( &ctx.sha512 );
if ( i == 0 )
sha512_8way_update( &ctx.sha512, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
sha512_8way_update( &ctx.sha512, vhash, size );
}
sha512_8way_close( &ctx.sha512, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
}
size = 64;
}
memcpy( output, hash0, 32 );
memcpy( output+32, hash1, 32 );
memcpy( output+64, hash2, 32 );
memcpy( output+96, hash3, 32 );
memcpy( output+128, hash4, 32 );
memcpy( output+160, hash5, 32 );
memcpy( output+192, hash6, 32 );
memcpy( output+224, hash7, 32 );
}
#if defined (X16R_8WAY)
int scanhash_x16rt_8way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr)
{
uint32_t hash[16*8] __attribute__ ((aligned (128)));
uint32_t vdata[20*8] __attribute__ ((aligned (64)));
uint32_t vdata2[20*8] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t _ALIGN(64) timeHash[8*8];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
@@ -490,74 +23,25 @@ int scanhash_x16rt_8way( struct work *work, uint32_t max_nonce,
if ( bench ) ptarget[7] = 0x0cff;
static __thread uint32_t s_ntime = UINT32_MAX;
uint32_t ntime = bswap_32( pdata[17] );
if ( s_ntime != ntime )
{
x16rt_getTimeHash( ntime, &timeHash );
x16rt_getAlgoString( &timeHash[0], hashOrder );
x16rt_getAlgoString( &timeHash[0], x16r_hash_order );
s_ntime = ntime;
if ( opt_debug && !thr_id )
applog( LOG_INFO, "hash order: %s time: (%08x) time hash: (%08x)",
hashOrder, ntime, timeHash );
}
// Do midstate prehash on hash functions with block size <= 64 bytes.
const char elem = hashOrder[0];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case JH:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
jh512_8way_init( &x16rt_ctx.jh );
jh512_8way_update( &x16rt_ctx.jh, vdata, 64 );
break;
case SKEIN:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
skein512_8way_init( &x16rt_ctx.skein );
skein512_8way_update( &x16rt_ctx.skein, vdata, 64 );
break;
case LUFFA:
mm128_bswap32_80( edata, pdata );
intrlv_4x128( vdata2, edata, edata, edata, edata, 640 );
luffa_4way_init( &x16rt_ctx.luffa, 512 );
luffa_4way_update( &x16rt_ctx.luffa, vdata2, 64 );
rintrlv_4x128_8x64( vdata, vdata2, vdata2, 640 );
break;
case CUBEHASH:
mm128_bswap32_80( edata, pdata );
cubehashInit( &x16rt_ctx.cube, 512, 16, 32 );
cubehashUpdate( &x16rt_ctx.cube, (const byte*)edata, 64 );
intrlv_8x64( vdata, edata, edata, edata, edata,
edata, edata, edata, edata, 640 );
break;
case HAMSI:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
hamsi512_8way_init( &x16rt_ctx.hamsi );
hamsi512_8way_update( &x16rt_ctx.hamsi, vdata, 64 );
break;
case SHABAL:
mm256_bswap32_intrlv80_8x32( vdata2, pdata );
shabal512_8way_init( &x16rt_ctx.shabal );
shabal512_8way_update( &x16rt_ctx.shabal, vdata2, 64 );
rintrlv_8x32_8x64( vdata, vdata2, 640 );
break;
case WHIRLPOOL:
mm128_bswap32_80( edata, pdata );
sph_whirlpool_init( &x16rt_ctx.whirlpool );
sph_whirlpool( &x16rt_ctx.whirlpool, edata, 64 );
intrlv_8x64( vdata, edata, edata, edata, edata,
edata, edata, edata, edata, 640 );
break;
default:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
x16r_hash_order, ntime, timeHash );
}
x16r_8way_prehash( vdata, pdata );
*noncev = mm512_intrlv_blend_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 );
do
{
x16rt_8way_hash( hash, vdata );
x16r_8way_hash( hash, vdata );
for ( int i = 0; i < 8; i++ )
if ( unlikely( valid_hash( hash + (i<<3), ptarget ) && !bench ) )
@@ -574,313 +58,13 @@ int scanhash_x16rt_8way( struct work *work, uint32_t max_nonce,
return 0;
}
#elif defined (X16RT_4WAY)
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
union _x16rt_4way_context_overlay
{
blake512_4way_context blake;
bmw512_4way_context bmw;
hashState_echo echo;
hashState_groestl groestl;
skein512_4way_context skein;
jh512_4way_context jh;
keccak512_4way_context keccak;
luffa_2way_context luffa;
hashState_luffa luffa1;
cubehashParam cube;
sph_shavite512_context shavite;
simd_2way_context simd;
hamsi512_4way_context hamsi;
sph_fugue512_context fugue;
shabal512_4way_context shabal;
sph_whirlpool_context whirlpool;
sha512_4way_context sha512;
};
typedef union _x16rt_4way_context_overlay x16rt_4way_context_overlay;
static __thread x16rt_4way_context_overlay x16rt_ctx;
void x16rt_4way_hash( void* output, const void* input )
{
uint32_t hash0[20] __attribute__ ((aligned (64)));
uint32_t hash1[20] __attribute__ ((aligned (64)));
uint32_t hash2[20] __attribute__ ((aligned (64)));
uint32_t hash3[20] __attribute__ ((aligned (64)));
uint32_t vhash[20*4] __attribute__ ((aligned (64)));
x16rt_4way_context_overlay ctx;
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
void *in0 = (void*) hash0;
void *in1 = (void*) hash1;
void *in2 = (void*) hash2;
void *in3 = (void*) hash3;
int size = 80;
dintrlv_4x64( hash0, hash1, hash2, hash3, input, 640 );
// Input data is both 64 bit interleaved (input)
// and deinterleaved in inp0-3. First function has no need re-interleave.
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case BLAKE:
if ( i == 0 )
blake512_4way_full( &ctx.blake, vhash, input, size );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
blake512_4way_full( &ctx.blake, vhash, vhash, size );
}
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case BMW:
bmw512_4way_init( &ctx.bmw );
if ( i == 0 )
bmw512_4way_update( &ctx.bmw, input, size );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
bmw512_4way_update( &ctx.bmw, vhash, size );
}
bmw512_4way_close( &ctx.bmw, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case GROESTL:
groestl512_full( &ctx.groestl, (char*)hash0, (char*)in0, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash1, (char*)in1, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash2, (char*)in2, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash3, (char*)in3, size<<3 );
break;
case JH:
if ( i == 0 )
jh512_4way_update( &ctx.jh, input + (64<<2), 16 );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
jh512_4way_init( &ctx.jh );
jh512_4way_update( &ctx.jh, vhash, size );
}
jh512_4way_close( &ctx.jh, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case KECCAK:
keccak512_4way_init( &ctx.keccak );
if ( i == 0 )
keccak512_4way_update( &ctx.keccak, input, size );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
keccak512_4way_update( &ctx.keccak, vhash, size );
}
keccak512_4way_close( &ctx.keccak, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case SKEIN:
if ( i == 0 )
skein512_4way_update( &ctx.skein, input + (64<<2), 16 );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
skein512_4way_init( &ctx.skein );
skein512_4way_update( &ctx.skein, vhash, size );
}
skein512_4way_close( &ctx.skein, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case LUFFA:
if ( i == 0 )
{
update_and_final_luffa( &ctx.luffa1, (BitSequence*)hash0,
(const BitSequence*)in0 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
update_and_final_luffa( &ctx.luffa1, (BitSequence*)hash1,
(const BitSequence*)in1 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
update_and_final_luffa( &ctx.luffa1, (BitSequence*)hash2,
(const BitSequence*)in2 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
update_and_final_luffa( &ctx.luffa1, (BitSequence*)hash3,
(const BitSequence*)in3 + 64, 16 );
}
else
{
intrlv_2x128( vhash, in0, in1, size<<3 );
luffa512_2way_full( &ctx.luffa, vhash, vhash, size );
dintrlv_2x128_512( hash0, hash1, vhash );
intrlv_2x128( vhash, in2, in3, size<<3 );
luffa512_2way_full( &ctx.luffa, vhash, vhash, size );
dintrlv_2x128_512( hash2, hash3, vhash );
}
break;
case CUBEHASH:
if ( i == 0 )
{
cubehashUpdateDigest( &ctx.cube, (byte*)hash0,
(const byte*)in0 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*) hash1,
(const byte*)in1 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*) hash2,
(const byte*)in2 + 64, 16 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*) hash3,
(const byte*)in3 + 64, 16 );
}
else
{
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash0,
(const byte*)in0, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash1,
(const byte*)in1, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash2,
(const byte*)in2, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash3,
(const byte*)in3, size );
}
break;
case SHAVITE:
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in0, size );
sph_shavite512_close( &ctx.shavite, hash0 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in1, size );
sph_shavite512_close( &ctx.shavite, hash1 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in2, size );
sph_shavite512_close( &ctx.shavite, hash2 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in3, size );
sph_shavite512_close( &ctx.shavite, hash3 );
break;
case SIMD:
intrlv_2x128( vhash, in0, in1, size<<3 );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, size<<3 );
dintrlv_2x128( hash0, hash1, vhash, 512 );
intrlv_2x128( vhash, in2, in3, size<<3 );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, size<<3 );
dintrlv_2x128( hash2, hash3, vhash, 512 );
break;
case ECHO:
echo_full( &ctx.echo, (BitSequence *)hash0, 512,
(const BitSequence *)in0, size );
echo_full( &ctx.echo, (BitSequence *)hash1, 512,
(const BitSequence *)in1, size );
echo_full( &ctx.echo, (BitSequence *)hash2, 512,
(const BitSequence *)in2, size );
echo_full( &ctx.echo, (BitSequence *)hash3, 512,
(const BitSequence *)in3, size );
break;
case HAMSI:
if ( i == 0 )
hamsi512_4way_update( &ctx.hamsi, input + (64<<2), 16 );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
hamsi512_4way_init( &ctx.hamsi );
hamsi512_4way_update( &ctx.hamsi, vhash, size );
}
hamsi512_4way_close( &ctx.hamsi, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case FUGUE:
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in0, size );
sph_fugue512_close( &ctx.fugue, hash0 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in1, size );
sph_fugue512_close( &ctx.fugue, hash1 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in2, size );
sph_fugue512_close( &ctx.fugue, hash2 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in3, size );
sph_fugue512_close( &ctx.fugue, hash3 );
break;
case SHABAL:
intrlv_4x32( vhash, in0, in1, in2, in3, size<<3 );
if ( i == 0 )
shabal512_4way_update( &ctx.shabal, vhash + (16<<2), 16 );
else
{
shabal512_4way_init( &ctx.shabal );
shabal512_4way_update( &ctx.shabal, vhash, size );
}
shabal512_4way_close( &ctx.shabal, vhash );
dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case WHIRLPOOL:
if ( i == 0 )
{
sph_whirlpool( &ctx.whirlpool, in0 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in1 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash1 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in2 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash2 );
memcpy( &ctx, &x16rt_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in3 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
}
else
{
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in0, size );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in1, size );
sph_whirlpool_close( &ctx.whirlpool, hash1 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in2, size );
sph_whirlpool_close( &ctx.whirlpool, hash2 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in3, size );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
}
break;
case SHA_512:
sha512_4way_init( &ctx.sha512 );
if ( i == 0 )
sha512_4way_update( &ctx.sha512, input, size );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
sha512_4way_update( &ctx.sha512, vhash, size );
}
sha512_4way_close( &ctx.sha512, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
}
size = 64;
}
memcpy( output, hash0, 32 );
memcpy( output+32, hash1, 32 );
memcpy( output+64, hash2, 32 );
memcpy( output+96, hash3, 32 );
}
#elif defined (X16R_4WAY)
int scanhash_x16rt_4way( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr)
{
uint32_t hash[4*16] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t vdata32[20*4] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t _ALIGN(64) timeHash[4*8];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
@@ -894,70 +78,24 @@ int scanhash_x16rt_4way( struct work *work, uint32_t max_nonce,
if ( bench ) ptarget[7] = 0x0cff;
static __thread uint32_t s_ntime = UINT32_MAX;
uint32_t ntime = bswap_32( pdata[17] );
if ( s_ntime != ntime )
{
x16rt_getTimeHash( ntime, &timeHash );
x16rt_getAlgoString( &timeHash[0], hashOrder );
x16rt_getAlgoString( &timeHash[0], x16r_hash_order );
s_ntime = ntime;
if ( opt_debug && !thr_id )
applog( LOG_INFO, "hash order: %s time: (%08x) time hash: (%08x)",
hashOrder, ntime, timeHash );
}
const char elem = hashOrder[0];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case JH:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
jh512_4way_init( &x16rt_ctx.jh );
jh512_4way_update( &x16rt_ctx.jh, vdata, 64 );
break;
case SKEIN:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
skein512_4way_init( &x16rt_ctx.skein );
skein512_4way_update( &x16rt_ctx.skein, vdata, 64 );
break;
case LUFFA:
mm128_bswap32_80( edata, pdata );
init_luffa( &x16rt_ctx.luffa1, 512 );
update_luffa( &x16rt_ctx.luffa1, (const BitSequence*)edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
case CUBEHASH:
mm128_bswap32_80( edata, pdata );
cubehashInit( &x16rt_ctx.cube, 512, 16, 32 );
cubehashUpdate( &x16rt_ctx.cube, (const byte*)edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
case HAMSI:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
hamsi512_4way_init( &x16rt_ctx.hamsi );
hamsi512_4way_update( &x16rt_ctx.hamsi, vdata, 64 );
break;
case SHABAL:
mm128_bswap32_intrlv80_4x32( vdata32, pdata );
shabal512_4way_init( &x16rt_ctx.shabal );
shabal512_4way_update( &x16rt_ctx.shabal, vdata32, 64 );
rintrlv_4x32_4x64( vdata, vdata32, 640 );
break;
case WHIRLPOOL:
mm128_bswap32_80( edata, pdata );
sph_whirlpool_init( &x16rt_ctx.whirlpool );
sph_whirlpool( &x16rt_ctx.whirlpool, edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
default:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
x16r_hash_order, ntime, timeHash );
}
x16r_4way_prehash( vdata, pdata );
*noncev = mm256_intrlv_blend_32(
_mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
do
{
x16rt_4way_hash( hash, vdata );
x16r_4way_hash( hash, vdata );
for ( int i = 0; i < 4; i++ )
if ( unlikely( valid_hash( hash + (i<<3), ptarget ) && !bench ) )
{

221
algo/x16/x16rt.c
View File

@@ -1,234 +1,46 @@
#include "x16r-gate.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "algo/blake/sph_blake.h"
#include "algo/bmw/sph_bmw.h"
#include "algo/groestl/sph_groestl.h"
#include "algo/jh/sph_jh.h"
#include "algo/keccak/sph_keccak.h"
#include "algo/skein/sph_skein.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/nist.h"
#include "algo/echo/sph_echo.h"
#include "algo/hamsi/sph_hamsi.h"
#include "algo/fugue/sph_fugue.h"
#include "algo/shabal/sph_shabal.h"
#include "algo/whirlpool/sph_whirlpool.h"
#include <openssl/sha.h>
#if defined(__AES__)
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#endif
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread bool s_implemented = false;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
union _x16rt_context_overlay
{
#if defined(__AES__)
hashState_echo echo;
hashState_groestl groestl;
#else
sph_groestl512_context groestl;
sph_echo512_context echo;
#endif
sph_blake512_context blake;
sph_bmw512_context bmw;
sph_skein512_context skein;
sph_jh512_context jh;
sph_keccak512_context keccak;
hashState_luffa luffa;
cubehashParam cube;
sph_shavite512_context shavite;
hashState_sd simd;
sph_hamsi512_context hamsi;
sph_fugue512_context fugue;
sph_shabal512_context shabal;
sph_whirlpool_context whirlpool;
SHA512_CTX sha512;
};
typedef union _x16rt_context_overlay x16rt_context_overlay;
void x16rt_hash( void* output, const void* input )
{
uint32_t _ALIGN(128) hash[16];
x16rt_context_overlay ctx;
int size = 80;
void *in = (void*) input;
/*
void *in = (void*) input;
uint32_t *in32 = (uint32_t*) in;
uint32_t ntime = in32[17];
if ( s_ntime == UINT32_MAX )
{
uint32_t _ALIGN(64) timeHash[8];
x16rt_getTimeHash(ntime, &timeHash);
x16rt_getAlgoString(&timeHash[0], hashOrder);
}
*/
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case BLAKE:
sph_blake512_init( &ctx.blake );
sph_blake512( &ctx.blake, in, size );
sph_blake512_close( &ctx.blake, hash );
break;
case BMW:
sph_bmw512_init( &ctx.bmw );
sph_bmw512(&ctx.bmw, in, size);
sph_bmw512_close(&ctx.bmw, hash);
break;
case GROESTL:
#if defined(__AES__)
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash,
(const char*)in, size<<3 );
#else
sph_groestl512_init( &ctx.groestl );
sph_groestl512( &ctx.groestl, in, size );
sph_groestl512_close(&ctx.groestl, hash);
#endif
break;
case SKEIN:
sph_skein512_init( &ctx.skein );
sph_skein512( &ctx.skein, in, size );
sph_skein512_close( &ctx.skein, hash );
break;
case JH:
sph_jh512_init( &ctx.jh );
sph_jh512(&ctx.jh, in, size );
sph_jh512_close(&ctx.jh, hash );
break;
case KECCAK:
sph_keccak512_init( &ctx.keccak );
sph_keccak512( &ctx.keccak, in, size );
sph_keccak512_close( &ctx.keccak, hash );
break;
case LUFFA:
init_luffa( &ctx.luffa, 512 );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
(const BitSequence*)in, size );
break;
case CUBEHASH:
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash,
(const byte*)in, size );
break;
case SHAVITE:
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in, size );
sph_shavite512_close( &ctx.shavite, hash );
break;
case SIMD:
init_sd( &ctx.simd, 512 );
update_final_sd( &ctx.simd, (BitSequence *)hash,
(const BitSequence*)in, size<<3 );
break;
case ECHO:
#if defined(__AES__)
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash,
(const BitSequence*)in, size<<3 );
#else
sph_echo512_init( &ctx.echo );
sph_echo512( &ctx.echo, in, size );
sph_echo512_close( &ctx.echo, hash );
#endif
break;
case HAMSI:
sph_hamsi512_init( &ctx.hamsi );
sph_hamsi512( &ctx.hamsi, in, size );
sph_hamsi512_close( &ctx.hamsi, hash );
break;
case FUGUE:
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in, size );
sph_fugue512_close( &ctx.fugue, hash );
break;
case SHABAL:
sph_shabal512_init( &ctx.shabal );
sph_shabal512( &ctx.shabal, in, size );
sph_shabal512_close( &ctx.shabal, hash );
break;
case WHIRLPOOL:
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in, size );
sph_whirlpool_close( &ctx.whirlpool, hash );
break;
case SHA_512:
SHA512_Init( &ctx.sha512 );
SHA512_Update( &ctx.sha512, in, size );
SHA512_Final( (unsigned char*) hash, &ctx.sha512 );
break;
}
in = (void*) hash;
size = 64;
}
memcpy(output, hash, 32);
}
#if !defined(X16R_8WAY) && !defined(X16R_4WAY)
int scanhash_x16rt( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t _ALIGN(128) hash32[8];
uint32_t _ALIGN(128) endiandata[20];
uint32_t _ALIGN(128) edata[20];
uint32_t _ALIGN(64) timeHash[8];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t Htarg = ptarget[7];
const uint32_t first_nonce = pdata[19];
int thr_id = mythr->id; // thr_id arg is deprecated
const int thr_id = mythr->id;
uint32_t nonce = first_nonce;
volatile uint8_t *restart = &(work_restart[thr_id].restart);
const bool bench = opt_benchmark;
if ( bench ) ptarget[7] = 0x0cff;
casti_m128i( endiandata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
casti_m128i( endiandata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
casti_m128i( endiandata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
casti_m128i( endiandata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
casti_m128i( endiandata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
mm128_bswap32_80( edata, pdata );
static __thread uint32_t s_ntime = UINT32_MAX;
uint32_t ntime = swab32( pdata[17] );
if ( s_ntime != ntime )
{
x16rt_getTimeHash( ntime, &timeHash );
x16rt_getAlgoString( &timeHash[0], hashOrder );
x16rt_getAlgoString( &timeHash[0], x16r_hash_order );
s_ntime = ntime;
s_implemented = true;
if ( opt_debug && !thr_id )
applog( LOG_INFO, "hash order: %s time: (%08x) time hash: (%08x)",
hashOrder, ntime, timeHash );
}
if ( !s_implemented )
{
applog( LOG_WARNING, "s not implemented");
sleep(1);
return 0;
x16r_hash_order, ntime, timeHash );
}
if ( opt_benchmark )
ptarget[7] = 0x0cff;
x16r_prehash( edata, pdata );
do
{
be32enc( &endiandata[19], nonce );
x16rt_hash( hash32, endiandata );
edata[19] = nonce;
x16r_hash( hash32, edata );
if ( hash32[7] <= Htarg )
if (fulltest( hash32, ptarget ) && !opt_benchmark )
if ( valid_hash( hash32, ptarget ) && !bench )
{
pdata[19] = nonce;
pdata[19] = bswap_32( nonce );
submit_solution( work, hash32, mythr );
}
nonce++;
@@ -237,3 +49,6 @@ int scanhash_x16rt( struct work *work, uint32_t max_nonce,
*hashes_done = pdata[19] - first_nonce + 1;
return 0;
}
#endif // !defined(X16R_8WAY) && !defined(X16R_4WAY)

4
algo/x16/x16rv2.c
View File

@@ -6,6 +6,8 @@
*/
#include "x16r-gate.h"
#if !defined(X16R_8WAY) && !defined(X16R_4WAY)
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
@@ -245,3 +247,5 @@ int scanhash_x16rv2( struct work *work, uint32_t max_nonce,
*hashes_done = pdata[19] - first_nonce + 1;
return 0;
}
#endif

879
algo/x16/x21s-4way.c
View File

@@ -8,480 +8,43 @@
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "algo/blake/blake-hash-4way.h"
#include "algo/bmw/bmw-hash-4way.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#include "algo/skein/skein-hash-4way.h"
#include "algo/jh/jh-hash-4way.h"
#include "algo/keccak/keccak-hash-4way.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa-hash-2way.h"
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/cubehash/cube-hash-2way.h"
#include "algo/simd/simd-hash-2way.h"
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/hamsi/hamsi-hash-4way.h"
#include "algo/fugue/sph_fugue.h"
#include "algo/shabal/shabal-hash-4way.h"
#include "algo/whirlpool/sph_whirlpool.h"
#include "algo/sha/sha-hash-4way.h"
#include "algo/haval/haval-hash-4way.h"
#include "algo/tiger/sph_tiger.h"
#include "algo/gost/sph_gost.h"
#include "algo/lyra2/lyra2.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(__SHA__)
#include <openssl/sha.h>
#endif
#if defined(X21S_8WAY) || defined(X21S_4WAY)
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
#endif
#if defined (X21S_8WAY)
static __thread uint64_t* x21s_8way_matrix;
union _x21s_8way_context_overlay
{
blake512_8way_context blake;
bmw512_8way_context bmw;
skein512_8way_context skein;
jh512_8way_context jh;
keccak512_8way_context keccak;
luffa_4way_context luffa;
cubehashParam cube;
// cube_4way_context cube;
simd_4way_context simd;
hamsi512_8way_context hamsi;
sph_fugue512_context fugue;
shabal512_8way_context shabal;
sph_whirlpool_context whirlpool;
sha512_8way_context sha512;
haval256_5_8way_context haval;
sph_tiger_context tiger;
sph_gost512_context gost;
sha256_8way_context sha256;
#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
} __attribute__ ((aligned (64)));
typedef union _x21s_8way_context_overlay x21s_8way_context_overlay;
static __thread x21s_8way_context_overlay x21s_ctx;
void x21s_8way_hash( void* output, const void* input )
{
uint32_t vhash[20*8] __attribute__ ((aligned (128)));
uint32_t hash0[20] __attribute__ ((aligned (64)));
uint32_t hash1[20] __attribute__ ((aligned (64)));
uint32_t hash2[20] __attribute__ ((aligned (64)));
uint32_t hash3[20] __attribute__ ((aligned (64)));
uint32_t hash4[20] __attribute__ ((aligned (64)));
uint32_t hash5[20] __attribute__ ((aligned (64)));
uint32_t hash6[20] __attribute__ ((aligned (64)));
uint32_t hash7[20] __attribute__ ((aligned (64)));
uint32_t vhash[16*8] __attribute__ ((aligned (128)));
uint8_t shash[64*8] __attribute__ ((aligned (64)));
uint32_t *hash0 = (uint32_t*) shash;
uint32_t *hash1 = (uint32_t*)( shash+64 );
uint32_t *hash2 = (uint32_t*)( shash+128 );
uint32_t *hash3 = (uint32_t*)( shash+192 );
uint32_t *hash4 = (uint32_t*)( shash+256 );
uint32_t *hash5 = (uint32_t*)( shash+320 );
uint32_t *hash6 = (uint32_t*)( shash+384 );
uint32_t *hash7 = (uint32_t*)( shash+448 );
x21s_8way_context_overlay ctx;
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
void *in0 = (void*) hash0;
void *in1 = (void*) hash1;
void *in2 = (void*) hash2;
void *in3 = (void*) hash3;
void *in4 = (void*) hash4;
void *in5 = (void*) hash5;
void *in6 = (void*) hash6;
void *in7 = (void*) hash7;
int size = 80;
dintrlv_8x64( hash0, hash1, hash2, hash3, hash4, hash5, hash6, hash7,
input, 640 );
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case BLAKE:
if ( i == 0 )
blake512_8way_full( &ctx.blake, vhash, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
blake512_8way_full( &ctx.blake, vhash, vhash, size );
}
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5,
hash6, hash7, vhash );
break;
case BMW:
bmw512_8way_init( &ctx.bmw );
if ( i == 0 )
bmw512_8way_update( &ctx.bmw, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
bmw512_8way_update( &ctx.bmw, vhash, size );
}
bmw512_8way_close( &ctx.bmw, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case GROESTL:
#if defined(__VAES__)
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
groestl512_4way_full( &ctx.groestl, vhash, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
groestl512_4way_full( &ctx.groestl, vhash, vhash, size );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
#else
groestl512_full( &ctx.groestl, (char*)hash0, (char*)in0, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash1, (char*)in1, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash2, (char*)in2, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash3, (char*)in3, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash4, (char*)in4, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash5, (char*)in5, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash6, (char*)in6, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash7, (char*)in7, size<<3 );
#endif
break;
case JH:
if ( i == 0 )
jh512_8way_update( &ctx.jh, input + (64<<3), 16 );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
jh512_8way_init( &ctx.jh );
jh512_8way_update( &ctx.jh, vhash, size );
}
jh512_8way_close( &ctx.jh, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case KECCAK:
keccak512_8way_init( &ctx.keccak );
if ( i == 0 )
keccak512_8way_update( &ctx.keccak, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
keccak512_8way_update( &ctx.keccak, vhash, size );
}
keccak512_8way_close( &ctx.keccak, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case SKEIN:
if ( i == 0 )
skein512_8way_update( &ctx.skein, input + (64<<3), 16 );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
skein512_8way_init( &ctx.skein );
skein512_8way_update( &ctx.skein, vhash, size );
}
skein512_8way_close( &ctx.skein, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case LUFFA:
if ( i == 0 )
{
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
luffa_4way_update_close( &ctx.luffa, vhash,
vhash + (16<<2), 16 );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
luffa_4way_update_close( &ctx.luffa, vhash,
vhash + (16<<2), 16 );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
}
else
{
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
luffa512_4way_full( &ctx.luffa, vhash, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
luffa512_4way_full( &ctx.luffa, vhash, vhash, size );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
}
break;
case CUBEHASH:
if ( i == 0 )
{
cubehashUpdateDigest( &ctx.cube, (byte*)hash0,
(const byte*)in0 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash1,
(const byte*)in1 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash2,
(const byte*)in2 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash3,
(const byte*)in3 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash4,
(const byte*)in4 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash5,
(const byte*)in5 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash6,
(const byte*)in6 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash7,
(const byte*)in7 + 64, 16 );
}
else
{
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash0,
(const byte*)in0, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash1,
(const byte*)in1, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash2,
(const byte*)in2, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash3,
(const byte*)in3, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash4,
(const byte*)in4, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash5,
(const byte*)in5, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash6,
(const byte*)in6, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash7,
(const byte*)in7, size );
}
break;
case SHAVITE:
#if defined(__VAES__)
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
shavite512_4way_init( &ctx.shavite );
shavite512_4way_update_close( &ctx.shavite, vhash, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
shavite512_4way_init( &ctx.shavite );
shavite512_4way_update_close( &ctx.shavite, vhash, vhash, size );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
#else
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in0, size );
sph_shavite512_close( &ctx.shavite, hash0 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in1, size );
sph_shavite512_close( &ctx.shavite, hash1 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in2, size );
sph_shavite512_close( &ctx.shavite, hash2 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in3, size );
sph_shavite512_close( &ctx.shavite, hash3 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in4, size );
sph_shavite512_close( &ctx.shavite, hash4 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in5, size );
sph_shavite512_close( &ctx.shavite, hash5 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in6, size );
sph_shavite512_close( &ctx.shavite, hash6 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in7, size );
sph_shavite512_close( &ctx.shavite, hash7 );
#endif
break;
case SIMD:
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
simd512_4way_full( &ctx.simd, vhash, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
simd512_4way_full( &ctx.simd, vhash, vhash, size );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
break;
case ECHO:
#if defined(__VAES__)
intrlv_4x128( vhash, in0, in1, in2, in3, size<<3 );
echo_4way_full( &ctx.echo, vhash, 512, vhash, size );
dintrlv_4x128_512( hash0, hash1, hash2, hash3, vhash );
intrlv_4x128( vhash, in4, in5, in6, in7, size<<3 );
echo_4way_full( &ctx.echo, vhash, 512, vhash, size );
dintrlv_4x128_512( hash4, hash5, hash6, hash7, vhash );
#else
echo_full( &ctx.echo, (BitSequence *)hash0, 512,
(const BitSequence *)in0, size );
echo_full( &ctx.echo, (BitSequence *)hash1, 512,
(const BitSequence *)in1, size );
echo_full( &ctx.echo, (BitSequence *)hash2, 512,
(const BitSequence *)in2, size );
echo_full( &ctx.echo, (BitSequence *)hash3, 512,
(const BitSequence *)in3, size );
echo_full( &ctx.echo, (BitSequence *)hash4, 512,
(const BitSequence *)in4, size );
echo_full( &ctx.echo, (BitSequence *)hash5, 512,
(const BitSequence *)in5, size );
echo_full( &ctx.echo, (BitSequence *)hash6, 512,
(const BitSequence *)in6, size );
echo_full( &ctx.echo, (BitSequence *)hash7, 512,
(const BitSequence *)in7, size );
#endif
break;
case HAMSI:
if ( i == 0 )
hamsi512_8way_update( &ctx.hamsi, input + (64<<3), 16 );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
hamsi512_8way_init( &ctx.hamsi );
hamsi512_8way_update( &ctx.hamsi, vhash, size );
}
hamsi512_8way_close( &ctx.hamsi, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case FUGUE:
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in0, size );
sph_fugue512_close( &ctx.fugue, hash0 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in1, size );
sph_fugue512_close( &ctx.fugue, hash1 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in2, size );
sph_fugue512_close( &ctx.fugue, hash2 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in3, size );
sph_fugue512_close( &ctx.fugue, hash3 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in4, size );
sph_fugue512_close( &ctx.fugue, hash4 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in5, size );
sph_fugue512_close( &ctx.fugue, hash5 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in6, size );
sph_fugue512_close( &ctx.fugue, hash6 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in7, size );
sph_fugue512_close( &ctx.fugue, hash7 );
break;
case SHABAL:
intrlv_8x32( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
if ( i == 0 )
shabal512_8way_update( &ctx.shabal, vhash + (16<<3), 16 );
else
{
shabal512_8way_init( &ctx.shabal );
shabal512_8way_update( &ctx.shabal, vhash, size );
}
shabal512_8way_close( &ctx.shabal, vhash );
dintrlv_8x32_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
case WHIRLPOOL:
if ( i == 0 )
{
sph_whirlpool( &ctx.whirlpool, in0 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in1 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash1 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in2 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash2 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in3 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in4 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash4 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in5 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash5 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in6 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash6 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in7 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash7 );
}
else
{
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in0, size );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in1, size );
sph_whirlpool_close( &ctx.whirlpool, hash1 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in2, size );
sph_whirlpool_close( &ctx.whirlpool, hash2 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in3, size );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in4, size );
sph_whirlpool_close( &ctx.whirlpool, hash4 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in5, size );
sph_whirlpool_close( &ctx.whirlpool, hash5 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in6, size );
sph_whirlpool_close( &ctx.whirlpool, hash6 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in7, size );
sph_whirlpool_close( &ctx.whirlpool, hash7 );
}
break;
case SHA_512:
sha512_8way_init( &ctx.sha512 );
if ( i == 0 )
sha512_8way_update( &ctx.sha512, input, size );
else
{
intrlv_8x64( vhash, in0, in1, in2, in3, in4, in5, in6, in7,
size<<3 );
sha512_8way_update( &ctx.sha512, vhash, size );
}
sha512_8way_close( &ctx.sha512, vhash );
dintrlv_8x64_512( hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7, vhash );
break;
}
size = 64;
}
x16r_8way_hash_generic( shash, input );
intrlv_8x32_512( vhash, hash0, hash1, hash2, hash3, hash4, hash5, hash6,
hash7 );
@@ -568,8 +131,6 @@ int scanhash_x21s_8way( struct work *work, uint32_t max_nonce,
{
uint32_t hash[16*8] __attribute__ ((aligned (128)));
uint32_t vdata[20*8] __attribute__ ((aligned (64)));
uint32_t vdata2[20*8] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t *hash7 = &hash[7<<3];
uint32_t lane_hash[8] __attribute__ ((aligned (64)));
uint32_t bedata1[2] __attribute__((aligned(64)));
@@ -588,71 +149,21 @@ int scanhash_x21s_8way( struct work *work, uint32_t max_nonce,
bedata1[0] = bswap_32( pdata[1] );
bedata1[1] = bswap_32( pdata[2] );
static __thread uint32_t s_ntime = UINT32_MAX;
uint32_t ntime = bswap_32( pdata[17] );
if ( s_ntime != ntime )
{
x16_r_s_getAlgoString( (const uint8_t*)bedata1, hashOrder );
x16_r_s_getAlgoString( (const uint8_t*)bedata1, x16r_hash_order );
s_ntime = ntime;
if ( opt_debug && !thr_id )
applog( LOG_INFO, "hash order %s (%08x)", hashOrder, ntime );
}
// Do midstate prehash on hash functions with block size <= 64 bytes.
const char elem = hashOrder[0];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case JH:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
jh512_8way_init( &x21s_ctx.jh );
jh512_8way_update( &x21s_ctx.jh, vdata, 64 );
break;
case SKEIN:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
skein512_8way_init( &x21s_ctx.skein );
skein512_8way_update( &x21s_ctx.skein, vdata, 64 );
break;
case LUFFA:
mm128_bswap32_80( edata, pdata );
intrlv_4x128( vdata2, edata, edata, edata, edata, 640 );
luffa_4way_init( &x21s_ctx.luffa, 512 );
luffa_4way_update( &x21s_ctx.luffa, vdata2, 64 );
rintrlv_4x128_8x64( vdata, vdata2, vdata2, 640 );
break;
case CUBEHASH:
mm128_bswap32_80( edata, pdata );
cubehashInit( &x21s_ctx.cube, 512, 16, 32 );
cubehashUpdate( &x21s_ctx.cube, (const byte*)edata, 64 );
intrlv_8x64( vdata, edata, edata, edata, edata,
edata, edata, edata, edata, 640 );
break;
case HAMSI:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
hamsi512_8way_init( &x21s_ctx.hamsi );
hamsi512_8way_update( &x21s_ctx.hamsi, vdata, 64 );
break;
case SHABAL:
mm256_bswap32_intrlv80_8x32( vdata2, pdata );
shabal512_8way_init( &x21s_ctx.shabal );
shabal512_8way_update( &x21s_ctx.shabal, vdata2, 64 );
rintrlv_8x32_8x64( vdata, vdata2, 640 );
break;
case WHIRLPOOL:
mm128_bswap32_80( edata, pdata );
sph_whirlpool_init( &x21s_ctx.whirlpool );
sph_whirlpool( &x21s_ctx.whirlpool, edata, 64 );
intrlv_8x64( vdata, edata, edata, edata, edata,
edata, edata, edata, edata, 640 );
break;
default:
mm512_bswap32_intrlv80_8x64( vdata, pdata );
applog( LOG_INFO, "hash order %s (%08x)", x16r_hash_order, ntime );
}
x16r_8way_prehash( vdata, pdata );
*noncev = mm512_intrlv_blend_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 );
do
{
x21s_8way_hash( hash, vdata );
@@ -670,7 +181,7 @@ int scanhash_x21s_8way( struct work *work, uint32_t max_nonce,
*noncev = _mm512_add_epi32( *noncev,
m512_const1_64( 0x0000000800000000 ) );
n += 8;
} while ( ( n < last_nonce ) && !(*restart) );
} while ( likely( ( n < last_nonce ) && !(*restart) ) );
pdata[19] = n;
*hashes_done = n - first_nonce;
return 0;
@@ -692,23 +203,6 @@ static __thread uint64_t* x21s_4way_matrix;
union _x21s_4way_context_overlay
{
blake512_4way_context blake;
bmw512_4way_context bmw;
hashState_echo echo;
hashState_groestl groestl;
skein512_4way_context skein;
jh512_4way_context jh;
keccak512_4way_context keccak;
luffa_2way_context luffa;
hashState_luffa luffa1;
cubehashParam cube;
sph_shavite512_context shavite;
simd_2way_context simd;
hamsi512_4way_context hamsi;
sph_fugue512_context fugue;
shabal512_4way_context shabal;
sph_whirlpool_context whirlpool;
sha512_4way_context sha512;
haval256_5_4way_context haval;
sph_tiger_context tiger;
sph_gost512_context gost;
@@ -718,282 +212,21 @@ union _x21s_4way_context_overlay
sha256_4way_context sha256;
#endif
} __attribute__ ((aligned (64)));
typedef union _x21s_4way_context_overlay x21s_4way_context_overlay;
static __thread x21s_4way_context_overlay x21s_ctx;
typedef union _x21s_4way_context_overlay x21s_4way_context_overlay;
void x21s_4way_hash( void* output, const void* input )
{
uint32_t hash0[20] __attribute__ ((aligned (64)));
uint32_t hash1[20] __attribute__ ((aligned (64)));
uint32_t hash2[20] __attribute__ ((aligned (64)));
uint32_t hash3[20] __attribute__ ((aligned (64)));
uint32_t vhash[20*4] __attribute__ ((aligned (64)));
uint32_t vhash[16*4] __attribute__ ((aligned (64)));
uint8_t shash[64*4] __attribute__ ((aligned (64)));
x21s_4way_context_overlay ctx;
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
void *in0 = (void*) hash0;
void *in1 = (void*) hash1;
void *in2 = (void*) hash2;
void *in3 = (void*) hash3;
int size = 80;
dintrlv_4x64( hash0, hash1, hash2, hash3, input, 640 );
// Input data is both 64 bit interleaved (input)
// and deinterleaved in inp0-3.
// If First function uses 64 bit data it is not required to interleave inp
// first. It may use the inerleaved data dmost convenient, ie 4way 64 bit.
// All other functions assume data is deinterleaved in hash0-3
// All functions must exit with data deinterleaved in hash0-3.
// Alias in0-3 points to either inp0-3 or hash0-3 according to
// its hashOrder position. Size is also set accordingly.
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case BLAKE:
if ( i == 0 )
blake512_4way_full( &ctx.blake, vhash, input, size );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
blake512_4way_full( &ctx.blake, vhash, vhash, size );
}
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case BMW:
bmw512_4way_init( &ctx.bmw );
if ( i == 0 )
bmw512_4way_update( &ctx.bmw, input, size );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
bmw512_4way_update( &ctx.bmw, vhash, size );
}
bmw512_4way_close( &ctx.bmw, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case GROESTL:
groestl512_full( &ctx.groestl, (char*)hash0, (char*)in0, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash1, (char*)in1, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash2, (char*)in2, size<<3 );
groestl512_full( &ctx.groestl, (char*)hash3, (char*)in3, size<<3 );
break;
case JH:
if ( i == 0 )
jh512_4way_update( &ctx.jh, input + (64<<2), 16 );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
jh512_4way_init( &ctx.jh );
jh512_4way_update( &ctx.jh, vhash, size );
}
jh512_4way_close( &ctx.jh, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case KECCAK:
keccak512_4way_init( &ctx.keccak );
if ( i == 0 )
keccak512_4way_update( &ctx.keccak, input, size );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
keccak512_4way_update( &ctx.keccak, vhash, size );
}
keccak512_4way_close( &ctx.keccak, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case SKEIN:
if ( i == 0 )
skein512_4way_update( &ctx.skein, input + (64<<2), 16 );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
skein512_4way_init( &ctx.skein );
skein512_4way_update( &ctx.skein, vhash, size );
}
skein512_4way_close( &ctx.skein, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case LUFFA:
if ( i == 0 )
{
update_and_final_luffa( &ctx.luffa1, (BitSequence*)hash0,
(const BitSequence*)in0 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
update_and_final_luffa( &ctx.luffa1, (BitSequence*)hash1,
(const BitSequence*)in1 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
update_and_final_luffa( &ctx.luffa1, (BitSequence*)hash2,
(const BitSequence*)in2 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
update_and_final_luffa( &ctx.luffa1, (BitSequence*)hash3,
(const BitSequence*)in3 + 64, 16 );
}
else
{
intrlv_2x128( vhash, in0, in1, size<<3 );
luffa512_2way_full( &ctx.luffa, vhash, vhash, size );
dintrlv_2x128_512( hash0, hash1, vhash );
intrlv_2x128( vhash, in2, in3, size<<3 );
luffa512_2way_full( &ctx.luffa, vhash, vhash, size );
dintrlv_2x128_512( hash2, hash3, vhash );
}
break;
case CUBEHASH:
if ( i == 0 )
{
cubehashUpdateDigest( &ctx.cube, (byte*)hash0,
(const byte*)in0 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*) hash1,
(const byte*)in1 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*) hash2,
(const byte*)in2 + 64, 16 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
cubehashUpdateDigest( &ctx.cube, (byte*) hash3,
(const byte*)in3 + 64, 16 );
}
else
{
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash0,
(const byte*)in0, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash1,
(const byte*)in1, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash2,
(const byte*)in2, size );
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash3,
(const byte*)in3, size );
}
break;
case SHAVITE:
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in0, size );
sph_shavite512_close( &ctx.shavite, hash0 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in1, size );
sph_shavite512_close( &ctx.shavite, hash1 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in2, size );
sph_shavite512_close( &ctx.shavite, hash2 );
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in3, size );
sph_shavite512_close( &ctx.shavite, hash3 );
break;
case SIMD:
intrlv_2x128( vhash, in0, in1, size<<3 );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, size<<3 );
dintrlv_2x128( hash0, hash1, vhash, 512 );
intrlv_2x128( vhash, in2, in3, size<<3 );
simd_2way_init( &ctx.simd, 512 );
simd_2way_update_close( &ctx.simd, vhash, vhash, size<<3 );
dintrlv_2x128( hash2, hash3, vhash, 512 );
break;
case ECHO:
echo_full( &ctx.echo, (BitSequence *)hash0, 512,
(const BitSequence *)in0, size );
echo_full( &ctx.echo, (BitSequence *)hash1, 512,
(const BitSequence *)in1, size );
echo_full( &ctx.echo, (BitSequence *)hash2, 512,
(const BitSequence *)in2, size );
echo_full( &ctx.echo, (BitSequence *)hash3, 512,
(const BitSequence *)in3, size );
break;
case HAMSI:
if ( i == 0 )
hamsi512_4way_update( &ctx.hamsi, input + (64<<2), 16 );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
hamsi512_4way_init( &ctx.hamsi );
hamsi512_4way_update( &ctx.hamsi, vhash, size );
}
hamsi512_4way_close( &ctx.hamsi, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case FUGUE:
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in0, size );
sph_fugue512_close( &ctx.fugue, hash0 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in1, size );
sph_fugue512_close( &ctx.fugue, hash1 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in2, size );
sph_fugue512_close( &ctx.fugue, hash2 );
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in3, size );
sph_fugue512_close( &ctx.fugue, hash3 );
break;
case SHABAL:
intrlv_4x32( vhash, in0, in1, in2, in3, size<<3 );
if ( i == 0 )
shabal512_4way_update( &ctx.shabal, vhash + (16<<2), 16 );
else
{
shabal512_4way_init( &ctx.shabal );
shabal512_4way_update( &ctx.shabal, vhash, size );
}
shabal512_4way_close( &ctx.shabal, vhash );
dintrlv_4x32( hash0, hash1, hash2, hash3, vhash, 512 );
break;
case WHIRLPOOL:
if ( i == 0 )
{
sph_whirlpool( &ctx.whirlpool, in0 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in1 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash1 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in2 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash2 );
memcpy( &ctx, &x21s_ctx, sizeof(ctx) );
sph_whirlpool( &ctx.whirlpool, in3 + 64, 16 );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
}
else
{
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in0, size );
sph_whirlpool_close( &ctx.whirlpool, hash0 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in1, size );
sph_whirlpool_close( &ctx.whirlpool, hash1 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in2, size );
sph_whirlpool_close( &ctx.whirlpool, hash2 );
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in3, size );
sph_whirlpool_close( &ctx.whirlpool, hash3 );
}
break;
case SHA_512:
sha512_4way_init( &ctx.sha512 );
if ( i == 0 )
sha512_4way_update( &ctx.sha512, input, size );
else
{
intrlv_4x64( vhash, in0, in1, in2, in3, size<<3 );
sha512_4way_update( &ctx.sha512, vhash, size );
}
sha512_4way_close( &ctx.sha512, vhash );
dintrlv_4x64( hash0, hash1, hash2, hash3, vhash, 512 );
break;
}
size = 64;
}
uint32_t *hash0 = (uint32_t*) shash;
uint32_t *hash1 = (uint32_t*)( shash+64 );
uint32_t *hash2 = (uint32_t*)( shash+128 );
uint32_t *hash3 = (uint32_t*)( shash+192 );
x16r_4way_hash_generic( shash, input );
intrlv_4x32( vhash, hash0, hash1, hash2, hash3, 512 );
haval256_5_4way_init( &ctx.haval );
@@ -1073,8 +306,6 @@ int scanhash_x21s_4way( struct work *work, uint32_t max_nonce,
{
uint32_t hash[16*4] __attribute__ ((aligned (64)));
uint32_t vdata[20*4] __attribute__ ((aligned (64)));
uint32_t vdata32[20*4] __attribute__ ((aligned (64)));
uint32_t edata[20] __attribute__ ((aligned (64)));
uint32_t bedata1[2] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
@@ -1090,66 +321,20 @@ int scanhash_x21s_4way( struct work *work, uint32_t max_nonce,
bedata1[0] = bswap_32( pdata[1] );
bedata1[1] = bswap_32( pdata[2] );
static __thread uint32_t s_ntime = UINT32_MAX;
uint32_t ntime = bswap_32( pdata[17] );
if ( s_ntime != ntime )
{
x16_r_s_getAlgoString( (const uint8_t*)bedata1, hashOrder );
x16_r_s_getAlgoString( (const uint8_t*)bedata1, x16r_hash_order );
s_ntime = ntime;
if ( opt_debug && !thr_id )
applog( LOG_DEBUG, "hash order %s (%08x)", hashOrder, ntime );
}
const char elem = hashOrder[0];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case JH:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
jh512_4way_init( &x21s_ctx.jh );
jh512_4way_update( &x21s_ctx.jh, vdata, 64 );
break;
case SKEIN:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
skein512_4way_init( &x21s_ctx.skein );
skein512_4way_update( &x21s_ctx.skein, vdata, 64 );
break;
case LUFFA:
mm128_bswap32_80( edata, pdata );
init_luffa( &x21s_ctx.luffa1, 512 );
update_luffa( &x21s_ctx.luffa1, (const BitSequence*)edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
case CUBEHASH:
mm128_bswap32_80( edata, pdata );
cubehashInit( &x21s_ctx.cube, 512, 16, 32 );
cubehashUpdate( &x21s_ctx.cube, (const byte*)edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
case HAMSI:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
hamsi512_4way_init( &x21s_ctx.hamsi );
hamsi512_4way_update( &x21s_ctx.hamsi, vdata, 64 );
break;
case SHABAL:
mm128_bswap32_intrlv80_4x32( vdata32, pdata );
shabal512_4way_init( &x21s_ctx.shabal );
shabal512_4way_update( &x21s_ctx.shabal, vdata32, 64 );
rintrlv_4x32_4x64( vdata, vdata32, 640 );
break;
case WHIRLPOOL:
mm128_bswap32_80( edata, pdata );
sph_whirlpool_init( &x21s_ctx.whirlpool );
sph_whirlpool( &x21s_ctx.whirlpool, edata, 64 );
intrlv_4x64( vdata, edata, edata, edata, edata, 640 );
break;
default:
mm256_bswap32_intrlv80_4x64( vdata, pdata );
applog( LOG_DEBUG, "hash order %s (%08x)", x16r_hash_order, ntime );
}
x16r_4way_prehash( vdata, pdata );
*noncev = mm256_intrlv_blend_32(
_mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
do
{
x21s_4way_hash( hash, vdata );
@@ -1162,7 +347,7 @@ int scanhash_x21s_4way( struct work *work, uint32_t max_nonce,
*noncev = _mm256_add_epi32( *noncev,
m256_const1_64( 0x0000000400000000 ) );
n += 4;
} while ( ( n < last_nonce ) && !(*restart) );
} while ( likely( ( n < last_nonce ) && !(*restart) ) );
pdata[19] = n;
*hashes_done = n - first_nonce;
return 0;

181
algo/x16/x21s.c
View File

@@ -5,63 +5,21 @@
* Optimized by JayDDee@github Jan 2018
*/
#include "x16r-gate.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "algo/blake/sph_blake.h"
#include "algo/bmw/sph_bmw.h"
#include "algo/groestl/sph_groestl.h"
#include "algo/jh/sph_jh.h"
#include "algo/keccak/sph_keccak.h"
#include "algo/skein/sph_skein.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/luffa/luffa_for_sse2.h"
#include "algo/cubehash/cubehash_sse2.h"
#include "algo/simd/nist.h"
#include "algo/echo/sph_echo.h"
#include "algo/hamsi/sph_hamsi.h"
#include "algo/fugue/sph_fugue.h"
#include "algo/shabal/sph_shabal.h"
#include "algo/whirlpool/sph_whirlpool.h"
#include <openssl/sha.h>
#if defined(__AES__)
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#endif
#include "algo/haval/sph-haval.h"
#include "algo/tiger/sph_tiger.h"
#include "algo/gost/sph_gost.h"
#include "algo/lyra2/lyra2.h"
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread char hashOrder[X16R_HASH_FUNC_COUNT + 1] = { 0 };
#if !defined(X16R_8WAY) && !defined(X16R_4WAY)
static __thread uint64_t* x21s_matrix;
union _x21s_context_overlay
{
#if defined(__AES__)
hashState_echo echo;
hashState_groestl groestl;
#else
sph_groestl512_context groestl;
sph_echo512_context echo;
#endif
sph_blake512_context blake;
sph_bmw512_context bmw;
sph_skein512_context skein;
sph_jh512_context jh;
sph_keccak512_context keccak;
hashState_luffa luffa;
cubehashParam cube;
sph_shavite512_context shavite;
hashState_sd simd;
sph_hamsi512_context hamsi;
sph_fugue512_context fugue;
sph_shabal512_context shabal;
sph_whirlpool_context whirlpool;
SHA512_CTX sha512;
sph_haval256_5_context haval;
sph_tiger_context tiger;
sph_gost512_context gost;
@@ -73,112 +31,8 @@ void x21s_hash( void* output, const void* input )
{
uint32_t _ALIGN(128) hash[16];
x21s_context_overlay ctx;
void *in = (void*) input;
int size = 80;
for ( int i = 0; i < 16; i++ )
{
const char elem = hashOrder[i];
const uint8_t algo = elem >= 'A' ? elem - 'A' + 10 : elem - '0';
switch ( algo )
{
case BLAKE:
sph_blake512_init( &ctx.blake );
sph_blake512( &ctx.blake, in, size );
sph_blake512_close( &ctx.blake, hash );
break;
case BMW:
sph_bmw512_init( &ctx.bmw );
sph_bmw512(&ctx.bmw, in, size);
sph_bmw512_close(&ctx.bmw, hash);
break;
case GROESTL:
#if defined(__AES__)
init_groestl( &ctx.groestl, 64 );
update_and_final_groestl( &ctx.groestl, (char*)hash,
(const char*)in, size<<3 );
#else
sph_groestl512_init( &ctx.groestl );
sph_groestl512( &ctx.groestl, in, size );
sph_groestl512_close(&ctx.groestl, hash);
#endif
break;
case SKEIN:
sph_skein512_init( &ctx.skein );
sph_skein512( &ctx.skein, in, size );
sph_skein512_close( &ctx.skein, hash );
break;
case JH:
sph_jh512_init( &ctx.jh );
sph_jh512(&ctx.jh, in, size );
sph_jh512_close(&ctx.jh, hash );
break;
case KECCAK:
sph_keccak512_init( &ctx.keccak );
sph_keccak512( &ctx.keccak, in, size );
sph_keccak512_close( &ctx.keccak, hash );
break;
case LUFFA:
init_luffa( &ctx.luffa, 512 );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash,
(const BitSequence*)in, size );
break;
case CUBEHASH:
cubehashInit( &ctx.cube, 512, 16, 32 );
cubehashUpdateDigest( &ctx.cube, (byte*) hash,
(const byte*)in, size );
break;
case SHAVITE:
sph_shavite512_init( &ctx.shavite );
sph_shavite512( &ctx.shavite, in, size );
sph_shavite512_close( &ctx.shavite, hash );
break;
case SIMD:
init_sd( &ctx.simd, 512 );
update_final_sd( &ctx.simd, (BitSequence *)hash,
(const BitSequence*)in, size<<3 );
break;
case ECHO:
#if defined(__AES__)
init_echo( &ctx.echo, 512 );
update_final_echo ( &ctx.echo, (BitSequence *)hash,
(const BitSequence*)in, size<<3 );
#else
sph_echo512_init( &ctx.echo );
sph_echo512( &ctx.echo, in, size );
sph_echo512_close( &ctx.echo, hash );
#endif
break;
case HAMSI:
sph_hamsi512_init( &ctx.hamsi );
sph_hamsi512( &ctx.hamsi, in, size );
sph_hamsi512_close( &ctx.hamsi, hash );
break;
case FUGUE:
sph_fugue512_init( &ctx.fugue );
sph_fugue512( &ctx.fugue, in, size );
sph_fugue512_close( &ctx.fugue, hash );
break;
case SHABAL:
sph_shabal512_init( &ctx.shabal );
sph_shabal512( &ctx.shabal, in, size );
sph_shabal512_close( &ctx.shabal, hash );
break;
case WHIRLPOOL:
sph_whirlpool_init( &ctx.whirlpool );
sph_whirlpool( &ctx.whirlpool, in, size );
sph_whirlpool_close( &ctx.whirlpool, hash );
break;
case SHA_512:
SHA512_Init( &ctx.sha512 );
SHA512_Update( &ctx.sha512, in, size );
SHA512_Final( (unsigned char*) hash, &ctx.sha512 );
break;
}
in = (void*) hash;
size = 64;
}
x16r_hash_generic( hash, input );
sph_haval256_5_init( &ctx.haval );
sph_haval256_5( &ctx.haval, (const void*) hash, 64) ;
@@ -206,42 +60,38 @@ int scanhash_x21s( struct work *work, uint32_t max_nonce,
uint64_t *hashes_done, struct thr_info *mythr )
{
uint32_t _ALIGN(128) hash32[8];
uint32_t _ALIGN(128) endiandata[20];
uint32_t _ALIGN(128) edata[20];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t Htarg = ptarget[7];
const uint32_t first_nonce = pdata[19];
int thr_id = mythr->id; // thr_id arg is deprecated
const int thr_id = mythr->id;
uint32_t nonce = first_nonce;
volatile uint8_t *restart = &(work_restart[thr_id].restart);
const bool bench = opt_benchmark;
if ( bench ) ptarget[7] = 0x0cff;
casti_m128i( endiandata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
casti_m128i( endiandata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
casti_m128i( endiandata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
casti_m128i( endiandata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
casti_m128i( endiandata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
mm128_bswap32_80( edata, pdata );
static __thread uint32_t s_ntime = UINT32_MAX;
if ( s_ntime != pdata[17] )
{
uint32_t ntime = swab32(pdata[17]);
x16_r_s_getAlgoString( (const uint8_t*) (&endiandata[1]), hashOrder );
x16_r_s_getAlgoString( (const uint8_t*)(&edata[1]), x16r_hash_order );
s_ntime = ntime;
if ( opt_debug && !thr_id )
applog( LOG_DEBUG, "hash order %s (%08x)", hashOrder, ntime );
applog( LOG_INFO, "hash order %s (%08x)", x16r_hash_order, ntime );
}
if ( opt_benchmark )
ptarget[7] = 0x0cff;
x16r_prehash( edata, pdata );
do
{
be32enc( &endiandata[19], nonce );
x21s_hash( hash32, endiandata );
edata[19] = nonce;
x21s_hash( hash32, edata );
if ( hash32[7] <= Htarg )
if (fulltest( hash32, ptarget ) && !opt_benchmark )
if ( unlikely( valid_hash( hash32, ptarget ) && !bench ) )
{
pdata[19] = nonce;
pdata[19] = bswap_32( nonce );
submit_solution( work, hash32, mythr );
}
nonce++;
@@ -261,3 +111,4 @@ bool x21s_thread_init()
return x21s_matrix;
}
#endif