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

v3.7.10

Browse Source
This commit is contained in:
Jay D Dee
2018-01-16 15:11:44 -05:00
parent bee78eac76
commit a90d75b8f5
77 changed files with 3408 additions and 1214 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
algo/x11/c11.c
View File

@@ -162,7 +162,8 @@ int scanhash_c11( int thr_id, struct work *work, uint32_t max_nonce,
{
pdata[19] = nonce;
*hashes_done = pdata[19] - first_nonce;
return 1;
work_set_target_ratio( work, hash );
return 1;
}
nonce++;
} while ( nonce < max_nonce && !(*restart) );

274
algo/x11/timetravel-4way.c Normal file
View File

@@ -0,0 +1,274 @@
#include "timetravel-gate.h"
#if defined(__AVX2__) && defined(__AES__)
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include <stdio.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/skein/skein-hash-4way.h"
#include "algo/jh/jh-hash-4way.h"
#include "algo/keccak/keccak-hash-4way.h"
#include "algo/luffa/sse2/luffa_for_sse2.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread int permutation[TT8_FUNC_COUNT] = { 0 };
typedef struct {
blake512_4way_context blake;
bmw512_4way_context bmw;
hashState_groestl groestl;
skein512_4way_context skein;
jh512_4way_context jh;
keccak512_4way_context keccak;
hashState_luffa luffa;
cubehashParam cube;
} tt8_4way_ctx_holder;
tt8_4way_ctx_holder tt8_4way_ctx __attribute__ ((aligned (64)));
void init_tt8_4way_ctx()
{
blake512_4way_init( &tt8_4way_ctx.blake );
bmw512_4way_init( &tt8_4way_ctx.bmw );
init_groestl( &tt8_4way_ctx.groestl, 64 );
skein512_4way_init( &tt8_4way_ctx.skein );
jh512_4way_init( &tt8_4way_ctx.jh );
keccak512_4way_init( &tt8_4way_ctx.keccak );
init_luffa( &tt8_4way_ctx.luffa, 512 );
cubehashInit( &tt8_4way_ctx.cube, 512, 16, 32 );
};
void timetravel_4way_hash(void *output, const void *input)
{
uint64_t hash0[8] __attribute__ ((aligned (64)));
uint64_t hash1[8] __attribute__ ((aligned (64)));
uint64_t hash2[8] __attribute__ ((aligned (64)));
uint64_t hash3[8] __attribute__ ((aligned (64)));
uint64_t vhashX[8*4] __attribute__ ((aligned (64)));
uint64_t vhashY[8*4] __attribute__ ((aligned (64)));
uint64_t *vhashA, *vhashB;
tt8_4way_ctx_holder ctx __attribute__ ((aligned (64)));
uint32_t dataLen = 64;
int i;
memcpy( &ctx, &tt8_4way_ctx, sizeof(tt8_4way_ctx) );
for ( i = 0; i < TT8_FUNC_COUNT; i++ )
{
if (i == 0)
{
dataLen = 80;
vhashA = (uint64_t*)input;
vhashB = vhashX;
}
else
{
dataLen = 64;
if ( i % 2 == 0 )
{
vhashA = vhashY;
vhashB = vhashX;
}
else
{
vhashA = vhashX;
vhashB = vhashY;
}
}
switch ( permutation[i] )
{
case 0:
blake512_4way( &ctx.blake, vhashA, dataLen );
blake512_4way_close( &ctx.blake, vhashB );
if ( i == 7 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 1:
bmw512_4way( &ctx.bmw, vhashA, dataLen );
bmw512_4way_close( &ctx.bmw, vhashB );
if ( i == 7 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 2:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashA, dataLen<<3 );
update_and_final_groestl( &ctx.groestl, (char*)hash0,
(char*)hash0, dataLen<<3 );
reinit_groestl( &ctx.groestl );
update_and_final_groestl( &ctx.groestl, (char*)hash1,
(char*)hash1, dataLen<<3 );
reinit_groestl( &ctx.groestl );
update_and_final_groestl( &ctx.groestl, (char*)hash2,
(char*)hash2, dataLen<<3 );
reinit_groestl( &ctx.groestl );
update_and_final_groestl( &ctx.groestl, (char*)hash3,
(char*)hash3, dataLen<<3 );
if ( i != 7 )
mm256_interleave_4x64( vhashB,
hash0, hash1, hash2, hash3, dataLen<<3 );
break;
case 3:
skein512_4way( &ctx.skein, vhashA, dataLen );
skein512_4way_close( &ctx.skein, vhashB );
if ( i == 7 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 4:
jh512_4way( &ctx.jh, vhashA, dataLen );
jh512_4way_close( &ctx.jh, vhashB );
if ( i == 7 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 5:
keccak512_4way( &ctx.keccak, vhashA, dataLen );
keccak512_4way_close( &ctx.keccak, vhashB );
if ( i == 7 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 6:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashA, dataLen<<3 );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash0,
(const BitSequence *)hash0, dataLen );
memcpy( &ctx.luffa, &tt8_4way_ctx.luffa, sizeof(hashState_luffa) );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash1,
(const BitSequence*)hash1, dataLen );
memcpy( &ctx.luffa, &tt8_4way_ctx.luffa, sizeof(hashState_luffa) );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash2,
(const BitSequence*)hash2, dataLen );
memcpy( &ctx.luffa, &tt8_4way_ctx.luffa, sizeof(hashState_luffa) );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash3,
(const BitSequence*)hash3, dataLen );
if ( i != 7 )
mm256_interleave_4x64( vhashB,
hash0, hash1, hash2, hash3, dataLen<<3 );
break;
case 7:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashA, dataLen<<3 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash0,
(const byte*)hash0, dataLen );
memcpy( &ctx.cube, &tt8_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash1,
(const byte*)hash1, dataLen );
memcpy( &ctx.cube, &tt8_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash2,
(const byte*)hash2, dataLen );
memcpy( &ctx.cube, &tt8_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash3,
(const byte*)hash3, dataLen );
if ( i != 7 )
mm256_interleave_4x64( vhashB,
hash0, hash1, hash2, hash3, dataLen<<3 );
break;
default:
applog(LOG_ERR,"SWERR: timetravel invalid permutation");
break;
}
}
memcpy( output, hash0, 32 );
memcpy( output+32, hash1, 32 );
memcpy( output+64, hash2, 32 );
memcpy( output+96, hash3, 32 );
}
int scanhash_timetravel_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done )
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t endiandata[20] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
uint32_t n = pdata[19];
const uint32_t first_nonce = pdata[19];
uint32_t *nonces = work->nonces;
bool *found = work->nfound;
int num_found = 0;
uint32_t *noncep0 = vdata + 73; // 9*8 + 1
uint32_t *noncep1 = vdata + 75;
uint32_t *noncep2 = vdata + 77;
uint32_t *noncep3 = vdata + 79;
const uint32_t Htarg = ptarget[7];
volatile uint8_t *restart = &(work_restart[thr_id].restart);
int i;
if ( opt_benchmark )
ptarget[7] = 0x0cff;
for ( int k = 0; k < 19; k++ )
be32enc( &endiandata[k], pdata[k] );
const uint32_t timestamp = endiandata[17];
if ( timestamp != s_ntime )
{
const int steps = ( timestamp - TT8_FUNC_BASE_TIMESTAMP )
% TT8_FUNC_COUNT_PERMUTATIONS;
for ( i = 0; i < TT8_FUNC_COUNT; i++ )
permutation[i] = i;
for ( i = 0; i < steps; i++ )
tt8_next_permutation( permutation, permutation + TT8_FUNC_COUNT );
s_ntime = timestamp;
}
uint64_t *edata = (uint64_t*)endiandata;
mm256_interleave_4x64( (uint64_t*)vdata, edata, edata, edata, edata, 640 );
do
{
found[0] = found[1] = found[2] = found[3] = false;
be32enc( noncep0, n );
be32enc( noncep1, n+1 );
be32enc( noncep2, n+2 );
be32enc( noncep3, n+3 );
timetravel_4way_hash( hash, vdata );
pdata[19] = n;
if ( hash[7] <= Htarg && fulltest( hash, ptarget) )
{
found[0] = true;
num_found++;
nonces[0] = n;
work_set_target_ratio( work, hash );
}
if ( (hash+8)[7] <= Htarg && fulltest( hash+8, ptarget) )
{
found[1] = true;
num_found++;
nonces[1] = n+1;
work_set_target_ratio( work, hash+8 );
}
if ( (hash+16)[7] <= Htarg && fulltest( hash+16, ptarget) )
{
found[2] = true;
num_found++;
nonces[2] = n+2;
work_set_target_ratio( work, hash+16 );
}
if ( (hash+24)[7] <= Htarg && fulltest( hash+24, ptarget) )
{
found[3] = true;
num_found++;
nonces[3] = n+3;
work_set_target_ratio( work, hash+24 );
}
n += 4;
} while ( ( num_found == 0 ) && ( n < max_nonce ) && !(*restart) );
*hashes_done = n - first_nonce + 1;
return num_found;
}
#endif

78
algo/x11/timetravel-gate.c Normal file
View File

@@ -0,0 +1,78 @@
#include "timetravel-gate.h"
void tt8_set_target( struct work* work, double job_diff )
{
work_set_target( work, job_diff / (256.0 * opt_diff_factor) );
}
bool register_timetravel_algo( algo_gate_t* gate )
{
#ifdef TIMETRAVEL_4WAY
init_tt8_4way_ctx();
gate->scanhash = (void*)&scanhash_timetravel_4way;
gate->hash = (void*)&timetravel_4way_hash;
#else
init_tt8_ctx();
gate->scanhash = (void*)&scanhash_timetravel;
gate->hash = (void*)&timetravel_hash;
#endif
gate->set_target = (void*)&tt8_set_target;
gate->optimizations = SSE2_OPT | AES_OPT | AVX_OPT | AVX2_OPT | FOUR_WAY_OPT;
gate->get_max64 = (void*)&get_max64_0xffffLL;
return true;
};
inline void tt_swap( int *a, int *b )
{
int c = *a;
*a = *b;
*b = c;
}
inline void reverse( int *pbegin, int *pend )
{
while ( (pbegin != pend) && (pbegin != --pend) )
{
tt_swap( pbegin, pend );
pbegin++;
}
}
void tt8_next_permutation( int *pbegin, int *pend )
{
if ( pbegin == pend )
return;
int *i = pbegin;
++i;
if ( i == pend )
return;
i = pend;
--i;
while (1)
{
int *j = i;
--i;
if ( *i < *j )
{
int *k = pend;
while ( !(*i < *--k) ) /* do nothing */ ;
tt_swap( i, k );
reverse(j, pend);
return; // true
}
if ( i == pbegin )
{
reverse(pbegin, pend);
return; // false
}
// else?
}
}

40
algo/x11/timetravel-gate.h Normal file
View File

@@ -0,0 +1,40 @@
#ifndef TIMETRAVEL_GATE_H__
#define TIMETRAVEL_GATE_H__ 1
#include "algo-gate-api.h"
#include <stdint.h>
#if defined(HASH_4WAY) && defined(__AES__)
#define TIMETRAVEL_4WAY
#endif
// Machinecoin Genesis Timestamp
#define TT8_FUNC_BASE_TIMESTAMP 1389040865
#define TT8_FUNC_COUNT 8
#define TT8_FUNC_COUNT_PERMUTATIONS 40320
void tt8_next_permutation( int *pbegin, int *pend );
bool register_timetravel_algo( algo_gate_t* gate );
#if defined(TIMETRAVEL_4WAY)
void timetravel_4way_hash( void *state, const void *input );
int scanhash_timetravel_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done );
void init_tt8_4way_ctx();
#endif
void timetravel_hash( void *state, const void *input );
int scanhash_timetravel( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done );
void init_tt8_ctx();
#endif

311
algo/x11/timetravel.c Normal file
View File

@@ -0,0 +1,311 @@
#include "timetravel-gate.h"
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include <stdio.h>
#include "algo/blake/sph_blake.h"
#include "algo/bmw/sph_bmw.h"
#include "algo/jh/sph_jh.h"
#include "algo/keccak/sph_keccak.h"
#include "algo/skein/sph_skein.h"
#include "algo/luffa/sse2/luffa_for_sse2.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#ifdef NO_AES_NI
#include "algo/groestl/sph_groestl.h"
#else
#include "algo/groestl/aes_ni/hash-groestl.h"
#endif
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread int permutation[TT8_FUNC_COUNT] = { 0 };
typedef struct {
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;
#ifdef NO_AES_NI
sph_groestl512_context groestl;
#else
hashState_groestl groestl;
#endif
} tt_ctx_holder;
tt_ctx_holder tt_ctx __attribute__ ((aligned (64)));
__thread tt_ctx_holder tt_mid __attribute__ ((aligned (64)));
void init_tt8_ctx()
{
sph_blake512_init( &tt_ctx.blake );
sph_bmw512_init( &tt_ctx.bmw );
sph_skein512_init( &tt_ctx.skein );
sph_jh512_init( &tt_ctx.jh );
sph_keccak512_init( &tt_ctx.keccak );
init_luffa( &tt_ctx.luffa, 512 );
cubehashInit( &tt_ctx.cube, 512, 16, 32 );
#ifdef NO_AES_NI
sph_groestl512_init( &tt_ctx.groestl );
#else
init_groestl( &tt_ctx.groestl, 64 );
#endif
};
void timetravel_hash(void *output, const void *input)
{
uint32_t hash[ 16 * TT8_FUNC_COUNT ] __attribute__ ((aligned (64)));
uint32_t *hashA, *hashB;
tt_ctx_holder ctx __attribute__ ((aligned (64)));
uint32_t dataLen = 64;
uint32_t *work_data = (uint32_t *)input;
int i;
const int midlen = 64; // bytes
const int tail = 80 - midlen; // 16
memcpy( &ctx, &tt_ctx, sizeof(tt_ctx) );
for ( i = 0; i < TT8_FUNC_COUNT; i++ )
{
if (i == 0)
{
dataLen = 80;
hashA = work_data;
}
else
{
dataLen = 64;
hashA = &hash[16 * (i - 1)];
}
hashB = &hash[16 * i];
switch ( permutation[i] )
{
case 0:
if ( i == 0 )
{
memcpy( &ctx.blake, &tt_mid.blake, sizeof tt_mid.blake );
sph_blake512( &ctx.blake, input + midlen, tail );
sph_blake512_close( &ctx.blake, hashB );
}
else
{
sph_blake512( &ctx.blake, hashA, dataLen );
sph_blake512_close( &ctx.blake, hashB );
}
break;
case 1:
if ( i == 0 )
{
memcpy( &ctx.bmw, &tt_mid.bmw, sizeof tt_mid.bmw );
sph_bmw512( &ctx.bmw, input + midlen, tail );
sph_bmw512_close( &ctx.bmw, hashB );
}
else
{
sph_bmw512( &ctx.bmw, hashA, dataLen );
sph_bmw512_close( &ctx.bmw, hashB );
}
break;
case 2:
#ifdef NO_AES_NI
if ( i == 0 )
{
memcpy( &ctx.groestl, &tt_mid.groestl, sizeof tt_mid.groestl );
sph_groestl512( &ctx.groestl, input + midlen, tail );
sph_groestl512_close( &ctx.groestl, hashB );
}
else
{
sph_groestl512( &ctx.groestl, hashA, dataLen );
sph_groestl512_close( &ctx.groestl, hashB );
}
#else
// groestl midstate is slower
// if ( i == 0 )
// {
// memcpy( &ctx.groestl, &tt_mid.groestl, sizeof tt_mid.groestl );
// update_and_final_groestl( &ctx.groestl, (char*)hashB,
// (char*)input + midlen, tail*8 );
// }
// else
// {
update_and_final_groestl( &ctx.groestl, (char*)hashB,
(char*)hashA, dataLen*8 );
// }
#endif
break;
case 3:
if ( i == 0 )
{
memcpy( &ctx.skein, &tt_mid.skein, sizeof tt_mid.skein );
sph_skein512( &ctx.skein, input + midlen, tail );
sph_skein512_close( &ctx.skein, hashB );
}
else
{
sph_skein512( &ctx.skein, hashA, dataLen );
sph_skein512_close( &ctx.skein, hashB );
}
break;
case 4:
if ( i == 0 )
{
memcpy( &ctx.jh, &tt_mid.jh, sizeof tt_mid.jh );
sph_jh512( &ctx.jh, input + midlen, tail );
sph_jh512_close( &ctx.jh, hashB );
}
else
{
sph_jh512( &ctx.jh, hashA, dataLen );
sph_jh512_close( &ctx.jh, hashB);
}
break;
case 5:
if ( i == 0 )
{
memcpy( &ctx.keccak, &tt_mid.keccak, sizeof tt_mid.keccak );
sph_keccak512( &ctx.keccak, input + midlen, tail );
sph_keccak512_close( &ctx.keccak, hashB );
}
else
{
sph_keccak512( &ctx.keccak, hashA, dataLen );
sph_keccak512_close( &ctx.keccak, hashB );
}
break;
case 6:
if ( i == 0 )
{
memcpy( &ctx.luffa, &tt_mid.luffa, sizeof tt_mid.luffa );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hashB,
(const BitSequence *)input + 64, 16 );
}
else
{
update_and_final_luffa( &ctx.luffa, (BitSequence*)hashB,
(const BitSequence *)hashA, dataLen );
}
break;
case 7:
if ( i == 0 )
{
memcpy( &ctx.cube, &tt_mid.cube, sizeof tt_mid.cube );
cubehashUpdateDigest( &ctx.cube, (byte*)hashB,
(const byte*)input + midlen, tail );
}
else
{
cubehashUpdateDigest( &ctx.cube, (byte*)hashB, (const byte*)hashA,
dataLen );
}
break;
default:
break;
}
}
memcpy(output, &hash[16 * (TT8_FUNC_COUNT - 1)], 32);
}
int scanhash_timetravel( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done )
{
uint32_t _ALIGN(64) hash[8];
uint32_t _ALIGN(64) endiandata[20];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t Htarg = ptarget[7];
const uint32_t first_nonce = pdata[19];
uint32_t nonce = first_nonce;
volatile uint8_t *restart = &(work_restart[thr_id].restart);
int i;
if (opt_benchmark)
ptarget[7] = 0x0cff;
for (int k=0; k < 19; k++)
be32enc(&endiandata[k], pdata[k]);
const uint32_t timestamp = endiandata[17];
if ( timestamp != s_ntime )
{
const int steps = ( timestamp - TT8_FUNC_BASE_TIMESTAMP )
% TT8_FUNC_COUNT_PERMUTATIONS;
for ( i = 0; i < TT8_FUNC_COUNT; i++ )
permutation[i] = i;
for ( i = 0; i < steps; i++ )
tt8_next_permutation( permutation, permutation + TT8_FUNC_COUNT );
s_ntime = timestamp;
// do midstate precalc for first function
switch ( permutation[0] )
{
case 0:
memcpy( &tt_mid.blake, &tt_ctx.blake, sizeof(tt_mid.blake) );
sph_blake512( &tt_mid.blake, endiandata, 64 );
break;
case 1:
memcpy( &tt_mid.bmw, &tt_ctx.bmw, sizeof(tt_mid.bmw) );
sph_bmw512( &tt_mid.bmw, endiandata, 64 );
break;
case 2:
#ifdef NO_AES_NI
memcpy( &tt_mid.groestl, &tt_ctx.groestl, sizeof(tt_mid.groestl ) );
sph_groestl512( &tt_mid.groestl, endiandata, 64 );
#else
// groestl midstate is slower
// memcpy( &tt_mid.groestl, &tt_ctx.groestl, sizeof(tt_mid.groestl ) );
// update_groestl( &tt_mid.groestl, (char*)endiandata, 64*8 );
#endif
break;
case 3:
memcpy( &tt_mid.skein, &tt_ctx.skein, sizeof(tt_mid.skein ) );
sph_skein512( &tt_mid.skein, endiandata, 64 );
break;
case 4:
memcpy( &tt_mid.jh, &tt_ctx.jh, sizeof(tt_mid.jh ) );
sph_jh512( &tt_mid.jh, endiandata, 64 );
break;
case 5:
memcpy( &tt_mid.keccak, &tt_ctx.keccak, sizeof(tt_mid.keccak ) );
sph_keccak512( &tt_mid.keccak, endiandata, 64 );
break;
case 6:
memcpy( &tt_mid.luffa, &tt_ctx.luffa, sizeof(tt_mid.luffa ) );
update_luffa( &tt_mid.luffa, (const BitSequence*)endiandata, 64 );
break;
case 7:
memcpy( &tt_mid.cube, &tt_ctx.cube, sizeof(tt_mid.cube ) );
cubehashUpdate( &tt_mid.cube, (const byte*)endiandata, 64 );
break;
default:
break;
}
}
do {
be32enc( &endiandata[19], nonce );
timetravel_hash( hash, endiandata );
if ( hash[7] <= Htarg && fulltest( hash, ptarget) )
{
work_set_target_ratio( work, hash );
pdata[19] = nonce;
*hashes_done = pdata[19] - first_nonce;
work_set_target_ratio( work, hash );
return 1;
}
nonce++;
} while (nonce < max_nonce && !(*restart));
pdata[19] = nonce;
*hashes_done = pdata[19] - first_nonce + 1;
return 0;
}

316
algo/x11/timetravel10-4way.c Normal file
View File

@@ -0,0 +1,316 @@
#include "timetravel10-gate.h"
#if defined(__AVX2__) && defined(__AES__)
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include <stdio.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/skein/skein-hash-4way.h"
#include "algo/jh/jh-hash-4way.h"
#include "algo/keccak/keccak-hash-4way.h"
#include "algo/luffa/sse2/luffa_for_sse2.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/simd/sse2/nist.h"
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread int permutation[TT10_FUNC_COUNT] = { 0 };
typedef struct {
blake512_4way_context blake;
bmw512_4way_context bmw;
hashState_groestl groestl;
skein512_4way_context skein;
jh512_4way_context jh;
keccak512_4way_context keccak;
hashState_luffa luffa;
cubehashParam cube;
sph_shavite512_context shavite;
hashState_sd simd;
} tt10_4way_ctx_holder;
tt10_4way_ctx_holder tt10_4way_ctx __attribute__ ((aligned (64)));
void init_tt10_4way_ctx()
{
blake512_4way_init( &tt10_4way_ctx.blake );
bmw512_4way_init( &tt10_4way_ctx.bmw );
init_groestl( &tt10_4way_ctx.groestl, 64 );
skein512_4way_init( &tt10_4way_ctx.skein );
jh512_4way_init( &tt10_4way_ctx.jh );
keccak512_4way_init( &tt10_4way_ctx.keccak );
init_luffa( &tt10_4way_ctx.luffa, 512 );
cubehashInit( &tt10_4way_ctx.cube, 512, 16, 32 );
sph_shavite512_init( &tt10_4way_ctx.shavite );
init_sd( &tt10_4way_ctx.simd, 512 );
};
void timetravel10_4way_hash(void *output, const void *input)
{
uint64_t hash0[8] __attribute__ ((aligned (64)));
uint64_t hash1[8] __attribute__ ((aligned (64)));
uint64_t hash2[8] __attribute__ ((aligned (64)));
uint64_t hash3[8] __attribute__ ((aligned (64)));
uint64_t vhashX[8*4] __attribute__ ((aligned (64)));
uint64_t vhashY[8*4] __attribute__ ((aligned (64)));
uint64_t *vhashA, *vhashB;
tt10_4way_ctx_holder ctx __attribute__ ((aligned (64)));
uint32_t dataLen = 64;
int i;
memcpy( &ctx, &tt10_4way_ctx, sizeof(tt10_4way_ctx) );
for ( i = 0; i < TT10_FUNC_COUNT; i++ )
{
if (i == 0)
{
dataLen = 80;
vhashA = (uint64_t*)input;
vhashB = vhashX;
}
else
{
dataLen = 64;
if ( i % 2 == 0 )
{
vhashA = vhashY;
vhashB = vhashX;
}
else
{
vhashA = vhashX;
vhashB = vhashY;
}
}
switch ( permutation[i] )
{
case 0:
blake512_4way( &ctx.blake, vhashA, dataLen );
blake512_4way_close( &ctx.blake, vhashB );
if ( i == 9 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 1:
bmw512_4way( &ctx.bmw, vhashA, dataLen );
bmw512_4way_close( &ctx.bmw, vhashB );
if ( i == 9 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 2:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashA, dataLen<<3 );
update_and_final_groestl( &ctx.groestl, (char*)hash0,
(char*)hash0, dataLen<<3 );
reinit_groestl( &ctx.groestl );
update_and_final_groestl( &ctx.groestl, (char*)hash1,
(char*)hash1, dataLen<<3 );
reinit_groestl( &ctx.groestl );
update_and_final_groestl( &ctx.groestl, (char*)hash2,
(char*)hash2, dataLen<<3 );
reinit_groestl( &ctx.groestl );
update_and_final_groestl( &ctx.groestl, (char*)hash3,
(char*)hash3, dataLen<<3 );
if ( i != 9 )
mm256_interleave_4x64( vhashB,
hash0, hash1, hash2, hash3, dataLen<<3 );
break;
case 3:
skein512_4way( &ctx.skein, vhashA, dataLen );
skein512_4way_close( &ctx.skein, vhashB );
if ( i == 9 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 4:
jh512_4way( &ctx.jh, vhashA, dataLen );
jh512_4way_close( &ctx.jh, vhashB );
if ( i == 9 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 5:
keccak512_4way( &ctx.keccak, vhashA, dataLen );
keccak512_4way_close( &ctx.keccak, vhashB );
if ( i == 9 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashB, dataLen<<3 );
break;
case 6:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashA, dataLen<<3 );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash0,
(const BitSequence *)hash0, dataLen );
memcpy( &ctx.luffa, &tt10_4way_ctx.luffa, sizeof(hashState_luffa) );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash1,
(const BitSequence*)hash1, dataLen );
memcpy( &ctx.luffa, &tt10_4way_ctx.luffa, sizeof(hashState_luffa) );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash2,
(const BitSequence*)hash2, dataLen );
memcpy( &ctx.luffa, &tt10_4way_ctx.luffa, sizeof(hashState_luffa) );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash3,
(const BitSequence*)hash3, dataLen );
if ( i != 9 )
mm256_interleave_4x64( vhashB,
hash0, hash1, hash2, hash3, dataLen<<3 );
break;
case 7:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashA, dataLen<<3 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash0,
(const byte*)hash0, dataLen );
memcpy( &ctx.cube, &tt10_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash1,
(const byte*)hash1, dataLen );
memcpy( &ctx.cube, &tt10_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash2,
(const byte*)hash2, dataLen );
memcpy( &ctx.cube, &tt10_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash3,
(const byte*)hash3, dataLen );
if ( i != 9 )
mm256_interleave_4x64( vhashB,
hash0, hash1, hash2, hash3, dataLen<<3 );
break;
case 8:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashA, dataLen<<3 );
sph_shavite512( &ctx.shavite, hash0, dataLen );
sph_shavite512_close( &ctx.shavite, hash0 );
memcpy( &ctx.shavite, &tt10_4way_ctx.shavite, sizeof ctx.shavite );
sph_shavite512( &ctx.shavite, hash1, dataLen );
sph_shavite512_close( &ctx.shavite, hash1 );
memcpy( &ctx.shavite, &tt10_4way_ctx.shavite, sizeof ctx.shavite );
sph_shavite512( &ctx.shavite, hash2, dataLen );
sph_shavite512_close( &ctx.shavite, hash2 );
memcpy( &ctx.shavite, &tt10_4way_ctx.shavite, sizeof ctx.shavite );
sph_shavite512( &ctx.shavite, hash3, dataLen );
sph_shavite512_close( &ctx.shavite, hash3 );
if ( i != 9 )
mm256_interleave_4x64( vhashB,
hash0, hash1, hash2, hash3, dataLen<<3 );
break;
case 9:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhashA, dataLen<<3 );
update_final_sd( &ctx.simd, (BitSequence *)hash0,
(const BitSequence *)hash0, dataLen<<3 );
memcpy( &ctx.simd, &tt10_4way_ctx.simd, sizeof ctx.simd );
update_final_sd( &ctx.simd, (BitSequence *)hash1,
(const BitSequence *)hash1, dataLen<<3 );
memcpy( &ctx.simd, &tt10_4way_ctx.simd, sizeof ctx.simd );
update_final_sd( &ctx.simd, (BitSequence *)hash2,
(const BitSequence *)hash2, dataLen<<3 );
memcpy( &ctx.simd, &tt10_4way_ctx.simd, sizeof ctx.simd );
update_final_sd( &ctx.simd, (BitSequence *)hash3,
(const BitSequence *)hash3, dataLen<<3 );
if ( i != 9 )
mm256_interleave_4x64( vhashB,
hash0, hash1, hash2, hash3, dataLen<<3 );
break;
default:
applog(LOG_ERR,"SWERR: timetravel invalid permutation");
break;
}
}
memcpy( output, hash0, 32 );
memcpy( output+32, hash1, 32 );
memcpy( output+64, hash2, 32 );
memcpy( output+96, hash3, 32 );
}
int scanhash_timetravel10_4way( int thr_id, struct work *work,
uint32_t max_nonce, uint64_t *hashes_done )
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t endiandata[20] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
uint32_t n = pdata[19];
const uint32_t first_nonce = pdata[19];
uint32_t *nonces = work->nonces;
bool *found = work->nfound;
int num_found = 0;
uint32_t *noncep0 = vdata + 73; // 9*8 + 1
uint32_t *noncep1 = vdata + 75;
uint32_t *noncep2 = vdata + 77;
uint32_t *noncep3 = vdata + 79;
const uint32_t Htarg = ptarget[7];
volatile uint8_t *restart = &(work_restart[thr_id].restart);
int i;
if ( opt_benchmark )
ptarget[7] = 0x0cff;
for ( int k = 0; k < 19; k++ )
be32enc( &endiandata[k], pdata[k] );
const uint32_t timestamp = endiandata[17];
if ( timestamp != s_ntime )
{
const int steps = ( timestamp - TT10_FUNC_BASE_TIMESTAMP )
% TT10_FUNC_COUNT_PERMUTATIONS;
for ( i = 0; i < TT10_FUNC_COUNT; i++ )
permutation[i] = i;
for ( i = 0; i < steps; i++ )
tt10_next_permutation( permutation, permutation + TT10_FUNC_COUNT );
s_ntime = timestamp;
}
uint64_t *edata = (uint64_t*)endiandata;
mm256_interleave_4x64( (uint64_t*)vdata, edata, edata, edata, edata, 640 );
do
{
found[0] = found[1] = found[2] = found[3] = false;
be32enc( noncep0, n );
be32enc( noncep1, n+1 );
be32enc( noncep2, n+2 );
be32enc( noncep3, n+3 );
timetravel10_4way_hash( hash, vdata );
pdata[19] = n;
if ( hash[7] <= Htarg && fulltest( hash, ptarget) )
{
found[0] = true;
num_found++;
nonces[0] = n;
work_set_target_ratio( work, hash );
}
if ( (hash+8)[7] <= Htarg && fulltest( hash+8, ptarget) )
{
found[1] = true;
num_found++;
nonces[1] = n+1;
work_set_target_ratio( work, hash+8 );
}
if ( (hash+16)[7] <= Htarg && fulltest( hash+16, ptarget) )
{
found[2] = true;
num_found++;
nonces[2] = n+2;
work_set_target_ratio( work, hash+16 );
}
if ( (hash+24)[7] <= Htarg && fulltest( hash+24, ptarget) )
{
found[3] = true;
num_found++;
nonces[3] = n+3;
work_set_target_ratio( work, hash+24 );
}
n += 4;
} while ( ( num_found == 0 ) && ( n < max_nonce ) && !(*restart) );
*hashes_done = n - first_nonce + 1;
return num_found;
}
#endif

78
algo/x11/timetravel10-gate.c Normal file
View File

@@ -0,0 +1,78 @@
#include "timetravel10-gate.h"
void tt10_set_target( struct work* work, double job_diff )
{
work_set_target( work, job_diff / (256.0 * opt_diff_factor) );
}
bool register_timetravel10_algo( algo_gate_t* gate )
{
#ifdef TIMETRAVEL10_4WAY
init_tt10_4way_ctx();
gate->scanhash = (void*)&scanhash_timetravel10_4way;
gate->hash = (void*)&timetravel10_4way_hash;
#else
init_tt10_ctx();
gate->scanhash = (void*)&scanhash_timetravel10;
gate->hash = (void*)&timetravel10_hash;
#endif
gate->set_target = (void*)&tt10_set_target;
gate->optimizations = SSE2_OPT | AES_OPT | AVX_OPT | AVX2_OPT | FOUR_WAY_OPT;
gate->get_max64 = (void*)&get_max64_0xffffLL;
return true;
};
inline void tt10_swap( int *a, int *b )
{
int c = *a;
*a = *b;
*b = c;
}
inline void reverse( int *pbegin, int *pend )
{
while ( (pbegin != pend) && (pbegin != --pend) )
{
tt10_swap( pbegin, pend );
pbegin++;
}
}
void tt10_next_permutation( int *pbegin, int *pend )
{
if ( pbegin == pend )
return;
int *i = pbegin;
++i;
if ( i == pend )
return;
i = pend;
--i;
while (1)
{
int *j = i;
--i;
if ( *i < *j )
{
int *k = pend;
while ( !(*i < *--k) ) /* do nothing */ ;
tt10_swap( i, k );
reverse(j, pend);
return; // true
}
if ( i == pbegin )
{
reverse(pbegin, pend);
return; // false
}
// else?
}
}

39
algo/x11/timetravel10-gate.h Normal file
View File

@@ -0,0 +1,39 @@
#ifndef TIMETRAVEL10_GATE_H__
#define TIMETRAVEL10_GATE_H__ 1
#include "algo-gate-api.h"
#include <stdint.h>
#if defined(HASH_4WAY) && defined(__AES__)
#define TIMETRAVEL10_4WAY
#endif
// BitCore Genesis Timestamp
#define TT10_FUNC_BASE_TIMESTAMP 1492973331U
#define TT10_FUNC_COUNT 10
#define TT10_FUNC_COUNT_PERMUTATIONS 40320
void tt10_next_permutation( int *pbegin, int *pend );
bool register_timetravel10_algo( algo_gate_t* gate );
#if defined(TIMETRAVEL10_4WAY)
void timetravel10_4way_hash( void *state, const void *input );
int scanhash_timetravel10_4way( int thr_id, struct work *work,
uint32_t max_nonce, uint64_t *hashes_done );
void init_tt10_4way_ctx();
#endif
void timetravel10_hash( void *state, const void *input );
int scanhash_timetravel10( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done );
void init_tt10_ctx();
#endif

349
algo/x11/timetravel10.c Normal file
View File

@@ -0,0 +1,349 @@
#include "timetravel10-gate.h"
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include <stdio.h>
#include "algo/blake/sph_blake.h"
#include "algo/bmw/sph_bmw.h"
#include "algo/jh/sph_jh.h"
#include "algo/keccak/sph_keccak.h"
#include "algo/skein/sph_skein.h"
#include "algo/luffa/sse2/luffa_for_sse2.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/simd/sse2/nist.h"
#ifdef NO_AES_NI
#include "algo/groestl/sph_groestl.h"
#else
#include "algo/groestl/aes_ni/hash-groestl.h"
#endif
static __thread uint32_t s_ntime = UINT32_MAX;
static __thread int permutation[TT10_FUNC_COUNT] = { 0 };
typedef struct {
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;
#ifdef NO_AES_NI
sph_groestl512_context groestl;
#else
hashState_groestl groestl;
#endif
} tt10_ctx_holder;
tt10_ctx_holder tt10_ctx __attribute__ ((aligned (64)));
__thread tt10_ctx_holder tt10_mid __attribute__ ((aligned (64)));
void init_tt10_ctx()
{
sph_blake512_init( &tt10_ctx.blake );
sph_bmw512_init( &tt10_ctx.bmw );
sph_skein512_init( &tt10_ctx.skein );
sph_jh512_init( &tt10_ctx.jh );
sph_keccak512_init( &tt10_ctx.keccak );
init_luffa( &tt10_ctx.luffa, 512 );
cubehashInit( &tt10_ctx.cube, 512, 16, 32 );
sph_shavite512_init( &tt10_ctx.shavite );
init_sd( &tt10_ctx.simd, 512 );
#ifdef NO_AES_NI
sph_groestl512_init( &tt10_ctx.groestl );
#else
init_groestl( &tt10_ctx.groestl, 64 );
#endif
};
void timetravel10_hash(void *output, const void *input)
{
uint32_t hash[ 16 * TT10_FUNC_COUNT ] __attribute__ ((aligned (64)));
uint32_t *hashA, *hashB;
tt10_ctx_holder ctx __attribute__ ((aligned (64)));
uint32_t dataLen = 64;
uint32_t *work_data = (uint32_t *)input;
int i;
const int midlen = 64; // bytes
const int tail = 80 - midlen; // 16
memcpy( &ctx, &tt10_ctx, sizeof(tt10_ctx) );
for ( i = 0; i < TT10_FUNC_COUNT; i++ )
{
if (i == 0)
{
dataLen = 80;
hashA = work_data;
}
else
{
dataLen = 64;
hashA = &hash[16 * (i - 1)];
}
hashB = &hash[16 * i];
switch ( permutation[i] )
{
case 0:
if ( i == 0 )
{
memcpy( &ctx.blake, &tt10_mid.blake, sizeof tt10_mid.blake );
sph_blake512( &ctx.blake, input + midlen, tail );
sph_blake512_close( &ctx.blake, hashB );
}
else
{
sph_blake512( &ctx.blake, hashA, dataLen );
sph_blake512_close( &ctx.blake, hashB );
}
break;
case 1:
if ( i == 0 )
{
memcpy( &ctx.bmw, &tt10_mid.bmw, sizeof tt10_mid.bmw );
sph_bmw512( &ctx.bmw, input + midlen, tail );
sph_bmw512_close( &ctx.bmw, hashB );
}
else
{
sph_bmw512( &ctx.bmw, hashA, dataLen );
sph_bmw512_close( &ctx.bmw, hashB );
}
break;
case 2:
#ifdef NO_AES_NI
if ( i == 0 )
{
memcpy( &ctx.groestl, &tt10_mid.groestl, sizeof tt10_mid.groestl );
sph_groestl512( &ctx.groestl, input + midlen, tail );
sph_groestl512_close( &ctx.groestl, hashB );
}
else
{
sph_groestl512( &ctx.groestl, hashA, dataLen );
sph_groestl512_close( &ctx.groestl, hashB );
}
#else
// groestl midstate is slower
// if ( i == 0 )
// {
// memcpy( &ctx.groestl, &tt10_mid.groestl, sizeof tt10_mid.groestl );
// update_and_final_groestl( &ctx.groestl, (char*)hashB,
// (char*)input + midlen, tail*8 );
// }
// else
// {
update_and_final_groestl( &ctx.groestl, (char*)hashB,
(char*)hashA, dataLen*8 );
// }
#endif
break;
case 3:
if ( i == 0 )
{
memcpy( &ctx.skein, &tt10_mid.skein, sizeof tt10_mid.skein );
sph_skein512( &ctx.skein, input + midlen, tail );
sph_skein512_close( &ctx.skein, hashB );
}
else
{
sph_skein512( &ctx.skein, hashA, dataLen );
sph_skein512_close( &ctx.skein, hashB );
}
break;
case 4:
if ( i == 0 )
{
memcpy( &ctx.jh, &tt10_mid.jh, sizeof tt10_mid.jh );
sph_jh512( &ctx.jh, input + midlen, tail );
sph_jh512_close( &ctx.jh, hashB );
}
else
{
sph_jh512( &ctx.jh, hashA, dataLen );
sph_jh512_close( &ctx.jh, hashB);
}
break;
case 5:
if ( i == 0 )
{
memcpy( &ctx.keccak, &tt10_mid.keccak, sizeof tt10_mid.keccak );
sph_keccak512( &ctx.keccak, input + midlen, tail );
sph_keccak512_close( &ctx.keccak, hashB );
}
else
{
sph_keccak512( &ctx.keccak, hashA, dataLen );
sph_keccak512_close( &ctx.keccak, hashB );
}
break;
case 6:
if ( i == 0 )
{
memcpy( &ctx.luffa, &tt10_mid.luffa, sizeof tt10_mid.luffa );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hashB,
(const BitSequence *)input + 64, 16 );
}
else
{
update_and_final_luffa( &ctx.luffa, (BitSequence*)hashB,
(const BitSequence *)hashA, dataLen );
}
break;
case 7:
if ( i == 0 )
{
memcpy( &ctx.cube, &tt10_mid.cube, sizeof tt10_mid.cube );
cubehashUpdateDigest( &ctx.cube, (byte*)hashB,
(const byte*)input + midlen, tail );
}
else
{
cubehashUpdateDigest( &ctx.cube, (byte*)hashB, (const byte*)hashA,
dataLen );
}
break;
case 8:
if ( i == 0 )
{
memcpy( &ctx.shavite, &tt10_mid.shavite, sizeof tt10_mid.shavite );
sph_shavite512( &ctx.shavite, input + midlen, tail*8 );
sph_shavite512_close( &ctx.shavite, hashB );
}
else
{
sph_shavite512( &ctx.shavite, hashA, dataLen );
sph_shavite512_close( &ctx.shavite, hashB );
}
break;
case 9:
if ( i == 0 )
{
memcpy( &ctx.simd, &tt10_mid.simd, sizeof tt10_mid.simd );
update_final_sd( &ctx.simd, (BitSequence *)hashB,
(const BitSequence *)input + midlen, tail*8 );
}
else
{
update_final_sd( &ctx.simd, (BitSequence *)hashB,
(const BitSequence *)hashA, dataLen*8 );
}
break;
default:
break;
}
}
memcpy(output, &hash[16 * (TT10_FUNC_COUNT - 1)], 32);
}
int scanhash_timetravel10( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done )
{
uint32_t _ALIGN(64) hash[8];
uint32_t _ALIGN(64) endiandata[20];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t Htarg = ptarget[7];
const uint32_t first_nonce = pdata[19];
uint32_t nonce = first_nonce;
volatile uint8_t *restart = &(work_restart[thr_id].restart);
int i;
if (opt_benchmark)
ptarget[7] = 0x0cff;
for (int k=0; k < 19; k++)
be32enc(&endiandata[k], pdata[k]);
const uint32_t timestamp = endiandata[17];
if ( timestamp != s_ntime )
{
const int steps = ( timestamp - TT10_FUNC_BASE_TIMESTAMP )
% TT10_FUNC_COUNT_PERMUTATIONS;
for ( i = 0; i < TT10_FUNC_COUNT; i++ )
permutation[i] = i;
for ( i = 0; i < steps; i++ )
tt10_next_permutation( permutation, permutation + TT10_FUNC_COUNT );
s_ntime = timestamp;
// do midstate precalc for first function
switch ( permutation[0] )
{
case 0:
memcpy( &tt10_mid.blake, &tt10_ctx.blake, sizeof(tt10_mid.blake) );
sph_blake512( &tt10_mid.blake, endiandata, 64 );
break;
case 1:
memcpy( &tt10_mid.bmw, &tt10_ctx.bmw, sizeof(tt10_mid.bmw) );
sph_bmw512( &tt10_mid.bmw, endiandata, 64 );
break;
case 2:
#ifdef NO_AES_NI
memcpy( &tt10_mid.groestl, &tt10_ctx.groestl, sizeof(tt10_mid.groestl ) );
sph_groestl512( &tt10_mid.groestl, endiandata, 64 );
#else
// groestl midstate is slower
// memcpy( &tt10_mid.groestl, &tt10_ctx.groestl, sizeof(tt10_mid.groestl ) );
// update_groestl( &tt10_mid.groestl, (char*)endiandata, 64*8 );
#endif
break;
case 3:
memcpy( &tt10_mid.skein, &tt10_ctx.skein, sizeof(tt10_mid.skein ) );
sph_skein512( &tt10_mid.skein, endiandata, 64 );
break;
case 4:
memcpy( &tt10_mid.jh, &tt10_ctx.jh, sizeof(tt10_mid.jh ) );
sph_jh512( &tt10_mid.jh, endiandata, 64 );
break;
case 5:
memcpy( &tt10_mid.keccak, &tt10_ctx.keccak, sizeof(tt10_mid.keccak ) );
sph_keccak512( &tt10_mid.keccak, endiandata, 64 );
break;
case 6:
memcpy( &tt10_mid.luffa, &tt10_ctx.luffa, sizeof(tt10_mid.luffa ) );
update_luffa( &tt10_mid.luffa, (const BitSequence*)endiandata, 64 );
break;
case 7:
memcpy( &tt10_mid.cube, &tt10_ctx.cube, sizeof(tt10_mid.cube ) );
cubehashUpdate( &tt10_mid.cube, (const byte*)endiandata, 64 );
break;
case 8:
memcpy( &tt10_mid.shavite, &tt10_ctx.shavite, sizeof(tt10_mid.shavite ) );
sph_shavite512( &tt10_mid.shavite, endiandata, 64 );
break;
case 9:
memcpy( &tt10_mid.simd, &tt10_ctx.simd, sizeof(tt10_mid.simd ) );
update_sd( &tt10_mid.simd, (const BitSequence *)endiandata, 512 );
break;
default:
break;
}
}
do {
be32enc( &endiandata[19], nonce );
timetravel10_hash( hash, endiandata );
if ( hash[7] <= Htarg && fulltest( hash, ptarget) )
{
work_set_target_ratio( work, hash );
pdata[19] = nonce;
work_set_target_ratio( work, hash );
*hashes_done = pdata[19] - first_nonce;
return 1;
}
nonce++;
} while (nonce < max_nonce && !(*restart));
pdata[19] = nonce;
*hashes_done = pdata[19] - first_nonce + 1;
return 0;
}

12
algo/x11/x11.c
View File

@@ -179,6 +179,7 @@ int scanhash_x11( int thr_id, struct work *work, uint32_t max_nonce,
if ( fulltest( hash64, ptarget ) )
{
*hashes_done = n - first_nonce + 1;
work_set_target_ratio( work, hash64 );
return true;
}
}
@@ -189,14 +190,3 @@ int scanhash_x11( int thr_id, struct work *work, uint32_t max_nonce,
pdata[19] = n;
return 0;
}
/*
bool register_x11_algo( algo_gate_t* gate )
{
gate->optimizations = SSE2_OPT | AES_OPT | AVX_OPT | AVX2_OPT;
init_x11_ctx();
gate->scanhash = (void*)&scanhash_x11;
gate->hash = (void*)&x11_hash;
gate->get_max64 = (void*)&get_max64_0x3ffff;
return true;
};
*/

340
algo/x11/x11evo-4way.c Normal file
View File

@@ -0,0 +1,340 @@
#include "cpuminer-config.h"
#include "x11evo-gate.h"
#if defined(__AVX2__) && defined(__AES__)
#include <string.h>
#include <stdint.h>
#include <compat/portable_endian.h>
#include "algo/blake/blake-hash-4way.h"
#include "algo/bmw/bmw-hash-4way.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/sph_luffa.h"
#include "algo/cubehash/sph_cubehash.h"
#include "algo/shavite/sph_shavite.h"
#include "algo/simd/sph_simd.h"
#include "algo/groestl/aes_ni/hash-groestl.h"
#include "algo/echo/aes_ni/hash_api.h"
#include "algo/luffa/sse2/luffa_for_sse2.h"
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/simd/sse2/nist.h"
typedef struct {
blake512_4way_context blake;
bmw512_4way_context bmw;
hashState_groestl groestl;
skein512_4way_context skein;
jh512_4way_context jh;
keccak512_4way_context keccak;
hashState_luffa luffa;
cubehashParam cube;
sph_shavite512_context shavite;
hashState_sd simd;
hashState_echo echo;
} x11evo_4way_ctx_holder;
static x11evo_4way_ctx_holder x11evo_4way_ctx __attribute__ ((aligned (64)));
void init_x11evo_4way_ctx()
{
blake512_4way_init( &x11evo_4way_ctx.blake );
bmw512_4way_init( &x11evo_4way_ctx.bmw );
init_groestl( &x11evo_4way_ctx.groestl, 64 );
skein512_4way_init( &x11evo_4way_ctx.skein );
jh512_4way_init( &x11evo_4way_ctx.jh );
keccak512_4way_init( &x11evo_4way_ctx.keccak );
init_luffa( &x11evo_4way_ctx.luffa, 512 );
cubehashInit( &x11evo_4way_ctx.cube, 512, 16, 32 );
sph_shavite512_init( &x11evo_4way_ctx.shavite );
init_sd( &x11evo_4way_ctx.simd, 512 );
init_echo( &x11evo_4way_ctx.echo, 512 );
}
static char hashOrder[X11EVO_FUNC_COUNT + 1] = { 0 };
static __thread uint32_t s_ntime = UINT32_MAX;
void x11evo_4way_hash( void *state, const void *input )
{
uint32_t hash0[16] __attribute__ ((aligned (64)));
uint32_t hash1[16] __attribute__ ((aligned (64)));
uint32_t hash2[16] __attribute__ ((aligned (64)));
uint32_t hash3[16] __attribute__ ((aligned (64)));
uint32_t vhash[16*4] __attribute__ ((aligned (64)));
x11evo_4way_ctx_holder ctx __attribute__ ((aligned (64)));
memcpy( &ctx, &x11evo_4way_ctx, sizeof(x11evo_4way_ctx) );
if ( s_seq == -1 )
{
uint32_t *data = (uint32_t*) input;
const uint32_t ntime = data[17];
evo_twisted_code( ntime, hashOrder );
}
int i;
int len = strlen( hashOrder );
for ( i = 0; i < len; i++ )
{
char elem = hashOrder[i];
uint8_t idx;
if ( elem >= 'A' )
idx = elem - 'A' + 10;
else
idx = elem - '0';
// int size = 64;
switch ( idx )
{
case 0:
blake512_4way( &ctx.blake, input, 80 );
blake512_4way_close( &ctx.blake, vhash );
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
break;
case 1:
bmw512_4way( &ctx.bmw, vhash, 64 );
bmw512_4way_close( &ctx.bmw, vhash );
if ( i >= len-1 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
break;
case 2:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
update_and_final_groestl( &ctx.groestl, (char*)hash0,
(char*)hash0, 512 );
reinit_groestl( &ctx.groestl );
update_and_final_groestl( &ctx.groestl, (char*)hash1,
(char*)hash1, 512 );
reinit_groestl( &ctx.groestl );
update_and_final_groestl( &ctx.groestl, (char*)hash2,
(char*)hash2, 512 );
reinit_groestl( &ctx.groestl );
update_and_final_groestl( &ctx.groestl, (char*)hash3,
(char*)hash3, 512 );
if ( i < len-1 )
mm256_interleave_4x64( vhash,
hash0, hash1, hash2, hash3, 64<<3 );
break;
case 3:
skein512_4way( &ctx.skein, vhash, 64 );
skein512_4way_close( &ctx.skein, vhash );
if ( i >= len-1 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
break;
case 4:
jh512_4way( &ctx.jh, vhash, 64 );
jh512_4way_close( &ctx.jh, vhash );
if ( i >= len-1 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
break;
case 5:
keccak512_4way( &ctx.keccak, vhash, 64 );
keccak512_4way_close( &ctx.keccak, vhash );
if ( i >= len-1 )
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
break;
case 6:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash0,
(const BitSequence*)hash0, 64 );
memcpy( &ctx.luffa, &x11evo_4way_ctx.luffa,
sizeof(hashState_luffa) );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash1,
(const BitSequence*)hash1, 64 );
memcpy( &ctx.luffa, &x11evo_4way_ctx.luffa,
sizeof(hashState_luffa) );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash2,
(const BitSequence*)hash2, 64 );
memcpy( &ctx.luffa, &x11evo_4way_ctx.luffa,
sizeof(hashState_luffa) );
update_and_final_luffa( &ctx.luffa, (BitSequence*)hash3,
(const BitSequence*)hash3, 64 );
if ( i < len-1 )
mm256_interleave_4x64( vhash,
hash0, hash1, hash2, hash3, 64<<3 );
break;
case 7:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
cubehashUpdateDigest( &ctx.cube, (byte*)hash0,
(const byte*) hash0, 64 );
memcpy( &ctx.cube, &x11evo_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash1,
(const byte*) hash1, 64 );
memcpy( &ctx.cube, &x11evo_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash2,
(const byte*) hash2, 64 );
memcpy( &ctx.cube, &x11evo_4way_ctx.cube, sizeof(cubehashParam) );
cubehashUpdateDigest( &ctx.cube, (byte*)hash3,
(const byte*) hash3, 64 );
if ( i < len-1 )
mm256_interleave_4x64( vhash,
hash0, hash1, hash2, hash3, 64<<3 );
break;
case 8:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
sph_shavite512( &ctx.shavite, hash0, 64 );
sph_shavite512_close( &ctx.shavite, hash0 );
memcpy( &ctx.shavite, &x11evo_4way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash1, 64 );
sph_shavite512_close( &ctx.shavite, hash1 );
memcpy( &ctx.shavite, &x11evo_4way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash2, 64 );
sph_shavite512_close( &ctx.shavite, hash2 );
memcpy( &ctx.shavite, &x11evo_4way_ctx.shavite,
sizeof(sph_shavite512_context) );
sph_shavite512( &ctx.shavite, hash3, 64 );
sph_shavite512_close( &ctx.shavite, hash3 );
if ( i < len-1 )
mm256_interleave_4x64( vhash,
hash0, hash1, hash2, hash3, 64<<3 );
break;
case 9:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
update_final_sd( &ctx.simd, (BitSequence *)hash0,
(const BitSequence *)hash0, 512 );
memcpy( &ctx.simd, &x11evo_4way_ctx.simd, sizeof(hashState_sd) );
update_final_sd( &ctx.simd, (BitSequence *)hash1,
(const BitSequence *)hash1, 512 );
memcpy( &ctx.simd, &x11evo_4way_ctx.simd, sizeof(hashState_sd) );
update_final_sd( &ctx.simd, (BitSequence *)hash2,
(const BitSequence *)hash2, 512 );
memcpy( &ctx.simd, &x11evo_4way_ctx.simd, sizeof(hashState_sd) );
update_final_sd( &ctx.simd, (BitSequence *)hash3,
(const BitSequence *)hash3, 512 );
if ( i < len-1 )
mm256_interleave_4x64( vhash,
hash0, hash1, hash2, hash3, 64<<3 );
break;
case 10:
mm256_deinterleave_4x64( hash0, hash1, hash2, hash3,
vhash, 64<<3 );
update_final_echo( &ctx.echo, (BitSequence *)hash0,
(const BitSequence *) hash0, 512 );
memcpy( &ctx.echo, &x11evo_4way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash1,
(const BitSequence *) hash1, 512 );
memcpy( &ctx.echo, &x11evo_4way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash2,
(const BitSequence *) hash2, 512 );
memcpy( &ctx.echo, &x11evo_4way_ctx.echo, sizeof(hashState_echo) );
update_final_echo( &ctx.echo, (BitSequence *)hash3,
(const BitSequence *) hash3, 512 );
if ( i < len-1 )
mm256_interleave_4x64( vhash,
hash0, hash1, hash2, hash3, 64<<3 );
break;
}
}
memcpy( state, hash0, 32 );
memcpy( state+32, hash1, 32 );
memcpy( state+64, hash2, 32 );
memcpy( state+96, hash3, 32 );
}
//static const uint32_t diff1targ = 0x0000ffff;
int scanhash_x11evo_4way( int thr_id, struct work* work, uint32_t max_nonce,
uint64_t *hashes_done )
{
uint32_t hash[4*8] __attribute__ ((aligned (64)));
uint32_t vdata[24*4] __attribute__ ((aligned (64)));
uint32_t endiandata[20] __attribute__((aligned(64)));
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
uint32_t n = pdata[19];
const uint32_t first_nonce = pdata[19];
uint32_t *nonces = work->nonces;
bool *found = work->nfound;
int num_found = 0;
uint32_t *noncep0 = vdata + 73; // 9*8 + 1
uint32_t *noncep1 = vdata + 75;
uint32_t *noncep2 = vdata + 77;
uint32_t *noncep3 = vdata + 79;
const uint32_t Htarg = ptarget[7];
swab32_array( endiandata, pdata, 20 );
int ntime = endiandata[17];
if ( ntime != s_ntime || s_seq == -1 )
{
evo_twisted_code( ntime, hashOrder );
s_ntime = ntime;
}
uint32_t hmask = 0xFFFFFFFF;
if ( Htarg > 0 )
{
if ( Htarg <= 0xF )
hmask = 0xFFFFFFF0;
else if ( Htarg <= 0xFF )
hmask = 0xFFFFFF00;
else if ( Htarg <= 0xFFF )
hmask = 0xFFFF000;
else if ( Htarg <= 0xFFFF )
hmask = 0xFFFF000;
}
uint64_t *edata = (uint64_t*)endiandata;
mm256_interleave_4x64( (uint64_t*)vdata, edata, edata, edata, edata, 640 );
do
{
found[0] = found[1] = found[2] = found[3] = false;
be32enc( noncep0, n );
be32enc( noncep1, n+1 );
be32enc( noncep2, n+2 );
be32enc( noncep3, n+3 );
x11evo_4way_hash( hash, vdata );
pdata[19] = n;
if ( ( hash[7] & hmask ) == 0 && fulltest( hash, ptarget ) )
{
found[0] = true;
num_found++;
nonces[0] = n;
work_set_target_ratio( work, hash );
}
if ( ( (hash+8)[7] & hmask ) == 0 && fulltest( hash+8, ptarget ) )
{
found[1] = true;
num_found++;
nonces[1] = n+1;
work_set_target_ratio( work, hash+8 );
}
if ( ( (hash+16)[7] & hmask ) == 0 && fulltest( hash+16, ptarget ) )
{
found[2] = true;
num_found++;
nonces[2] = n+2;
work_set_target_ratio( work, hash+16 );
}
if ( ( (hash+24)[7] & hmask ) == 0 && fulltest( hash+24, ptarget ) )
{
found[3] = true;
num_found++;
nonces[3] = n+3;
work_set_target_ratio( work, hash+24 );
}
n += 4;
} while ( ( num_found == 0 ) && ( n < max_nonce )
&& !work_restart[thr_id].restart );
*hashes_done = n - first_nonce + 1;
return num_found;
}
#endif

95
algo/x11/x11evo-gate.c Normal file
View File

@@ -0,0 +1,95 @@
#include "x11evo-gate.h"
int s_seq = -1;
static inline int getCurrentAlgoSeq( uint32_t current_time )
{
// change once per day
return (int) (current_time - X11EVO_INITIAL_DATE) / (60 * 60 * 24);
}
// swap_vars doesn't work here
void evo_swap( uint8_t *a, uint8_t *b )
{
uint8_t __tmp = *a;
*a = *b;
*b = __tmp;
}
void initPerm( uint8_t n[], uint8_t count )
{
int i;
for ( i = 0; i<count; i++ )
n[i] = i;
}
int nextPerm( uint8_t n[], uint32_t count )
{
uint32_t tail = 0, i = 0, j = 0;
if (unlikely( count <= 1 ))
return 0;
for ( i = count - 1; i>0 && n[i - 1] >= n[i]; i-- );
tail = i;
if ( tail > 0 )
for ( j = count - 1; j>tail && n[j] <= n[tail - 1]; j-- );
evo_swap( &n[tail - 1], &n[j] );
for ( i = tail, j = count - 1; i<j; i++, j-- )
evo_swap( &n[i], &n[j] );
return ( tail != 0 );
}
void getAlgoString( char *str, uint32_t count )
{
uint8_t algoList[X11EVO_FUNC_COUNT];
char *sptr;
int j;
int k;
initPerm( algoList, X11EVO_FUNC_COUNT );
for ( k = 0; k < count; k++ )
nextPerm( algoList, X11EVO_FUNC_COUNT );
sptr = str;
for ( j = 0; j < X11EVO_FUNC_COUNT; j++ )
{
if ( algoList[j] >= 10 )
sprintf( sptr, "%c", 'A' + (algoList[j] - 10) );
else
sprintf( sptr, "%u", algoList[j] );
sptr++;
}
*sptr = 0;
//applog(LOG_DEBUG, "nextPerm %s", str);
}
void evo_twisted_code( uint32_t ntime, char *permstr )
{
int seq = getCurrentAlgoSeq( ntime );
if ( s_seq != seq )
{
getAlgoString( permstr, seq );
s_seq = seq;
}
}
bool register_x11evo_algo( algo_gate_t* gate )
{
#if defined (X11EVO_4WAY)
init_x11evo_4way_ctx();
gate->scanhash = (void*)&scanhash_x11evo_4way;
gate->hash = (void*)&x11evo_4way_hash;
#else
init_x11evo_ctx();
gate->scanhash = (void*)&scanhash_x11evo;
gate->hash = (void*)&x11evo_hash;
#endif
gate->optimizations = SSE2_OPT | AES_OPT | AVX_OPT | AVX2_OPT | FOUR_WAY_OPT;
return true;
};

39
algo/x11/x11evo-gate.h Normal file
View File

@@ -0,0 +1,39 @@
#ifndef X11EVO_GATE_H__
#define X11EVO_GATE_H__ 1
#include "algo-gate-api.h"
#include <stdint.h>
#if defined(HASH_4WAY) && defined(__AES__)
#define X11EVO_4WAY
#endif
#define X11EVO_INITIAL_DATE 1462060800
#define X11EVO_FUNC_COUNT 11
extern int s_seq;
bool register_x11evo_algo( algo_gate_t* gate );
#if defined(X11EVO_4WAY)
void x11evo_4way_hash( void *state, const void *input );
int scanhash_x11evo_4way( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done );
void init_x11evo_4way_ctx();
#endif
void x11evo_hash( void *state, const void *input );
int scanhash_x11evo( int thr_id, struct work *work, uint32_t max_nonce,
uint64_t *hashes_done );
void init_x11evo_ctx();
void evo_twisted_code( uint32_t ntime, char *permstr );
#endif

112
algo/x11/x11evo.c
View File

@@ -1,5 +1,5 @@
#include "cpuminer-config.h"
#include "algo-gate-api.h"
#include "x11evo-gate.h"
#include <string.h>
#include <stdint.h>
@@ -26,9 +26,6 @@
#include "algo/cubehash/sse2/cubehash_sse2.h"
#include "algo/simd/sse2/nist.h"
#define INITIAL_DATE 1462060800
#define HASH_FUNC_COUNT 11
typedef struct {
#ifdef NO_AES_NI
sph_groestl512_context groestl;
@@ -70,94 +67,10 @@ void init_x11evo_ctx()
sph_shavite512_init( &x11evo_ctx.shavite );
}
/*
uint32_t getCurrentAlgoSeq(uint32_t current_time, uint32_t base_time)
{
return (current_time - base_time) / (60 * 60 * 24);
}
*/
static inline int getCurrentAlgoSeq( uint32_t current_time )
{
// change once per day
return (int) (current_time - INITIAL_DATE) / (60 * 60 * 24);
}
// swap_vars doesn't work here
void evo_swap( uint8_t *a, uint8_t *b )
{
uint8_t __tmp = *a;
*a = *b;
*b = __tmp;
}
void initPerm( uint8_t n[], uint8_t count )
{
int i;
for ( i = 0; i<count; i++ )
n[i] = i;
}
int nextPerm( uint8_t n[], uint32_t count )
{
uint32_t tail = 0, i = 0, j = 0;
if (unlikely( count <= 1 ))
return 0;
for ( i = count - 1; i>0 && n[i - 1] >= n[i]; i-- );
tail = i;
if ( tail > 0 )
for ( j = count - 1; j>tail && n[j] <= n[tail - 1]; j-- );
evo_swap( &n[tail - 1], &n[j] );
for ( i = tail, j = count - 1; i<j; i++, j-- )
evo_swap( &n[i], &n[j] );
return ( tail != 0 );
}
void getAlgoString( char *str, uint32_t count )
{
uint8_t algoList[HASH_FUNC_COUNT];
char *sptr;
int j;
int k;
initPerm( algoList, HASH_FUNC_COUNT );
for ( k = 0; k < count; k++ )
nextPerm( algoList, HASH_FUNC_COUNT );
sptr = str;
for ( j = 0; j < HASH_FUNC_COUNT; j++ )
{
if ( algoList[j] >= 10 )
sprintf( sptr, "%c", 'A' + (algoList[j] - 10) );
else
sprintf( sptr, "%u", algoList[j] );
sptr++;
}
*sptr = 0;
//applog(LOG_DEBUG, "nextPerm %s", str);
}
static char hashOrder[HASH_FUNC_COUNT + 1] = { 0 };
static char hashOrder[X11EVO_FUNC_COUNT + 1] = { 0 };
static __thread uint32_t s_ntime = UINT32_MAX;
static int s_seq = -1;
static void evo_twisted_code(uint32_t ntime, char *permstr)
{
int seq = getCurrentAlgoSeq(ntime);
if (s_seq != seq)
{
getAlgoString(permstr, seq);
s_seq = seq;
}
}
static inline void x11evo_hash( void *state, const void *input )
void x11evo_hash( void *state, const void *input )
{
uint32_t hash[16] __attribute__ ((aligned (64)));
x11evo_ctx_holder ctx __attribute__ ((aligned (64)));
@@ -242,10 +155,10 @@ static inline void x11evo_hash( void *state, const void *input )
memcpy( state, hash, 32 );
}
static const uint32_t diff1targ = 0x0000ffff;
//static const uint32_t diff1targ = 0x0000ffff;
int scanhash_x11evo( int thr_id, struct work* work, uint32_t max_nonce,
unsigned long *hashes_done )
uint64_t *hashes_done )
{
uint32_t endiandata[20] __attribute__((aligned(64)));
uint32_t hash64[8] __attribute__((aligned(64)));
@@ -274,19 +187,20 @@ int scanhash_x11evo( int thr_id, struct work* work, uint32_t max_nonce,
else if ( Htarg <= 0xFFF )
hmask = 0xFFFF000;
else if ( Htarg <= 0xFFFF )
hmask = 0xFFFF000;
hmask = 0xFFFF000;
}
do
{
pdata[19] = ++n;
be32enc( &endiandata[19], n );
x11evo_hash( hash64, &endiandata );
x11evo_hash( hash64, endiandata );
if ( ( hash64[7] & hmask ) == 0 )
{
if ( fulltest( hash64, ptarget ) )
{
*hashes_done = n - first_nonce + 1;
work_set_target_ratio( work, hash64 );
return true;
}
}
@@ -296,13 +210,3 @@ int scanhash_x11evo( int thr_id, struct work* work, uint32_t max_nonce,
pdata[19] = n;
return 0;
}
bool register_x11evo_algo( algo_gate_t* gate )
{
gate->optimizations = SSE2_OPT | AES_OPT | AVX_OPT | AVX2_OPT;
gate->scanhash = (void*)&scanhash_x11evo;
gate->hash = (void*)&x11evo_hash;
init_x11evo_ctx();
return true;
};

1
algo/x11/x11gost.c
View File

@@ -161,6 +161,7 @@ int scanhash_x11gost( int thr_id, struct work *work, uint32_t max_nonce,
if (hash[7] <= Htarg && fulltest(hash, ptarget)) {
pdata[19] = nonce;
*hashes_done = pdata[19] - first_nonce;
work_set_target_ratio( work, hash );
return 1;
}
nonce++;