You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
138 lines
4.0 KiB
138 lines
4.0 KiB
/** |
|
* S3 Hash (Also called 3S - Used by 1Coin) |
|
*/ |
|
|
|
extern "C" { |
|
#include "sph/sph_skein.h" |
|
#include "sph/sph_shavite.h" |
|
#include "sph/sph_simd.h" |
|
} |
|
|
|
#include "miner.h" |
|
#include "cuda_helper.h" |
|
|
|
#include <stdint.h> |
|
|
|
static uint32_t *d_hash[MAX_GPUS]; |
|
|
|
extern void x11_shavite512_cpu_init(int thr_id, uint32_t threads); |
|
extern void x11_shavite512_cpu_hash_80(int thr_id, uint32_t threads, uint32_t startNounce, uint32_t *d_hash, int order); |
|
extern void x11_shavite512_setBlock_80(void *pdata); |
|
|
|
extern int x11_simd512_cpu_init(int thr_id, uint32_t threads); |
|
extern void x11_simd512_cpu_hash_64(int thr_id, uint32_t threads, uint32_t startNounce, uint32_t *d_nonceVector, uint32_t *d_hash, int order); |
|
|
|
extern void quark_skein512_cpu_init(int thr_id, uint32_t threads); |
|
extern void quark_skein512_cpu_hash_64(int thr_id, uint32_t threads, uint32_t startNounce, uint32_t *d_nonceVector, uint32_t *d_hash, int order); |
|
|
|
/* CPU HASH */ |
|
extern "C" void s3hash(void *output, const void *input) |
|
{ |
|
sph_shavite512_context ctx_shavite; |
|
sph_simd512_context ctx_simd; |
|
sph_skein512_context ctx_skein; |
|
|
|
unsigned char hash[64]; |
|
|
|
sph_shavite512_init(&ctx_shavite); |
|
sph_shavite512(&ctx_shavite, input, 80); |
|
sph_shavite512_close(&ctx_shavite, (void*) hash); |
|
|
|
sph_simd512_init(&ctx_simd); |
|
sph_simd512(&ctx_simd, (const void*) hash, 64); |
|
sph_simd512_close(&ctx_simd, (void*) hash); |
|
|
|
sph_skein512_init(&ctx_skein); |
|
sph_skein512(&ctx_skein, (const void*) hash, 64); |
|
sph_skein512_close(&ctx_skein, (void*) hash); |
|
|
|
memcpy(output, hash, 32); |
|
} |
|
|
|
static bool init[MAX_GPUS] = { 0 }; |
|
|
|
/* Main S3 entry point */ |
|
extern "C" int scanhash_s3(int thr_id, struct work* work, uint32_t max_nonce, unsigned long *hashes_done) |
|
{ |
|
uint32_t *pdata = work->data; |
|
uint32_t *ptarget = work->target; |
|
const uint32_t first_nonce = pdata[19]; |
|
int intensity = 20; // 256*256*8*2; |
|
#ifdef WIN32 |
|
// reduce by one the intensity on windows |
|
intensity--; |
|
#endif |
|
uint32_t throughput = device_intensity(thr_id, __func__, 1 << intensity); |
|
throughput = min(throughput, max_nonce - first_nonce); |
|
|
|
if (opt_benchmark) |
|
((uint32_t*)ptarget)[7] = 0xF; |
|
|
|
if (!init[thr_id]) |
|
{ |
|
cudaSetDevice(device_map[thr_id]); |
|
|
|
x11_shavite512_cpu_init(thr_id, throughput); |
|
x11_simd512_cpu_init(thr_id, throughput); |
|
quark_skein512_cpu_init(thr_id, throughput); |
|
|
|
CUDA_CALL_OR_RET_X(cudaMalloc(&d_hash[thr_id], 16 * sizeof(uint32_t) * throughput), 0); |
|
|
|
cuda_check_cpu_init(thr_id, throughput); |
|
|
|
init[thr_id] = true; |
|
} |
|
|
|
uint32_t endiandata[20]; |
|
for (int k=0; k < 20; k++) |
|
be32enc(&endiandata[k], pdata[k]); |
|
|
|
x11_shavite512_setBlock_80((void*)endiandata); |
|
cuda_check_cpu_setTarget(ptarget); |
|
|
|
do { |
|
const uint32_t Htarg = ptarget[7]; |
|
uint32_t foundNonce; |
|
int order = 0; |
|
|
|
x11_shavite512_cpu_hash_80(thr_id, throughput, pdata[19], d_hash[thr_id], order++); |
|
x11_simd512_cpu_hash_64(thr_id, throughput, pdata[19], NULL, d_hash[thr_id], order++); |
|
quark_skein512_cpu_hash_64(thr_id, throughput, pdata[19], NULL, d_hash[thr_id], order++); |
|
|
|
*hashes_done = pdata[19] - first_nonce + throughput; |
|
|
|
foundNonce = cuda_check_hash(thr_id, throughput, pdata[19], d_hash[thr_id]); |
|
|
|
if (foundNonce != UINT32_MAX) |
|
{ |
|
uint32_t vhash64[8]; |
|
be32enc(&endiandata[19], foundNonce); |
|
s3hash(vhash64, endiandata); |
|
|
|
if (vhash64[7] <= Htarg && fulltest(vhash64, ptarget)) { |
|
int res = 1; |
|
uint32_t secNonce = cuda_check_hash_suppl(thr_id, throughput, pdata[19], d_hash[thr_id], 1); |
|
bn_store_hash_target_ratio(vhash64, ptarget, work); |
|
if (secNonce != 0) { |
|
be32enc(&endiandata[19], secNonce); |
|
s3hash(vhash64, endiandata); |
|
if (bn_hash_target_ratio(vhash64, ptarget) > work->shareratio) |
|
bn_store_hash_target_ratio(vhash64, ptarget, work); |
|
pdata[21] = secNonce; |
|
res++; |
|
} |
|
pdata[19] = foundNonce; |
|
return res; |
|
|
|
} else { |
|
applog(LOG_WARNING, "GPU #%d: result for nonce $%08X does not validate on CPU!", device_map[thr_id], foundNonce); |
|
} |
|
} |
|
|
|
pdata[19] += throughput; |
|
|
|
} while (pdata[19] < max_nonce && !work_restart[thr_id].restart); |
|
|
|
*hashes_done = pdata[19] - first_nonce + 1; |
|
return 0; |
|
}
|
|
|