GOSTCoin CUDA miner project, compatible with most nvidia cards, containing only gostd algo
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.

124 lines
3.6 KiB

/*
* whirlpool routine (djm)
*/
extern "C"
{
#include "sph/sph_whirlpool.h"
#include "miner.h"
}
#include "cuda_helper.h"
static uint32_t *d_hash[8];
extern void x15_whirlpool_cpu_init(int thr_id, int threads, int mode);
extern void x15_whirlpool_cpu_hash_64(int thr_id, int threads, uint32_t startNounce, uint32_t *d_nonceVector, uint32_t *d_hash, int order);
extern void whirlpool512_setBlock_80(void *pdata, const void *ptarget);
extern void whirlpool512_cpu_hash_80(int thr_id, int threads, uint32_t startNounce, uint32_t *d_hash, int order);
extern uint32_t whirlpool512_cpu_finalhash_64(int thr_id, int threads, uint32_t startNounce, uint32_t *d_nonceVector, uint32_t *d_hash, int order);
// CPU Hash function
extern "C" void wcoinhash(void *state, const void *input)
{
sph_whirlpool_context ctx_whirlpool;
unsigned char hash[128]; // uint32_t hashA[16], hashB[16];
#define hashB hash+64
memset(hash, 0, sizeof hash);
// shavite 1
sph_whirlpool1_init(&ctx_whirlpool);
sph_whirlpool1(&ctx_whirlpool, input, 80);
sph_whirlpool1_close(&ctx_whirlpool, hash);
sph_whirlpool1_init(&ctx_whirlpool);
sph_whirlpool1(&ctx_whirlpool, hash, 64);
sph_whirlpool1_close(&ctx_whirlpool, hashB);
sph_whirlpool1_init(&ctx_whirlpool);
sph_whirlpool1(&ctx_whirlpool, hashB, 64);
sph_whirlpool1_close(&ctx_whirlpool, hash);
sph_whirlpool1_init(&ctx_whirlpool);
sph_whirlpool1(&ctx_whirlpool, hash, 64);
sph_whirlpool1_close(&ctx_whirlpool, hash);
memcpy(state, hash, 32);
}
static bool init[8] = { 0 };
extern "C" int scanhash_whc(int thr_id, uint32_t *pdata,
const uint32_t *ptarget, uint32_t max_nonce,
unsigned long *hashes_done)
{
const uint32_t first_nonce = pdata[19];
uint32_t endiandata[20];
int throughput = opt_work_size ? opt_work_size : (1 << 19); // 256*256*8;
throughput = min(throughput, (int)(max_nonce - first_nonce));
if (opt_benchmark)
((uint32_t*)ptarget)[7] = 0x0000ff;
if (!init[thr_id]) {
cudaSetDevice(device_map[thr_id]);
// Konstanten kopieren, Speicher belegen
cudaMalloc(&d_hash[thr_id], 16 * sizeof(uint32_t) * throughput);
x15_whirlpool_cpu_init(thr_id, throughput, 1 /* old whirlpool */);
init[thr_id] = true;
}
for (int k=0; k < 20; k++) {
be32enc(&endiandata[k], ((uint32_t*)pdata)[k]);
}
whirlpool512_setBlock_80((void*)endiandata, ptarget);
do {
uint32_t foundNonce;
int order = 0;
whirlpool512_cpu_hash_80(thr_id, throughput, pdata[19], d_hash[thr_id], order++);
x15_whirlpool_cpu_hash_64(thr_id, throughput, pdata[19], NULL, d_hash[thr_id], order++);
x15_whirlpool_cpu_hash_64(thr_id, throughput, pdata[19], NULL, d_hash[thr_id], order++);
foundNonce = whirlpool512_cpu_finalhash_64(thr_id, throughput, pdata[19], NULL, d_hash[thr_id], order++);
if (foundNonce != UINT32_MAX)
{
const uint32_t Htarg = ptarget[7];
uint32_t vhash64[8];
be32enc(&endiandata[19], foundNonce);
wcoinhash(vhash64, endiandata);
if (vhash64[7] <= Htarg && fulltest(vhash64, ptarget)) {
int res = 1;
*hashes_done = pdata[19] - first_nonce + throughput;
10 years ago
#if 0
uint32_t secNonce = cuda_check_hash_suppl(thr_id, throughput, pdata[19], d_hash[thr_id], 1);
if (secNonce != 0) {
pdata[21] = secNonce;
res++;
}
10 years ago
#endif
pdata[19] = foundNonce;
return res;
}
else if (vhash64[7] > Htarg) {
applog(LOG_INFO, "GPU #%d: result for %08x is not in range: %x > %x", thr_id, foundNonce, vhash64[7], Htarg);
}
else {
applog(LOG_INFO, "GPU #%d: result for %08x does not validate on CPU!", thr_id, foundNonce);
}
}
pdata[19] += throughput;
} while (pdata[19] < max_nonce && !work_restart[thr_id].restart);
10 years ago
*hashes_done = pdata[19] - first_nonce;
return 0;
}