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.

125 lines
3.1 KiB

#include <string.h>
#include <stdint.h>
#include <cuda_runtime.h>
#include <openssl/sha.h>
#include "sph/sph_groestl.h"
#include "miner.h"
void myriadgroestl_cpu_init(int thr_id, uint32_t threads);
void myriadgroestl_cpu_free(int thr_id);
void myriadgroestl_cpu_setBlock(int thr_id, void *data, uint32_t *target);
void myriadgroestl_cpu_hash(int thr_id, uint32_t threads, uint32_t startNonce, uint32_t *resNonces);
void myriadhash(void *state, const void *input)
{
uint32_t _ALIGN(64) hash[16];
10 years ago
sph_groestl512_context ctx_groestl;
SHA256_CTX sha256;
10 years ago
sph_groestl512_init(&ctx_groestl);
sph_groestl512(&ctx_groestl, input, 80);
sph_groestl512_close(&ctx_groestl, hash);
10 years ago
SHA256_Init(&sha256);
SHA256_Update(&sha256,(unsigned char *)hash, 64);
SHA256_Final((unsigned char *)hash, &sha256);
memcpy(state, hash, 32);
}
static bool init[MAX_GPUS] = { 0 };
int scanhash_myriad(int thr_id, struct work *work, uint32_t max_nonce, unsigned long *hashes_done)
{
uint32_t _ALIGN(64) endiandata[32];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
uint32_t start_nonce = pdata[19];
int dev_id = device_map[thr_id];
int intensity = (device_sm[dev_id] >= 600) ? 20 : 18;
uint32_t throughput = cuda_default_throughput(thr_id, 1U << intensity);
if (init[thr_id]) throughput = min(throughput, max_nonce - start_nonce);
if (opt_benchmark)
ptarget[7] = 0x0000ff;
// init
if(!init[thr_id])
{
cudaSetDevice(dev_id);
if (opt_cudaschedule == -1 && gpu_threads == 1) {
cudaDeviceReset();
// reduce cpu usage
cudaSetDeviceFlags(cudaDeviceScheduleBlockingSync);
CUDA_LOG_ERROR();
}
myriadgroestl_cpu_init(thr_id, throughput);
init[thr_id] = true;
}
for (int k=0; k < 20; k++)
be32enc(&endiandata[k], pdata[k]);
myriadgroestl_cpu_setBlock(thr_id, endiandata, ptarget);
do {
// GPU
uint32_t foundNonces[2] = { UINT32_MAX, UINT32_MAX };
myriadgroestl_cpu_hash(thr_id, throughput, pdata[19], foundNonces);
*hashes_done = pdata[19] - start_nonce + throughput;
if (foundNonces[0] < UINT32_MAX && bench_algo < 0)
{
uint32_t _ALIGN(64) vhash[8];
endiandata[19] = swab32(foundNonces[0]);
myriadhash(vhash, endiandata);
if (vhash[7] <= ptarget[7] && fulltest(vhash, ptarget)) {
work_set_target_ratio(work, vhash);
pdata[19] = foundNonces[0];
// search for another nonce
if (foundNonces[1] != UINT32_MAX) {
endiandata[19] = swab32(foundNonces[1]);
myriadhash(vhash, endiandata);
pdata[21] = foundNonces[1];
if(bn_hash_target_ratio(vhash, ptarget) > work->shareratio) {
work_set_target_ratio(work, vhash);
}
return 2;
}
return 1;
} else if (vhash[7] > ptarget[7]) {
gpulog(LOG_WARNING, thr_id, "result for %08x does not validate on CPU!", foundNonces[0]);
}
}
if ((uint64_t) throughput + pdata[19] >= max_nonce) {
pdata[19] = max_nonce;
break;
}
pdata[19] += throughput;
} while (!work_restart[thr_id].restart);
*hashes_done = max_nonce - start_nonce;
return 0;
}
// cleanup
void free_myriad(int thr_id)
{
if (!init[thr_id])
return;
cudaThreadSynchronize();
myriadgroestl_cpu_free(thr_id);
init[thr_id] = false;
cudaDeviceSynchronize();
}