GOSTcoin support for ccminer CUDA miner project, compatible with most nvidia cards
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.

107 lines
2.5 KiB

#include <string.h>
#include <stdint.h>
#include <cuda_runtime.h>
#include <openssl/sha.h>
#include "sph/sph_groestl.h"
#include "cuda_groestlcoin.h"
#include "miner.h"
// CPU hash
void groestlhash(void *state, const void *input)
{
uint32_t _ALIGN(64) hash[16];
sph_groestl512_context ctx_groestl;
sph_groestl512_init(&ctx_groestl);
sph_groestl512(&ctx_groestl, input, 80);
sph_groestl512_close(&ctx_groestl, hash);
sph_groestl512_init(&ctx_groestl);
sph_groestl512(&ctx_groestl, hash, 64);
sph_groestl512_close(&ctx_groestl, hash);
memcpy(state, hash, 32);
}
static bool init[MAX_GPUS] = { 0 };
int scanhash_groestlcoin(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];
uint32_t throughput = cuda_default_throughput(thr_id, 1 << 19); // 256*256*8
if (init[thr_id]) throughput = min(throughput, max_nonce - start_nonce);
uint32_t *outputHash = (uint32_t*)malloc(throughput * 64);
if (opt_benchmark)
((uint32_t*)ptarget)[7] = 0x000000ff;
if (!init[thr_id])
{
cudaSetDevice(device_map[thr_id]);
groestlcoin_cpu_init(thr_id, throughput);
init[thr_id] = true;
}
for (int k=0; k < 20; k++)
be32enc(&endiandata[k], pdata[k]);
groestlcoin_cpu_setBlock(thr_id, endiandata, (void*)ptarget);
do {
uint32_t foundNounce = UINT32_MAX;
*hashes_done = pdata[19] - start_nonce + throughput;
// GPU hash
groestlcoin_cpu_hash(thr_id, throughput, pdata[19], outputHash, &foundNounce);
if (foundNounce < UINT32_MAX)
{
uint32_t _ALIGN(64) vhash[8];
endiandata[19] = swab32(foundNounce);
groestlhash(vhash, endiandata);
if (vhash[7] <= ptarget[7] && fulltest(vhash, ptarget)) {
work_set_target_ratio(work, vhash);
pdata[19] = foundNounce;
free(outputHash);
return true;
} else {
applog(LOG_WARNING, "GPU #%d: result for nonce %08x does not validate on CPU!",
device_map[thr_id], foundNounce);
}
}
if ((uint64_t) pdata[19] + throughput > max_nonce) {
*hashes_done = pdata[19] - start_nonce + 1;
pdata[19] = max_nonce;
break;
}
pdata[19] += throughput;
} while (pdata[19] < max_nonce && !work_restart[thr_id].restart);
free(outputHash);
return 0;
}
// cleanup
void free_groestlcoin(int thr_id)
{
if (!init[thr_id])
return;
cudaSetDevice(device_map[thr_id]);
groestlcoin_cpu_free(thr_id);
init[thr_id] = false;
cudaDeviceSynchronize();
}