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.
 
 
 
 
 

122 lines
2.9 KiB

/*
* Keccak 256
*
*/
extern "C"
{
#include "sph/sph_shavite.h"
#include "sph/sph_simd.h"
#include "sph/sph_keccak.h"
#include "miner.h"
}
#include "cuda_helper.h"
static uint32_t *d_hash[MAX_GPUS];
extern void keccak256_cpu_init(int thr_id, uint32_t threads);
extern void keccak256_cpu_free(int thr_id);
extern void keccak256_setBlock_80(void *pdata,const void *ptarget);
extern uint32_t keccak256_cpu_hash_80(int thr_id, uint32_t threads, uint32_t startNounce, uint32_t *d_hash, int order);
// CPU Hash
extern "C" void keccak256_hash(void *state, const void *input)
{
uint32_t _ALIGN(64) hash[16];
sph_keccak_context ctx_keccak;
sph_keccak256_init(&ctx_keccak);
sph_keccak256 (&ctx_keccak, input, 80);
sph_keccak256_close(&ctx_keccak, (void*) hash);
memcpy(state, hash, 32);
}
static bool init[MAX_GPUS] = { 0 };
extern "C" int scanhash_keccak256(int thr_id, struct work* work, uint32_t max_nonce, unsigned long *hashes_done)
{
uint32_t _ALIGN(64) endiandata[20];
uint32_t *pdata = work->data;
uint32_t *ptarget = work->target;
const uint32_t first_nonce = pdata[19];
uint32_t throughput = cuda_default_throughput(thr_id, 1U << 21); // 256*256*8*4
if (init[thr_id]) throughput = min(throughput, max_nonce - first_nonce);
if (opt_benchmark)
ptarget[7] = 0x000f;
if (!init[thr_id])
{
cudaSetDevice(device_map[thr_id]);
if (opt_cudaschedule == -1 && gpu_threads == 1) {
cudaDeviceReset();
// reduce cpu usage
cudaSetDeviceFlags(cudaDeviceScheduleBlockingSync);
CUDA_LOG_ERROR();
}
gpulog(LOG_INFO, thr_id, "Intensity set to %g, %u cuda threads", throughput2intensity(throughput), throughput);
CUDA_SAFE_CALL(cudaMalloc(&d_hash[thr_id], throughput * 64));
keccak256_cpu_init(thr_id, throughput);
init[thr_id] = true;
}
for (int k=0; k < 20; k++) {
be32enc(&endiandata[k], pdata[k]);
}
keccak256_setBlock_80((void*)endiandata, ptarget);
do {
int order = 0;
*hashes_done = pdata[19] - first_nonce + throughput;
uint32_t foundNonce = keccak256_cpu_hash_80(thr_id, throughput, pdata[19], d_hash[thr_id], order++);
if (foundNonce != UINT32_MAX && bench_algo < 0)
{
uint32_t _ALIGN(64) vhash64[8];
be32enc(&endiandata[19], foundNonce);
keccak256_hash(vhash64, endiandata);
if (vhash64[7] <= ptarget[7] && fulltest(vhash64, ptarget)) {
work_set_target_ratio(work, vhash64);
pdata[19] = foundNonce;
return 1;
}
else {
gpulog(LOG_WARNING, thr_id, "result for %08x does not validate on CPU!", foundNonce);
}
}
if ((uint64_t) throughput + pdata[19] >= max_nonce) {
pdata[19] = max_nonce;
break;
}
pdata[19] += throughput;
} while (!work_restart[thr_id].restart);
*hashes_done = pdata[19] - first_nonce;
return 0;
}
// cleanup
extern "C" void free_keccak256(int thr_id)
{
if (!init[thr_id])
return;
cudaThreadSynchronize();
cudaFree(d_hash[thr_id]);
keccak256_cpu_free(thr_id);
cudaDeviceSynchronize();
init[thr_id] = false;
}