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.
110 lines
2.8 KiB
110 lines
2.8 KiB
#include <cuda_runtime.h> |
|
#include "miner.h" |
|
#include "neoscrypt/neoscrypt.h" |
|
|
|
extern void neoscrypt_setBlockTarget(uint32_t * data, const void *ptarget); |
|
|
|
extern void neoscrypt_init_2stream(int thr_id, uint32_t threads); |
|
extern void neoscrypt_free_2stream(int thr_id); |
|
extern void neoscrypt_hash_k4_2stream(int thr_id, uint32_t threads, uint32_t startNounce, uint32_t *result, bool stratum); |
|
|
|
static bool init[MAX_GPUS] = { 0 }; |
|
|
|
int scanhash_neoscrypt(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]; |
|
|
|
int dev_id = device_map[thr_id]; |
|
int intensity = is_windows() ? 18 : 19; |
|
uint32_t throughput = cuda_default_throughput(thr_id, 1U << intensity); |
|
throughput = throughput / 32; /* set for max intensity ~= 20 */ |
|
api_set_throughput(thr_id, throughput); |
|
|
|
if (init[thr_id]) throughput = min(throughput, max_nonce - first_nonce + 1); |
|
|
|
if (opt_benchmark) |
|
ptarget[7] = 0x00ff; |
|
|
|
if (!init[thr_id]) |
|
{ |
|
cudaDeviceSynchronize(); |
|
cudaSetDevice(dev_id); |
|
cudaDeviceSetCacheConfig(cudaFuncCachePreferL1); |
|
cudaGetLastError(); // reset errors if device is not "reset" |
|
|
|
if (device_sm[dev_id] <= 300) { |
|
gpulog(LOG_ERR, thr_id, "Sorry neoscrypt is not supported on SM 3.0 devices"); |
|
proper_exit(EXIT_CODE_CUDA_ERROR); |
|
} |
|
|
|
gpulog(LOG_INFO, thr_id, "Using %d cuda threads", throughput); |
|
neoscrypt_init_2stream(thr_id, throughput); |
|
|
|
init[thr_id] = true; |
|
} |
|
|
|
if (have_stratum) { |
|
for (int k = 0; k < 20; k++) |
|
be32enc(&endiandata[k], pdata[k]); |
|
} else { |
|
for (int k = 0; k < 20; k++) |
|
endiandata[k] = pdata[k]; |
|
} |
|
|
|
neoscrypt_setBlockTarget(endiandata,ptarget); |
|
|
|
do { |
|
uint32_t foundNonces[2] = { UINT32_MAX, UINT32_MAX }; |
|
neoscrypt_hash_k4_2stream(thr_id, throughput, pdata[19], foundNonces, have_stratum); |
|
|
|
*hashes_done = pdata[19] - first_nonce + throughput; |
|
|
|
if (foundNonces[0] != UINT32_MAX) |
|
{ |
|
uint32_t _ALIGN(64) vhash[8]; |
|
|
|
if (have_stratum) { |
|
be32enc(&endiandata[19], foundNonces[0]); |
|
} else { |
|
endiandata[19] = foundNonces[0]; |
|
} |
|
neoscrypt((uchar*)vhash, (uchar*) endiandata, 0x80000620U); |
|
|
|
if (vhash[7] <= ptarget[7] && fulltest(vhash, ptarget)) { |
|
work_set_target_ratio(work, vhash); |
|
pdata[19] = foundNonces[0]; |
|
return 1; |
|
} else { |
|
gpulog(LOG_WARNING, thr_id, "nonce %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 = pdata[19] - first_nonce; |
|
return 0; |
|
} |
|
|
|
// cleanup |
|
void free_neoscrypt(int thr_id) |
|
{ |
|
if (!init[thr_id]) |
|
return; |
|
|
|
cudaThreadSynchronize(); |
|
|
|
neoscrypt_free_2stream(thr_id); |
|
init[thr_id] = false; |
|
|
|
cudaDeviceSynchronize(); |
|
}
|
|
|