From 6169bf683b8aa30aa509c0df480a6f550448c5f3 Mon Sep 17 00:00:00 2001 From: Tanguy Pruvot Date: Sun, 26 Oct 2014 08:59:49 +0100 Subject: [PATCH] Add S3 Algo (1Coin) Simple addition of the algo using existing X11 code --- Makefile.am | 3 +- ccminer.vcxproj | 1 + ccminer.vcxproj.filters | 3 + cpu-miner.c | 8 +++ miner.h | 5 ++ util.c | 4 ++ x11/s3.cu | 123 ++++++++++++++++++++++++++++++++++++++++ 7 files changed, 146 insertions(+), 1 deletion(-) create mode 100644 x11/s3.cu diff --git a/Makefile.am b/Makefile.am index 1d39ec7..85d2439 100644 --- a/Makefile.am +++ b/Makefile.am @@ -46,7 +46,8 @@ ccminer_SOURCES = elist.h miner.h compat.h \ x13/x13.cu x13/cuda_x13_hamsi512.cu x13/cuda_x13_fugue512.cu \ x15/x14.cu x15/x15.cu x15/cuda_x14_shabal512.cu x15/cuda_x15_whirlpool.cu \ x15/whirlpool.cu \ - x17/x17.cu x17/cuda_x17_haval512.cu x17/cuda_x17_sha512.cu + x17/x17.cu x17/cuda_x17_haval512.cu x17/cuda_x17_sha512.cu \ + x11/s3.cu ccminer_LDFLAGS = $(PTHREAD_FLAGS) @CUDA_LDFLAGS@ diff --git a/ccminer.vcxproj b/ccminer.vcxproj index e683b9c..11b2aea 100644 --- a/ccminer.vcxproj +++ b/ccminer.vcxproj @@ -509,6 +509,7 @@ %(AdditionalOptions) 64 + true diff --git a/ccminer.vcxproj.filters b/ccminer.vcxproj.filters index c77708d..334ca64 100644 --- a/ccminer.vcxproj.filters +++ b/ccminer.vcxproj.filters @@ -469,5 +469,8 @@ Source Files\CUDA\keccak + + Source Files\CUDA\x11 + \ No newline at end of file diff --git a/cpu-miner.c b/cpu-miner.c index 1082ec6..6b9ed7b 100644 --- a/cpu-miner.c +++ b/cpu-miner.c @@ -148,6 +148,7 @@ typedef enum { ALGO_PENTABLAKE, ALGO_QUARK, ALGO_QUBIT, + ALGO_S3, ALGO_WHC, ALGO_X11, ALGO_X13, @@ -176,6 +177,7 @@ static const char *algo_names[] = { "penta", "quark", "qubit", + "s3", "whirl", "x11", "x13", @@ -265,6 +267,7 @@ Options:\n\ penta Pentablake hash (5x Blake 512)\n\ quark Quark\n\ qubit Qubit\n\ + s3 S3 (1Coin)\n\ x11 X11 (DarkCoin)\n\ x13 X13 (MaruCoin)\n\ x14 X14\n\ @@ -1235,6 +1238,11 @@ continue_scan: max_nonce, &hashes_done); break; + case ALGO_S3: + rc = scanhash_s3(thr_id, work.data, work.target, + max_nonce, &hashes_done); + break; + case ALGO_WHC: rc = scanhash_whc(thr_id, work.data, work.target, max_nonce, &hashes_done); diff --git a/miner.h b/miner.h index 8a75322..12a4b24 100644 --- a/miner.h +++ b/miner.h @@ -295,6 +295,10 @@ extern int scanhash_scrypt(int thr_id, uint32_t *pdata, unsigned char *scratchbuf, const uint32_t *ptarget, uint32_t max_nonce, unsigned long *hashes_done); +extern int scanhash_s3(int thr_id, uint32_t *pdata, + const uint32_t *ptarget, uint32_t max_nonce, + unsigned long *hashes_done); + extern int scanhash_whc(int thr_id, uint32_t *pdata, const uint32_t *ptarget, uint32_t max_nonce, unsigned long *hashes_done); @@ -483,6 +487,7 @@ void nist5hash(void *state, const void *input); void pentablakehash(void *output, const void *input); void quarkhash(void *state, const void *input); void qubithash(void *state, const void *input); +void s3hash(void *output, const void *input); void wcoinhash(void *state, const void *input); void x11hash(void *output, const void *input); void x13hash(void *output, const void *input); diff --git a/util.c b/util.c index 639baa7..eb973cb 100644 --- a/util.c +++ b/util.c @@ -1534,6 +1534,10 @@ void print_hash_tests(void) qubithash(&hash[0], &buf[0]); printpfx("qubit", hash); + memset(hash, 0, sizeof hash); + s3hash(&hash[0], &buf[0]); + printpfx("S3", hash); + memset(hash, 0, sizeof hash); wcoinhash(&hash[0], &buf[0]); printpfx("whirl", hash); diff --git a/x11/s3.cu b/x11/s3.cu new file mode 100644 index 0000000..d834f08 --- /dev/null +++ b/x11/s3.cu @@ -0,0 +1,123 @@ +/** + * S3 Hash (Also called 3S - Used by 1Coin) + */ + +extern "C" { +#include "sph/sph_skein.h" +#include "sph/sph_shavite.h" +#include "sph/sph_simd.h" +} + +#include "miner.h" +#include "cuda_helper.h" + +#include + +extern int device_map[8]; + +static uint32_t *d_hash[8]; + +extern void x11_shavite512_cpu_init(int thr_id, int threads); +extern void x11_shavite512_cpu_hash_80(int thr_id, int threads, uint32_t startNounce, uint32_t *d_hash, int order); +extern void x11_shavite512_setBlock_80(void *pdata); + +extern void x11_simd512_cpu_init(int thr_id, int threads); +extern void x11_simd512_cpu_hash_64(int thr_id, int threads, uint32_t startNounce, uint32_t *d_nonceVector, uint32_t *d_hash, int order); + +extern void quark_skein512_cpu_init(int thr_id, int threads); +extern void quark_skein512_cpu_hash_64(int thr_id, int threads, uint32_t startNounce, uint32_t *d_nonceVector, uint32_t *d_hash, int order); + +/* CPU HASH */ +extern "C" void s3hash(void *output, const void *input) +{ + sph_shavite512_context ctx_shavite; + sph_simd512_context ctx_simd; + sph_skein512_context ctx_skein; + + unsigned char hash[64]; + + sph_shavite512_init(&ctx_shavite); + sph_shavite512(&ctx_shavite, input, 80); + sph_shavite512_close(&ctx_shavite, (void*) hash); + + sph_simd512_init(&ctx_simd); + sph_simd512(&ctx_simd, (const void*) hash, 64); + sph_simd512_close(&ctx_simd, (void*) hash); + + sph_skein512_init(&ctx_skein); + sph_skein512(&ctx_skein, (const void*) hash, 64); + sph_skein512_close(&ctx_skein, (void*) hash); + + memcpy(output, hash, 32); +} + +/* Main S3 entry point */ +extern "C" int scanhash_s3(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]; + const int throughput = 256*256*8*2; + + static bool init[8] = { 0, 0, 0, 0, 0, 0, 0, 0 }; + + if (opt_benchmark) + ((uint32_t*)ptarget)[7] = 0xF; + + if (!init[thr_id]) + { + CUDA_SAFE_CALL(cudaSetDevice(device_map[thr_id])); + + CUDA_SAFE_CALL(cudaMalloc(&d_hash[thr_id], 64 * throughput)); + + x11_shavite512_cpu_init(thr_id, throughput); + x11_simd512_cpu_init(thr_id, throughput); + quark_skein512_cpu_init(thr_id, throughput); + + cuda_check_cpu_init(thr_id, throughput); + + init[thr_id] = true; + } + + uint32_t endiandata[20]; + for (int k=0; k < 20; k++) + be32enc(&endiandata[k], ((uint32_t*)pdata)[k]); + + x11_shavite512_setBlock_80((void*)endiandata); + cuda_check_cpu_setTarget(ptarget); + + do { + const uint32_t Htarg = ptarget[7]; + uint32_t foundNonce; + int order = 0; + + x11_shavite512_cpu_hash_80(thr_id, throughput, pdata[19], d_hash[thr_id], order++); + x11_simd512_cpu_hash_64(thr_id, throughput, pdata[19], NULL, d_hash[thr_id], order++); + quark_skein512_cpu_hash_64(thr_id, throughput, pdata[19], NULL, d_hash[thr_id], order++); + + foundNonce = cuda_check_cpu_hash_64(thr_id, throughput, pdata[19], NULL, d_hash[thr_id], order++); + + if (foundNonce != 0xffffffff) + { + uint32_t vhash64[8]; + be32enc(&endiandata[19], foundNonce); + s3hash(vhash64, endiandata); + + if (vhash64[7] <= Htarg && fulltest(vhash64, ptarget)) { + + pdata[19] = foundNonce; + *hashes_done = foundNonce - first_nonce + 1; + return 1; + + } else { + applog(LOG_INFO, "GPU #%d: result for nonce $%08X does not validate on CPU!", thr_id, foundNonce); + } + } + + pdata[19] += throughput; + + } while (pdata[19] < max_nonce && !work_restart[thr_id].restart); + + *hashes_done = pdata[19] - first_nonce + 1; + return 0; +}