GOSTSec
/
ccminer-gostd-lite
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
Browse Source

scrypt: clean some defines 

and prevent error reports on Ctrl+C
master
Tanguy Pruvot 10 years ago
parent
d9b0312897
commit
9aace79718
9 changed files with 36 additions and 72 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 40
      scrypt/blake.cu
  2. 2
      scrypt/fermi_kernel.cu
  3. 11
      scrypt/keccak.cu
  4. 2
      scrypt/kepler_kernel.cu
  5. 2
      scrypt/nv_kernel2.cu
  6. 32
      scrypt/salsa_kernel.cu
  7. 13
      scrypt/sha256.cu
  8. 2
      scrypt/test_kernel.cu
  9. 2
      scrypt/titan_kernel.cu

40
scrypt/blake.cu
Unescape View File

@ -12,36 +12,29 @@
#include <map> #include <map>
#include <stdint.h> #include <stdint.h>
#include "cuda_runtime.h"
#include "salsa_kernel.h"
#include "miner.h" #include "miner.h"
#include "salsa_kernel.h"
#include "cuda_helper.h"
typedef uint32_t sph_u32; typedef uint32_t sph_u32;
#define SPH_C32(x) ((sph_u32)(x)) #define SPH_ROTL32 ROTL32
#define SPH_T32(x) ((x) & SPH_C32(0xFFFFFFFF)) #define SPH_ROTR32 ROTR32
#define SPH_ROTL32(x, n) SPH_T32(((x) << (n)) | ((x) >> (32 - (n))))
#define SPH_ROTR32(x, n) SPH_ROTL32(x, (32 - (n)))
__constant__ uint64_t ptarget64[4]; __constant__ uint64_t ptarget64[4];
__constant__ uint32_t pdata[20]; __constant__ uint32_t pdata[20];
// define some error checking macros // define some error checking macros
#undef checkCudaErrors
#if WIN32
#define DELIMITER '/'
#else
#define DELIMITER '/' #define DELIMITER '/'
#endif
#define __FILENAME__ ( strrchr(__FILE__, DELIMITER) != NULL ? strrchr(__FILE__, DELIMITER)+1 : __FILE__ ) #define __FILENAME__ ( strrchr(__FILE__, DELIMITER) != NULL ? strrchr(__FILE__, DELIMITER)+1 : __FILE__ )
#undef checkCudaErrors
#define checkCudaErrors(x) \ #define checkCudaErrors(x) \
{ \ { \
cudaGetLastError(); \ cudaGetLastError(); \
x; \ x; \
cudaError_t err = cudaGetLastError(); \ cudaError_t err = cudaGetLastError(); \
if (err != cudaSuccess) \ if (err != cudaSuccess && !abort_flag) \
applog(LOG_ERR, "GPU #%d: cudaError %d (%s) calling '%s' (%s line %d)\n", device_map[thr_id], err, cudaGetErrorString(err), #x, __FILENAME__, __LINE__); \ applog(LOG_ERR, "GPU #%d: cudaError %d (%s) (%s line %d)\n", device_map[thr_id], err, cudaGetErrorString(err), __FILENAME__, __LINE__); \
} }
// from salsa_kernel.cu // from salsa_kernel.cu
@ -54,12 +47,6 @@ extern std::map<int, uint32_t *> context_hash[2];
#pragma warning (disable: 4146) #pragma warning (disable: 4146)
#endif #endif
static __device__ sph_u32 cuda_sph_bswap32(sph_u32 x)
{
return (((x << 24) & 0xff000000u) | ((x << 8) & 0x00ff0000u)
| ((x >> 8) & 0x0000ff00u) | ((x >> 24) & 0x000000ffu));
}
/** /**
* Encode a 32-bit value into the provided buffer (big endian convention). * Encode a 32-bit value into the provided buffer (big endian convention).
* *
@ -69,7 +56,7 @@ static __device__ sph_u32 cuda_sph_bswap32(sph_u32 x)
static __device__ void static __device__ void
cuda_sph_enc32be(void *dst, sph_u32 val) cuda_sph_enc32be(void *dst, sph_u32 val)
{ {
*(sph_u32 *)dst = cuda_sph_bswap32(val); *(sph_u32 *)dst = cuda_swab32(val);
} }
#define Z00 0 #define Z00 0
@ -344,12 +331,13 @@ cuda_sph_enc32be(void *dst, sph_u32 val)
H7 ^= S3 ^ V7 ^ VF; \ H7 ^= S3 ^ V7 ^ VF; \
} while (0) } while (0)
__global__ void cuda_blake256_hash( uint64_t *g_out, uint32_t nonce, uint32_t *g_good, bool validate ) __global__
void cuda_blake256_hash( uint64_t *g_out, uint32_t nonce, uint32_t *g_good, bool validate )
{ {
uint32_t input[16]; uint32_t input[16];
uint64_t output[4]; uint64_t output[4];
#pragma unroll 16 #pragma unroll
for (int i=0; i < 16; ++i) input[i] = pdata[i]; for (int i=0; i < 16; ++i) input[i] = pdata[i];
sph_u32 H0 = 0x6A09E667; sph_u32 H0 = 0x6A09E667;
@ -369,11 +357,11 @@ __global__ void cuda_blake256_hash( uint64_t *g_out, uint32_t nonce, uint32_t *g
T0 = SPH_T32(T0 + 512); T0 = SPH_T32(T0 + 512);
COMPRESS32; COMPRESS32;
#pragma unroll 3 #pragma unroll
for (int i=0; i < 3; ++i) input[i] = pdata[16+i]; for (int i=0; i < 3; ++i) input[i] = pdata[16+i];
input[3] = nonce + ((blockIdx.x * blockDim.x) + threadIdx.x); input[3] = nonce + ((blockIdx.x * blockDim.x) + threadIdx.x);
input[4] = 0x80000000; input[4] = 0x80000000;
#pragma unroll 8 #pragma unroll 8
for (int i=5; i < 13; ++i) input[i] = 0; for (int i=5; i < 13; ++i) input[i] = 0;
input[13] = 0x00000001; input[13] = 0x00000001;
input[14] = T1; input[14] = T1;
@ -398,7 +386,7 @@ __global__ void cuda_blake256_hash( uint64_t *g_out, uint32_t nonce, uint32_t *g
if (validate) if (validate)
{ {
g_out += 4 * ((blockIdx.x * blockDim.x) + threadIdx.x); g_out += 4 * ((blockIdx.x * blockDim.x) + threadIdx.x);
#pragma unroll 4 #pragma unroll
for (int i=0; i < 4; ++i) g_out[i] = output[i]; for (int i=0; i < 4; ++i) g_out[i] = output[i];
} }

2
scrypt/fermi_kernel.cu
Unescape View File

@ -14,7 +14,7 @@
#include <map> #include <map>
#include "cuda_runtime.h" #include <cuda_runtime.h>
#include "miner.h" #include "miner.h"
#include "salsa_kernel.h" #include "salsa_kernel.h"

11
scrypt/keccak.cu
Unescape View File

@ -16,22 +16,17 @@
#include "salsa_kernel.h" #include "salsa_kernel.h"
// define some error checking macros // define some error checking macros
#undef checkCudaErrors
#if WIN32
#define DELIMITER '/'
#else
#define DELIMITER '/' #define DELIMITER '/'
#endif
#define __FILENAME__ ( strrchr(__FILE__, DELIMITER) != NULL ? strrchr(__FILE__, DELIMITER)+1 : __FILE__ ) #define __FILENAME__ ( strrchr(__FILE__, DELIMITER) != NULL ? strrchr(__FILE__, DELIMITER)+1 : __FILE__ )
#undef checkCudaErrors
#define checkCudaErrors(x) \ #define checkCudaErrors(x) \
{ \ { \
cudaGetLastError(); \ cudaGetLastError(); \
x; \ x; \
cudaError_t err = cudaGetLastError(); \ cudaError_t err = cudaGetLastError(); \
if (err != cudaSuccess) \ if (err != cudaSuccess && !abort_flag) \
applog(LOG_ERR, "GPU #%d: cudaError %d (%s) calling '%s' (%s line %d)\n", device_map[thr_id], err, cudaGetErrorString(err), #x, __FILENAME__, __LINE__); \ applog(LOG_ERR, "GPU #%d: cudaError %d (%s) (%s line %d)\n", device_map[thr_id], err, cudaGetErrorString(err), __FILENAME__, __LINE__); \
} }
// from salsa_kernel.cu // from salsa_kernel.cu

2
scrypt/kepler_kernel.cu
Unescape View File

@ -9,7 +9,7 @@
#include <map> #include <map>
#include "cuda_runtime.h" #include <cuda_runtime.h>
#include "miner.h" #include "miner.h"
#include "salsa_kernel.h" #include "salsa_kernel.h"

2
scrypt/nv_kernel2.cu
Unescape View File

@ -11,7 +11,7 @@
#include <map> #include <map>
#include "cuda_runtime.h" #include <cuda_runtime.h>
#include "miner.h" #include "miner.h"
#include "salsa_kernel.h" #include "salsa_kernel.h"

32
scrypt/salsa_kernel.cu
Unescape View File

@ -12,26 +12,16 @@
#include "salsa_kernel.h" #include "salsa_kernel.h"
#include "nv_kernel2.h"
#include "titan_kernel.h" #include "titan_kernel.h"
#include "fermi_kernel.h"
#include "test_kernel.h"
#include "nv_kernel.h" #include "nv_kernel.h"
#include "nv_kernel2.h"
#include "kepler_kernel.h" #include "kepler_kernel.h"
#include "fermi_kernel.h"
#include "test_kernel.h"
#include "miner.h" #include "miner.h"
#if WIN32 #if defined(_WIN64) || defined(__x86_64__) || defined(__64BIT__)
#ifdef _WIN64
#define _64BIT 1
#endif
#else
#if __x86_64__
#define _64BIT 1
#endif
#endif
#if _64BIT
#define MAXMEM 0x300000000ULL // 12 GB (the largest Kepler) #define MAXMEM 0x300000000ULL // 12 GB (the largest Kepler)
#else #else
#define MAXMEM 0xFFFFFFFFULL // nearly 4 GB (32 bit limitations) #define MAXMEM 0xFFFFFFFFULL // nearly 4 GB (32 bit limitations)
@ -42,21 +32,16 @@
#define DMIN 5 #define DMIN 5
// define some error checking macros // define some error checking macros
#undef checkCudaErrors
#if WIN32
#define DELIMITER '/'
#else
#define DELIMITER '/' #define DELIMITER '/'
#endif
#define __FILENAME__ ( strrchr(__FILE__, DELIMITER) != NULL ? strrchr(__FILE__, DELIMITER)+1 : __FILE__ ) #define __FILENAME__ ( strrchr(__FILE__, DELIMITER) != NULL ? strrchr(__FILE__, DELIMITER)+1 : __FILE__ )
#undef checkCudaErrors
#define checkCudaErrors(x) \ #define checkCudaErrors(x) \
{ \ { \
cudaGetLastError(); \ cudaGetLastError(); \
x; \ x; \
cudaError_t err = cudaGetLastError(); \ cudaError_t err = cudaGetLastError(); \
if (err != cudaSuccess) \ if (err != cudaSuccess && !abort_flag) \
applog(LOG_ERR, "GPU #%d: Err %d: %s (%s:%d)", device_map[thr_id], err, cudaGetErrorString(err), __FILENAME__, __LINE__); \ applog(LOG_ERR, "GPU #%d: Err %d: %s (%s:%d)", device_map[thr_id], err, cudaGetErrorString(err), __FILENAME__, __LINE__); \
} }
@ -78,7 +63,7 @@ KernelInterface *Best_Kernel_Heuristics(cudaDeviceProp *props)
kernel = new NV2Kernel(); // we don't want this for Keccak though kernel = new NV2Kernel(); // we don't want this for Keccak though
else if (props->major == 3 && props->minor == 0) else if (props->major == 3 && props->minor == 0)
kernel = new NVKernel(); kernel = new NVKernel();
else if (props->major == 2 || props->major == 1) else
kernel = new FermiKernel(); kernel = new FermiKernel();
} }
else else
@ -88,7 +73,7 @@ KernelInterface *Best_Kernel_Heuristics(cudaDeviceProp *props)
kernel = new TitanKernel(); kernel = new TitanKernel();
else if (props->major == 3 && props->minor == 0) else if (props->major == 3 && props->minor == 0)
kernel = new KeplerKernel(); kernel = new KeplerKernel();
else if (props->major == 2 || props->major == 1) else
kernel = new TestKernel(); kernel = new TestKernel();
} }
return kernel; return kernel;
@ -861,6 +846,7 @@ bool cuda_scrypt_sync(int thr_id, int stream)
if(err != cudaSuccess) if(err != cudaSuccess)
{ {
if (!abort_flag)
applog(LOG_ERR, "GPU #%d: CUDA error `%s` while executing the kernel.", device_map[thr_id], cudaGetErrorString(err)); applog(LOG_ERR, "GPU #%d: CUDA error `%s` while executing the kernel.", device_map[thr_id], cudaGetErrorString(err));
return false; return false;
} }

13
scrypt/sha256.cu
Unescape View File

@ -6,7 +6,7 @@
#include <map> #include <map>
#include "cuda_runtime.h" #include <cuda_runtime.h>
#include "miner.h" #include "miner.h"
#include "salsa_kernel.h" #include "salsa_kernel.h"
@ -14,21 +14,16 @@
#include "sha256.h" #include "sha256.h"
// define some error checking macros // define some error checking macros
#undef checkCudaErrors
#if WIN32
#define DELIMITER '/' #define DELIMITER '/'
#else
#define DELIMITER '/'
#endif
#define __FILENAME__ ( strrchr(__FILE__, DELIMITER) != NULL ? strrchr(__FILE__, DELIMITER)+1 : __FILE__ ) #define __FILENAME__ ( strrchr(__FILE__, DELIMITER) != NULL ? strrchr(__FILE__, DELIMITER)+1 : __FILE__ )
#undef checkCudaErrors
#define checkCudaErrors(x) { \ #define checkCudaErrors(x) { \
cudaGetLastError(); \ cudaGetLastError(); \
x; \ x; \
cudaError_t err = cudaGetLastError(); \ cudaError_t err = cudaGetLastError(); \
if (err != cudaSuccess) \ if (err != cudaSuccess && !abort_flag) \
applog(LOG_ERR, "GPU #%d: cudaError %d (%s) calling '%s' (%s line %d)\n", (int) device_map[thr_id], err, cudaGetErrorString(err), #x, __FILENAME__, __LINE__); \ applog(LOG_ERR, "GPU #%d: cudaError %d (%s) (%s line %d)\n", (int) device_map[thr_id], err, cudaGetErrorString(err), __FILENAME__, __LINE__); \
} }
// from salsa_kernel.cu // from salsa_kernel.cu

2
scrypt/test_kernel.cu
Unescape View File

@ -14,7 +14,7 @@
#include <map> #include <map>
#include "cuda_runtime.h" #include <cuda_runtime.h>
#include "miner.h" #include "miner.h"
#include "salsa_kernel.h" #include "salsa_kernel.h"

2
scrypt/titan_kernel.cu
Unescape View File

@ -9,7 +9,7 @@
#include <map> #include <map>
#include "cuda_runtime.h" #include <cuda_runtime.h>
#include "miner.h" #include "miner.h"
#include "salsa_kernel.h" #include "salsa_kernel.h"

Write Preview
Loading…
Cancel
Save