3.3.1.2.2 : Le fichier asterics_cuda.cpp
Développons le fichier asterics_cuda.cpp :
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 |
#include <stdlib.h> #include <cuda.h> #include <cuda_runtime.h> #include <sstream> #include <iomanip> #include <iostream> #include "asterics_cuda.h" ///Get the number of cuda device /** @return number of Cuda device */ int asterics_getNbCudaDevice(){ int deviceCount = 0; cudaError_t error_id = cudaGetDeviceCount(&deviceCount); if(error_id != cudaSuccess){ printf("asterics_getNbCudaDevice : cudaGetDeviceCount returned %d\n-> %s\n", (int)error_id, cudaGetErrorString(error_id)); printf("asterics_getNbCudaDevice : deviceCount = %d\n", deviceCount); printf("asterics_getNbCudaDevice : Result = FAIL\n"); exit(EXIT_FAILURE); } // This function call returns 0 if there are no CUDA capable devices. if(deviceCount == 0){ printf("asterics_getNbCudaDevice : There are no available device(s) that support CUDA\n"); }else{ printf("asterics_getNbCudaDevice : Detected %d CUDA Capable device(s)\n", deviceCount); } return deviceCount; } ///Get some usefull GPU info /** @param[out] maxNbThreadPerBlockX : maximum number of thread per block on X * @param[out] maxNbBlockX : maximum number of block in the grid on X * @param[out] warpSize : number of thread per warp * @param deviceIndex : index of the device to be used */ void asterics_getGpuInfo(int & maxNbThreadPerBlockX, int & maxNbBlockX, int & warpSize, int deviceIndex){ cudaSetDevice(deviceIndex); cudaDeviceProp deviceProp; cudaGetDeviceProperties(&deviceProp, deviceIndex); maxNbThreadPerBlockX = deviceProp.maxThreadsDim[0]; maxNbBlockX = deviceProp.maxGridSize[0]; warpSize = deviceProp.warpSize; } ///Get some usefull GPU info /** @param[out] maxNbThreadPerBlockX : maximum number of thread per block on X * @param[out] maxNbBlockX : maximum number of block in the grid on X * @param[out] memSize : number of bytes in the GPU memory * @param deviceIndex : index of the device to be used */ void asterics_getGpuInfoMem(size_t & maxNbThreadPerBlockX, size_t & maxNbBlockX, size_t & memSize, int deviceIndex){ cudaSetDevice(deviceIndex); cudaDeviceProp deviceProp; cudaGetDeviceProperties(&deviceProp, deviceIndex); maxNbThreadPerBlockX = deviceProp.maxThreadsDim[0]; maxNbBlockX = deviceProp.maxGridSize[0]; memSize = deviceProp.totalGlobalMem; } ///Get the UUID of the given Cuda Device /** @param deviceProp : Cuda Device to be used * @return corresponding UUID */ std::string asterics_getUuid(const cudaDeviceProp & deviceProp){ const unsigned char * tabByte = (const unsigned char *)deviceProp.uuid.bytes; std::stringstream out; for(int i(0); i < 16; ++i){ // printf("%02x,", (unsigned int)(tabByte[i])); out << std::hex << std::setfill('0') << std::setw(2) << (unsigned int)tabByte[i]; } return "GPU-" + out.str(); } ///Check if two string start the same way /** @param str : string to be tested @param beginig : begining to be checked @return true if str starts as beginig */ bool asterics_isSameBegining(const std::string & str, const std::string & beginig){ if(str.size() < beginig.size()) return false; std::string::const_iterator it = str.begin(); std::string::const_iterator it2 = beginig.begin(); while(it != str.end() && it2 != beginig.end()){ if(*it != *it2){ return false;} it++; it2++; } return true; } ///Set the cuda device /** @param uuidSelectedDevice : uuid of the device to be used (or the begining) * @return id of the choosen device */ int asterics_setDevice(const std::string & uuidSelectedDevice){ int deviceCount = asterics_getNbCudaDevice(); for(int dev = 0; dev < deviceCount; ++dev){ cudaSetDevice(dev); cudaDeviceProp deviceProp; cudaGetDeviceProperties(&deviceProp, dev); std::string uuid(asterics_getUuid(deviceProp)); if(asterics_isSameBegining(uuid, uuidSelectedDevice)){ std::cout << "asterics_setDevice : set device " << dev << " with UUID = " << uuid << std::endl; cudaSetDevice(dev); return dev; } } return 0; } |
Le fichier asterics_cuda.cpp complet.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 |
/*************************************** Auteur : Pierre Aubert Mail : pierre.aubert@lapp.in2p3.fr Licence : CeCILL-C ****************************************/ #include <stdlib.h> #include <cuda.h> #include <cuda_runtime.h> #include <sstream> #include <iomanip> #include <iostream> #include "asterics_cuda.h" ///Get the number of cuda device /** @return number of Cuda device */ int asterics_getNbCudaDevice(){ int deviceCount = 0; cudaError_t error_id = cudaGetDeviceCount(&deviceCount); if(error_id != cudaSuccess){ printf("asterics_getNbCudaDevice : cudaGetDeviceCount returned %d\n-> %s\n", (int)error_id, cudaGetErrorString(error_id)); printf("asterics_getNbCudaDevice : deviceCount = %d\n", deviceCount); printf("asterics_getNbCudaDevice : Result = FAIL\n"); exit(EXIT_FAILURE); } // This function call returns 0 if there are no CUDA capable devices. if(deviceCount == 0){ printf("asterics_getNbCudaDevice : There are no available device(s) that support CUDA\n"); }else{ printf("asterics_getNbCudaDevice : Detected %d CUDA Capable device(s)\n", deviceCount); } return deviceCount; } ///Get some usefull GPU info /** @param[out] maxNbThreadPerBlockX : maximum number of thread per block on X * @param[out] maxNbBlockX : maximum number of block in the grid on X * @param[out] warpSize : number of thread per warp * @param deviceIndex : index of the device to be used */ void asterics_getGpuInfo(int & maxNbThreadPerBlockX, int & maxNbBlockX, int & warpSize, int deviceIndex){ cudaSetDevice(deviceIndex); cudaDeviceProp deviceProp; cudaGetDeviceProperties(&deviceProp, deviceIndex); maxNbThreadPerBlockX = deviceProp.maxThreadsDim[0]; maxNbBlockX = deviceProp.maxGridSize[0]; warpSize = deviceProp.warpSize; } ///Get some usefull GPU info /** @param[out] maxNbThreadPerBlockX : maximum number of thread per block on X * @param[out] maxNbBlockX : maximum number of block in the grid on X * @param[out] memSize : number of bytes in the GPU memory * @param deviceIndex : index of the device to be used */ void asterics_getGpuInfoMem(size_t & maxNbThreadPerBlockX, size_t & maxNbBlockX, size_t & memSize, int deviceIndex){ cudaSetDevice(deviceIndex); cudaDeviceProp deviceProp; cudaGetDeviceProperties(&deviceProp, deviceIndex); maxNbThreadPerBlockX = deviceProp.maxThreadsDim[0]; maxNbBlockX = deviceProp.maxGridSize[0]; memSize = deviceProp.totalGlobalMem; } ///Get the UUID of the given Cuda Device /** @param deviceProp : Cuda Device to be used * @return corresponding UUID */ std::string asterics_getUuid(const cudaDeviceProp & deviceProp){ const unsigned char * tabByte = (const unsigned char *)deviceProp.uuid.bytes; std::stringstream out; for(int i(0); i < 16; ++i){ // printf("%02x,", (unsigned int)(tabByte[i])); out << std::hex << std::setfill('0') << std::setw(2) << (unsigned int)tabByte[i]; } return "GPU-" + out.str(); } ///Check if two string start the same way /** @param str : string to be tested @param beginig : begining to be checked @return true if str starts as beginig */ bool asterics_isSameBegining(const std::string & str, const std::string & beginig){ if(str.size() < beginig.size()) return false; std::string::const_iterator it = str.begin(); std::string::const_iterator it2 = beginig.begin(); while(it != str.end() && it2 != beginig.end()){ if(*it != *it2){ return false;} it++; it2++; } return true; } ///Set the cuda device /** @param uuidSelectedDevice : uuid of the device to be used (or the begining) * @return id of the choosen device */ int asterics_setDevice(const std::string & uuidSelectedDevice){ int deviceCount = asterics_getNbCudaDevice(); for(int dev = 0; dev < deviceCount; ++dev){ cudaSetDevice(dev); cudaDeviceProp deviceProp; cudaGetDeviceProperties(&deviceProp, dev); std::string uuid(asterics_getUuid(deviceProp)); if(asterics_isSameBegining(uuid, uuidSelectedDevice)){ std::cout << "asterics_setDevice : set device " << dev << " with UUID = " << uuid << std::endl; cudaSetDevice(dev); return dev; } } return 0; } |
Vous pouvez télécharger le fichier ici.