关于矩阵相乘

#include<stdio.h>

#include<stdlib.h>

#include<CL/cl.hpp>

//矩阵相乘核心算法

const char *KernelSource=

"__kernel void SimpleMultiply(__global int *outputC,int widthA,int heightA,int widthB,int heightB,__global int *inputA,__global int *inputB)"

"{"

"    int row=get_global_id(1);"

"    int col=get_global_id(0);"

"    int sum=0;"

"    for(int i=0;i<widthA;i++)"

"    {"

"        sum+=inputA[row*widthA+i]*inputB[i*widthB+col];"

"     }"

"     outputC[row*widthB+col]=sum;"

"};";

int main()

{

const int heightA=6;

const int widthA=8;

const int heightB=8;

const int widthB=10;

const int heightC=6;

const int widthC=10;

int i,j,k;

int A[heightA][widthA]=

{

1,2,3,4,5,6,7,8,

1,2,3,4,5,6,7,8,

1,2,3,4,5,6,7,8,

1,2,3,4,5,6,7,8,

1,2,3,4,5,6,7,8,

1,2,3,4,5,6,7,8

};

int B[heightB][widthB]=

{

10,9,8,7,6,5,4,3,2,1,

10,9,8,7,6,5,4,3,2,1,

10,9,8,7,6,5,4,3,2,1,

10,9,8,7,6,5,4,3,2,1,

10,9,8,7,6,5,4,3,2,1,

10,9,8,7,6,5,4,3,2,1,

10,9,8,7,6,5,4,3,2,1,

10,9,8,7,6,5,4,3,2,1

};

int C[heightC][widthC];

cl_int err;

cl_platform_id platform;

err=clGetPlatformIDs(1,&platform,NULL);

cl_device_id device;

err=clGetDeviceIDs(platform,CL_DEVICE_TYPE_GPU,1,&device,NULL);

cl_context context;

context=clCreateContext(NULL,1,&device,NULL,NULL,&err);

cl_command_queue cmdQueue;

cmdQueue=clCreateCommandQueue(context,device,NULL,&err);

cl_mem BufferA;

cl_mem BufferB;

cl_mem BufferC;

BufferA=clCreateBuffer(context,CL_MEM_READ_ONLY,sizeof(int)*widthA*heightA,NULL,&err);

BufferB=clCreateBuffer(context,CL_MEM_READ_ONLY,sizeof(int)*widthB*heightB,NULL,&err);

BufferC=clCreateBuffer(context,CL_MEM_WRITE_ONLY,sizeof(int)*widthC*heightC,NULL,&err);

err=clEnqueueWriteBuffer(cmdQueue,BufferA,CL_TRUE,0,sizeof(int)*widthA*heightA,A,0,NULL,NULL);

err=clEnqueueWriteBuffer(cmdQueue,BufferB,CL_TRUE,0,sizeof(int)*widthB*heightB,B,0,NULL,NULL);

cl_program program;

program=clCreateProgramWithSource(context,1,&KernelSource,NULL,&err);

err=clBuildProgram(program,1,&device,NULL,NULL,NULL);

cl_kernel kernel;

kernel=clCreateKernel(program,"SimpleMultiply",&err);

err=clSetKernelArg(kernel,0,sizeof(cl_mem),&BufferC);

err=clSetKernelArg(kernel,1,sizeof(int),&widthA);

err=clSetKernelArg(kernel,2,sizeof(int),&heightA);

err=clSetKernelArg(kernel,3,sizeof(int),&widthB);

err=clSetKernelArg(kernel,4,sizeof(int),&heightB);

err=clSetKernelArg(kernel,5,sizeof(cl_mem),&BufferA);

err=clSetKernelArg(kernel,6,sizeof(cl_mem),&BufferB);

//size_t localWorkSize[2]={16,16};

size_t globalWorkSize[2]={widthC,heightC};

err=clEnqueueNDRangeKernel(cmdQueue,kernel,2,NULL,globalWorkSize,NULL,0,NULL,NULL);

err=clEnqueueReadBuffer(cmdQueue,BufferC,CL_TRUE,0,sizeof(int)*widthC*heightC,C,0,NULL,NULL);

for(i=0;i<heightC;i++)

{

for(j=0;j<widthC;j++)

{

printf("%d ",C[i][j]);

}

printf("\n");

}

system("pause");

}