Windows Opencl编程：第二天（Opencl GPU的加法运算）

Windows C++ Opencl编程：第二天

今日份踩坑，Opencl GPU的加法运算

第一部分：main.cpp

#include<CL/opencl.h>
#include<iostream>
#pragma comment(lib,"OPENCL.lib")
using namespace std;
void main()
{
	cl_int err;
	cl_uint counts;
	cl_context context = NULL;
	err=clGetPlatformIDs(0, 0, &counts);
	if (err != CL_SUCCESS)puts("获取可用平台数量失败！"); else printf("获取可用平台数量成功！%d\n",counts);
	cl_platform_id *platm = new cl_platform_id[counts];
	err = clGetPlatformIDs(counts, platm, NULL);
	if (err != CL_SUCCESS)puts("获取可用平台失败！"); else puts("获取可用平台成功！");
	for (int i = 0; i < counts; i++)
	{
		size_t size;
		err = clGetPlatformInfo(platm[i], CL_PLATFORM_NAME, 0, NULL, &size);
		char *name = new char[size];
		err = clGetPlatformInfo(platm[i], CL_PLATFORM_NAME, size, name, NULL);
		cout << "GPU" << i + 1 << "名称：" << name << endl;
		delete name;
	}
	cl_device_id cdi1;
	clGetDeviceIDs(platm[1], CL_DEVICE_TYPE_GPU, 1, &cdi1, NULL);//查询平台上的设备
	context = clCreateContext(NULL, 1, &cdi1, 0, 0, &err);//创建上下文
	cl_command_queue ccq= clCreateCommandQueue(context, cdi1, 0, &err);//创建命令队列
	FILE *fp = fopen("add.cl", "rb");
	char t[50];
	fseek(fp, 0, SEEK_END);
	int fileset = ftell(fp);
	fseek(fp, 0, SEEK_SET);
	char ch; int i = 0; char *clcode =new char[fileset+1];
	fread(clcode, 1, fileset, fp);
	clcode[fileset] = 0;
	fclose(fp);
	printf("%s\n", clcode);
	cl_program cp= clCreateProgramWithSource(context, 1,(const char**) &clcode, NULL, &err);
	err=clBuildProgram(cp, 0, 0, 0, 0, 0);
	cl_kernel ck= clCreateKernel(cp, "add", &err);

	int *a = new int[100], *b = new int[100];
	for (int j = 0; j < 100; j++)
		a[j] = j;
	cl_mem mema=clCreateBuffer(context, CL_MEM_COPY_HOST_PTR | CL_MEM_READ_ONLY, sizeof(int) * 100, a, &err);
	cl_mem memb= clCreateBuffer(context, CL_MEM_READ_WRITE, sizeof(int) * 100, NULL, &err);
	err=clSetKernelArg(ck, 0, sizeof(cl_mem), &mema);
	err = clSetKernelArg(ck, 1, sizeof(cl_mem), &memb);
	size_t  gws[] = { 100 }, lws = { 1 };
	err=clEnqueueNDRangeKernel(ccq, ck, 1, 0,(const size_t *) &gws ,&lws,0, 0, 0);
	err=clEnqueueReadBuffer(ccq, memb, CL_TRUE, 0, sizeof(int) * 100,b, 0, 0,0);
	for (int j = 0; j < 100; j++)
	{
		j % 10 == 0 ? puts("") : 0;
		printf("%04d ", a[j]);
	}
	puts("");
	puts("GPU累加运算");
	for (int j = 0; j < 100; j++)
	{
		j % 10 == 0 ? puts("") : 0;
		printf("%04d ", b[j]);
	}
	getchar();
}

第二部分：add.cl

kernel void add(global const int *a,global int *b)
{
	int gid=get_global_id(0);
	b[gid]=a[gid]+a[gid];
}

运行结果