C++ clGetDeviceIDs函数代码示例

int main() {
	char buf[]="Hello, World!";
	size_t srcsize, worksize=strlen(buf);
	
	cl_int error;
	cl_platform_id platform;
	cl_device_id device;
	cl_uint platforms, devices;

	// Fetch the Platform and Device IDs; we only want one.
	error=clGetPlatformIDs(1, &platform, &platforms);
	error=clGetDeviceIDs(platform, CL_DEVICE_TYPE_ALL, 1, &device, &devices);
	cl_context_properties properties[]={
		CL_CONTEXT_PLATFORM, (cl_context_properties)platform,
		0};
	// Note that nVidia's OpenCL requires the platform property
	cl_context context=clCreateContext(properties, 1, &device, NULL, NULL, &error);
	cl_command_queue cq = clCreateCommandQueue(context, device, 0, &error);
	
	rot13(buf);	// scramble using the CPU
	puts(buf);	// Just to demonstrate the plaintext is destroyed

	//char src[8192];
	//FILE *fil=fopen("rot13.cl","r");
	//srcsize=fread(src, sizeof src, 1, fil);
	//fclose(fil);
	
	const char *src=rot13_cl;
	srcsize=strlen(rot13_cl);

	const char *srcptr[]={src};
	// Submit the source code of the rot13 kernel to OpenCL
	cl_program prog=clCreateProgramWithSource(context,
		1, srcptr, &srcsize, &error);
	// and compile it (after this we could extract the compiled version)
	error=clBuildProgram(prog, 0, NULL, "", NULL, NULL);

	// Allocate memory for the kernel to work with
	cl_mem mem1, mem2;
	mem1=clCreateBuffer(context, CL_MEM_READ_ONLY, worksize, NULL, &error);
	mem2=clCreateBuffer(context, CL_MEM_WRITE_ONLY, worksize, NULL, &error);
	
	// get a handle and map parameters for the kernel
	cl_kernel k_rot13=clCreateKernel(prog, "rot13", &error);
	clSetKernelArg(k_rot13, 0, sizeof(mem1), &mem1);
	clSetKernelArg(k_rot13, 1, sizeof(mem2), &mem2);

	// Target buffer just so we show we got the data from OpenCL
	char buf2[sizeof buf];
	buf2[0]='?';
	buf2[worksize]=0;

	// Send input data to OpenCL (async, don't alter the buffer!)
	error=clEnqueueWriteBuffer(cq, mem1, CL_FALSE, 0, worksize, buf, 0, NULL, NULL);
	// Perform the operation
	error=clEnqueueNDRangeKernel(cq, k_rot13, 1, NULL, &worksize, &worksize, 0, NULL, NULL);
	// Read the result back into buf2
	error=clEnqueueReadBuffer(cq, mem2, CL_FALSE, 0, worksize, buf2, 0, NULL, NULL);
	// Await completion of all the above
	error=clFinish(cq);
	
	// Finally, output out happy message.
	puts(buf2);
}

int main() {
    // Get platform information
    err = clGetPlatformIDs(0, NULL, &numOfPlatforms);
    if (err) Error("Fail to get the number of platforms");
    printf("The machine has %d platform(s) for OpenCL.\n", numOfPlatforms);

    platformIDs = new cl_platform_id [numOfPlatforms];
    err = clGetPlatformIDs(numOfPlatforms, platformIDs, NULL);
    if (err) Error("Fail to get the platform list");

    int cudaPlatformID = -1;

    for (int i = 0; i < numOfPlatforms; i++) {
        char platformName[50];
        err = clGetPlatformInfo(platformIDs[i], CL_PLATFORM_NAME, 50, platformName, NULL);
        if (err) Error("Fail to get the platform name");
        printf("Platform %d is %s\n", i + 1, platformName);
        if (!strcmp(platformName, "NVIDIA CUDA")) cudaPlatformID = i;
    }
    printf("\n");

    if (cudaPlatformID == -1) Error("Fail to find an NVIDIA CUDA platform");

    printf("Platform %d (NVIDIA CUDA) is chosen for use.\n", cudaPlatformID + 1);
    printf("\n");

    // Get device information
    err = clGetDeviceIDs(platformIDs[cudaPlatformID], CL_DEVICE_TYPE_GPU, 0, NULL, &numOfDevices);
    if (err) Error("Fail to get the number of devices");
    printf("CUDA platform has %d device(s).\n", numOfDevices);

    deviceIDs = new cl_device_id [numOfDevices];
    err = clGetDeviceIDs(platformIDs[cudaPlatformID], CL_DEVICE_TYPE_GPU, numOfDevices, deviceIDs, NULL);
    if (err) Error("Fail to get the device list");
    for (int i = 0; i < numOfDevices; i++) {
        char deviceName[50];
        err = clGetDeviceInfo(deviceIDs[i], CL_DEVICE_NAME, 50, deviceName, NULL);
        if (err) Error("Fail to get the device name");
        printf("Device %d is %s\n", i + 1, deviceName);
    }
    printf("\n");

    // Create a context
    context = clCreateContext(NULL, numOfDevices, deviceIDs, NULL, NULL, &err);
    if (err) Error("Fail to create a context");

    printf("Device 1 is chosen for use.\n");
    printf("\n");

    // Create a command queue for the first device
    commandQueue = clCreateCommandQueue(context, deviceIDs[0],
                                        CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE | CL_QUEUE_PROFILING_ENABLE, &err);
    if (err) Error("Fail to create a command queue");

    // create the program
    cl_program program = CreateProgram(exclusiveScanKernels, "exclusive scan");

    // create two kernels
    cl_kernel scanKernel = clCreateKernel(program, "Scan", &err);
    if (err) Error("Fail to create the kernel for scan");

    cl_kernel reverseUpdateKernel = clCreateKernel(program, "ReverseUpdate", &err);
    if (err) Error("Fail to create the kernel for reverse update");

    // Get the work group size
    size_t maxWorkGroupSize;
    err = clGetKernelWorkGroupInfo(scanKernel, deviceIDs[0], CL_KERNEL_WORK_GROUP_SIZE,
                                   sizeof(size_t), &maxWorkGroupSize, NULL);
    printf("maxWorkGroupSize = %d\n", maxWorkGroupSize);

    err = clGetKernelWorkGroupInfo(reverseUpdateKernel, deviceIDs[0], CL_KERNEL_WORK_GROUP_SIZE,
                                   sizeof(size_t), &maxWorkGroupSize, NULL);
    printf("maxWorkGroupSize = %d\n", maxWorkGroupSize);

    // Set work group size to 64

    int workGroupSize = 512;

    int length = 2048000;
    int *arr = new int [length];
    for (int i = 0; i < length; i++)
        arr[i] = rand() % 100;

    int *prefixSum = new int [length];
    prefixSum[0] = 0;

    int t0 = clock();

    for (int i = 1; i < length; i++)
        prefixSum[i] = prefixSum[i - 1] + arr[i - 1];

    int t1 = clock();

    printf("time1: %lf\n", (t1 - t0) * 1.0 / CLOCKS_PER_SEC);

    cl_mem d_arr = clCreateBuffer(context, CL_MEM_READ_WRITE, sizeof(int) * length, NULL, &err);
    if (err) Error("Fail to create d_arr");

    err = clEnqueueWriteBuffer(commandQueue, d_arr, CL_TRUE, 0, sizeof(int) * length, arr, 0, NULL, NULL);
    if (err) Error("Fail to write d_arr");
//.........这里部分代码省略.........

void sum_gpu(long long *in, long long *out, unsigned int n)
{
	size_t global_size;
	size_t local_size;

	char *kernel_src;

	cl_int err;
	cl_platform_id platform_id;
	cl_device_id device_id;
	cl_uint max_compute_units;
	size_t max_workgroup_size;

	cl_context context;
	cl_command_queue commands;
	cl_program program;
	cl_kernel kernel;
	cl_mem d_array;

	cl_event event;
	cl_ulong start, end;

	/* start OpenCL */
	err = clGetPlatformIDs(1, &platform_id,NULL);
	clErrorHandling("clGetPlatformIDs");

	err = clGetDeviceIDs(platform_id, CL_DEVICE_TYPE_GPU, 1, &device_id, NULL);
	clErrorHandling("clGetDeviceIDs");

	context = clCreateContext(0, 1, &device_id, NULL, NULL, &err);
	clErrorHandling("clCreateContext");

	commands = clCreateCommandQueue(context, device_id, CL_QUEUE_PROFILING_ENABLE, &err);
	clErrorHandling("clCreateCommandQueue");

	/* create kernel */
	kernel_src = file_to_string(KERNEL_SRC);
	program = clCreateProgramWithSource(context, 1, (const char**) &kernel_src, NULL, &err);
	free(kernel_src);
	clErrorHandling("clCreateProgramWithSource");

	err = clBuildProgram(program, 0, NULL, NULL, NULL, NULL);
	clErrorHandling("clBuildProgram");

	kernel = clCreateKernel(program, "matrix_mult", &err);
	clErrorHandling("clCreateKernel");

	/* allocate memory and send to gpu */
	d_array = clCreateBuffer(context, CL_MEM_READ_WRITE, sizeof(long long) * n, NULL, &err);
	clErrorHandling("clCreateBuffer");

	err = clEnqueueWriteBuffer(commands, d_array, CL_TRUE, 0, sizeof(long long) * n, in, 0, NULL, NULL);
	clErrorHandling("clEnqueueWriteBuffer");

	err  = clGetDeviceInfo(device_id, CL_DEVICE_MAX_COMPUTE_UNITS, sizeof(cl_uint), &max_compute_units, NULL);
	err |= clGetDeviceInfo(device_id, CL_DEVICE_MAX_WORK_GROUP_SIZE, sizeof(size_t), &max_workgroup_size, NULL);
	clErrorHandling("clGetDeviceInfo");

	/* prepare kernel args */
	err  = clSetKernelArg(kernel, 0, sizeof(cl_mem), &d_array);
	err |= clSetKernelArg(kernel, 1, sizeof(unsigned int), &n);

	/* execute */
	local_size = n / max_compute_units / 8;
	if (local_size > max_workgroup_size)
		local_size = max_workgroup_size;

	/*
	 *	Usually it would be
	 *	global_size = local_size * max_compute_units;
	 *	but that would only be valid if local_size = n / max_compute_units;
	 *	local_size is n / max_compute_units / 8 because it obtains its hightest performance.
	 */
	for (global_size = local_size; global_size < n; global_size += local_size);

	err = clEnqueueNDRangeKernel(commands, kernel, 1, NULL, &global_size, &local_size, 0, NULL, &event);
	clErrorHandling("clEnqueueNDRangeKernel");

	clWaitForEvents(1, &event);
	clGetEventProfilingInfo(event, CL_PROFILING_COMMAND_START, sizeof(cl_ulong), &start, NULL);
	clGetEventProfilingInfo(event, CL_PROFILING_COMMAND_END, sizeof(cl_ulong), &end, NULL);
	fprintf(stderr, "Time for event (ms): %10.5f \n", (end - start) / 1000000.0);

	err = clFinish(commands);
	clErrorHandling("clFinish");

	/* transfer back */
	err = clEnqueueReadBuffer(commands, d_array, CL_TRUE, 0, sizeof(long long), out, 0, NULL, NULL); // a single long long
	clErrorHandling("clEnqueueReadBuffer");

	/* cleanup*/
	clReleaseMemObject(d_array);
	clReleaseProgram(program);
	clReleaseKernel(kernel);
	clReleaseCommandQueue(commands);
	clReleaseContext(context);
	clReleaseEvent(event);
}

void InitOpenCL()
{
    // 1. Get a platform.
    cl_platform_id platform;
    
    clGetPlatformIDs( 1, &platform, NULL );
    // 2. Find a gpu device.
    cl_device_id device;
    
    clGetDeviceIDs( platform, CL_DEVICE_TYPE_GPU,
                   1,
                   &device,
                   NULL);
    // 3. Create a context and command queue on that device.
    cl_context context = clCreateContext( NULL,
                                         1,
                                         &device,
                                         NULL, NULL, NULL);
    queue = clCreateCommandQueue( context,
                                 device,
                                 0, NULL );
    // 4. Perform runtime source compilation, and obtain kernel entry point.
    std::ifstream file("scene.cl");
    std::string source;
    if (file){
    while(!file.eof()){
        char line[256];
        file.getline(line,255);
        source += std::string(line) + "\n";
    }
    }
    if (source.length()==0)
    {
        std::string err = "fail to load shader";
    }
    
    cl_ulong maxSize;
    clGetDeviceInfo(device, CL_DEVICE_MAX_MEM_ALLOC_SIZE , sizeof(cl_ulong), &maxSize, 0);
    
    const char* str = source.c_str();
    cl_program program = clCreateProgramWithSource( context,
                                                   1,
                                                   &str,
                                                   NULL, NULL );
    cl_int result = clBuildProgram( program, 1, &device, NULL, NULL, NULL );
    if ( result ){
        char* build_log;
        size_t log_size;
        clGetProgramBuildInfo(program, device, CL_PROGRAM_BUILD_LOG, 0, NULL, &log_size);
        build_log = new char[log_size+1];
        clGetProgramBuildInfo(program, device, CL_PROGRAM_BUILD_LOG, log_size, build_log, NULL);
        build_log[log_size] = '\0';
        if( log_size > 2 ) {
            std::cout << "build log: " << build_log << std::endl;
        }
        delete[] build_log;
        std::cout << "Error during compilation! (" << result << ")" << std::endl;
    }
    kernel = clCreateKernel( program, "tracekernel", NULL );
    // 5. Create a data buffer.
    buffer        = clCreateBuffer( context,
                                   CL_MEM_WRITE_ONLY,
                                   kWidth * kHeight *sizeof(cl_float4),
                                   NULL, 0 );
    viewTransform = clCreateBuffer( context,
                                   CL_MEM_READ_WRITE,
                                   16 *sizeof(cl_float),
                                   NULL, 0 );
    
    worldTransforms = clCreateBuffer( context,
                                     CL_MEM_READ_WRITE,
                                     16 *sizeof(cl_float)*2,
                                     NULL, 0 );
    
    clSetKernelArg(kernel, 0, sizeof(buffer), (void*) &buffer);
    clSetKernelArg(kernel, 1, sizeof(cl_uint), (void*) &kWidth);
    clSetKernelArg(kernel, 2, sizeof(cl_uint), (void*) &kWidth);
    clSetKernelArg(kernel, 3, sizeof(viewTransform), (void*) &viewTransform);
    clSetKernelArg(kernel, 4, sizeof(worldTransforms), (void*) &worldTransforms);
}

//.........这里部分代码省略.........
            perror("Couldn't read CL_PLATFORM_VENDER data.");
        }
        ext_data = (char *)malloc(ext_size);
        clGetPlatformInfo(platform, CL_PLATFORM_VENDER, ext_size, ext_data, NULL);
        printf("Platform %d vender: %s\n", i, ext_data);
        free(ext_data)

        /*VERSION*/
        err = clGetPlatformInfo(platform, CL_PLATFORM_VERSION, 0, NULL, &ext_size);
        if(err < 0) {
            perror("Couldn't read CL_PLATFORM_VERSION data.");
        }
        ext_data = (char *)malloc(ext_size);
        clGetPlatformInfo(platform, CL_PLATFORM_VERSION, ext_size, ext_data, NULL);
        printf("Platform %d support OpenCL version: %s\n", i, ext_data);
        free(ext_data)
        
        /*PROFILE*/
        err = clGetPlatformInfo(platform, CL_PLATFORM_PROFILE, 0, NULL, &ext_size);
        if(err < 0) {
            perror("Couldn't read CL_PLATFORM_PROFILE data.");
        }
        ext_data = (char *)malloc(ext_size);
        clGetPlatformInfo(platform, CL_PLATFORM_PROFILE, ext_size, ext_data, NULL);
        printf("Platform %d support OpenCL profile: %s\n", i, ext_data);
        free(ext_data)
        
        /*EXTENSIONS*/
        err = clGetPlatformInfo(platform, CL_PLATFORM_EXTENSIONS, 0, NULL, &ext_size);
        if(err < 0) {
            perror("Couldn't read CL_PLATFORM_EXTENSIONS data.");
        }
        ext_data = (char *)malloc(ext_size);
        clGetPlatformInfo(platform, CL_PLATFORM_EXTENSIONS, ext_size, ext_data, NULL);
        printf("Platform %d support OpenCL extensions: %s\n", i, ext_data);
        free(ext_data)
        
        /*现在对这个platform进一步的提取信息*/
        /*获取Device信息*/
        /*clGetDeviceIDs*/
        /*
        参数1:平台句柄
        参数2:要获取设备的类型
        参数3:要获取的数量
        参数4:返回信息的数据指针
        参数5:返回信息的实际条数
        */
        for (cnt=0; cnt<2; cnt++)
        {
            if (0==cnt) err = clGetDeviceIDs(platform, CL_DEVICE_TYPE_GPU, 0, NULL, &num_devs);
            if (1==cnt) err = clGetDeviceIDs(platform, CL_DEVICE_TYPE_CPU, 0, NULL, &num_devs);
            if(err == CL_DEVICE_NOT_FOUND)
            {
                if (0==cnt) printf("No GPU support OpenCL found.\n");
                if (1==cnt) printf("No CPU support OpenCL found.\n");
            }
            else if(err < 0)
            {
                if (0==cnt) printf("Couldn't access any GPU devices.\n");
                if (1==cnt) printf("Couldn't access any CPU devices.\n");
            }
            else
            {
                devs = (cl_device_id *)malloc( sizeof(cl_device_id) * num_devs );
                clGetDeviceIDs(platform, CL_DEVICE_TYPE_GPU, num_devs, devs, NULL);
                for (j=0; j<num_devs; j++)
                {
                    cl_device_id dev;
                    char dev_name_data[48];
                    cl_uint addr_data;
                    cl_ulong global_mem_size;
                    
                    /*name*/
                    err = clGetDeviceInfo(dev, CL_DEVICE_NAME, 48 * sizeof(char), dev_name_data, NULL);
                    if(err < 0) {
                        perror("Couldn't read dev name data");
                        exit(1);
                    }
                    printf("Dev %d: NAME: %s\n", j, name_data);
                                    
                    /*address size*/
                    err = clGetDeviceInfo(dev, CL_DEVICE_ADDRESS_BITS, sizeof(addr_data), &addr_data, NULL);
                    printf("Dev %d: ADDRESS_WIDTH: %u\n", j, addr_data);
                    
                    /*device extensions*/
                    ext_data=malloc(4096)
                    clGetDeviceInfo(dev, CL_DEVICE_EXTENSIONS, 4096 * sizeof(char), ext_data, NULL);
                    printf("Dev %d: EXTENSIONS: %s\n", j, ext_data);
                    free(ext_data)
                    
                    err = clGetDeviceInfo(dev, CL_DEVICE_GLOBAL_MEM_SIZE, sizeof(global_mem_size), &global_mem_size, NULL);
                    printf("Dev %d: GLOBAL_MEM_SIZE: %u\n", j, global_mem_size);
                }
            }

        }
    }

    return 0;
}

int main(void) {
    const size_t ARRAY_BYTES = ARRAY_SIZE * sizeof(float);

    // Generate the input array on the host.
    float h_a[ARRAY_SIZE];
    float h_b[ARRAY_SIZE];
    for (int i = 0; i < ARRAY_SIZE; i++) {
        h_a[i] = (float)i;
        h_b[i] = (float)(2 * i);
    }

    float h_c[ARRAY_SIZE];

    FILE *fp;
    char *source_str;
    size_t source_size;

    fp = fopen("vectors_cl.cl", "r");
    if (!fp) {
        fprintf(stderr, "Failed to load kernel.\n");
        exit(1);
    }
    source_str = (char *)malloc(MAX_SOURCE_SIZE);
    source_size = fread(source_str, 1, MAX_SOURCE_SIZE, fp);
    fclose(fp);

    // Get platform and device information
    cl_platform_id platform_id = NULL;
    cl_device_id device_id = NULL;
    cl_uint ret_num_devices;
    cl_uint ret_num_platforms;
    cl_int ret = clGetPlatformIDs(1, &platform_id, &ret_num_platforms);
    ret = clGetDeviceIDs(platform_id, CL_DEVICE_TYPE_DEFAULT, 1,
                         &device_id, &ret_num_devices);

    // Create an OpenCL context
    cl_context context = clCreateContext(NULL, 1, &device_id, NULL, NULL, &ret);

    // Create a command queue
    cl_command_queue command_queue = clCreateCommandQueue(context, device_id, 0, &ret);

    // Create memory buffers on the device for each vector
    cl_mem a_mem_obj = clCreateBuffer(context, CL_MEM_READ_ONLY,
                                      ARRAY_BYTES, NULL, &ret);
    cl_mem b_mem_obj = clCreateBuffer(context, CL_MEM_READ_ONLY,
                                      ARRAY_BYTES, NULL, &ret);
    cl_mem c_mem_obj = clCreateBuffer(context, CL_MEM_WRITE_ONLY,
                                      ARRAY_BYTES, NULL, &ret);

    // Copy h_a and h_b to memory buffer
    ret = clEnqueueWriteBuffer(command_queue, a_mem_obj, CL_TRUE, 0,
                               ARRAY_BYTES, h_a, 0, NULL, NULL);
    ret = clEnqueueWriteBuffer(command_queue, b_mem_obj, CL_TRUE, 0,
                               ARRAY_BYTES, h_b, 0, NULL, NULL);

    // Create a program from the kernel source
    cl_program program = clCreateProgramWithSource(context, 1,
        (const char **)&source_str, (const size_t *)&source_size, &ret);
    if (ret != 0) {
        printf("clCreateProgramWithSource returned non-zero status %d\n\n", ret);
        exit(1);
    }

    // Build the program
    ret = clBuildProgram(program, 1, &device_id, NULL, NULL, NULL);
    if (ret != 0) {
        printf("clBuildProgram returned non-zero status %d: ", ret);

        if (ret == CL_INVALID_PROGRAM) {
            printf("invalid program\n");
        } else if (ret == CL_INVALID_VALUE) {
            printf("invalid value\n");
        } else if (ret == CL_INVALID_DEVICE) {
            printf("invalid device\n");
        } else if (ret == CL_INVALID_BINARY) {
            printf("invalid binary\n");
        } else if (ret == CL_INVALID_BUILD_OPTIONS) {
            printf("invalid build options\n");
        } else if (ret == CL_INVALID_OPERATION) {
            printf("invalid operation\n");
        } else if (ret == CL_COMPILER_NOT_AVAILABLE) {
            printf("compiler not available\n");
        } else if (ret == CL_BUILD_PROGRAM_FAILURE) {
            printf("build program failure\n");

            // Determine the size of the log
            size_t log_size;
            clGetProgramBuildInfo(program, device_id, CL_PROGRAM_BUILD_LOG, 0, NULL, &log_size);

            // Allocate memory for the log
            char *log = (char *) malloc(log_size);

            // Get the log
            clGetProgramBuildInfo(program, device_id, CL_PROGRAM_BUILD_LOG, log_size, log, NULL);

            // Print the log
            printf("%s\n", log);
        } else if (ret == CL_OUT_OF_HOST_MEMORY) {
            printf("out of host memory\n");
        }
//.........这里部分代码省略.........

int main()
{
    cl_platform_id platform_id = NULL;
    cl_device_id device_id = NULL;
    cl_context context = NULL;
    cl_command_queue command_queue = NULL;
    cl_mem memobj = NULL;
    cl_program program = NULL;
    cl_kernel kernel = NULL;
    cl_uint ret_num_devices;
    cl_uint ret_num_platforms;
    cl_int ret;

    float mem[MEM_SIZE];

    FILE *fp;
    char fileName[] = "./kernel.clbin";
    size_t binary_size;
    char *binary_buf;
    cl_int binary_status;
    cl_int i;

    /* カーネルを含むオブジェクトファイルをロード */
    fp = fopen(fileName, "r");
    if (!fp) {
        fprintf(stderr, "Failed to load kernel.\n");
        exit(1);
    }
    binary_buf = (char *)malloc(MAX_BINARY_SIZE);
    binary_size = fread( binary_buf, 1, MAX_BINARY_SIZE, fp );
    fclose( fp );

    /* データを初期化 */
    for( i = 0; i < MEM_SIZE; i++ ) {
        mem[i] = i;
    }

    /* プラットフォーム・デバイスの情報の取得 */
    ret = clGetPlatformIDs(1, &platform_id, &ret_num_platforms);
    ret = clGetDeviceIDs(platform_id, CL_DEVICE_TYPE_DEFAULT, 1, &device_id, &ret_num_devices);

    /* OpenCLコンテキストの作成 */
    context = clCreateContext( NULL, 1, &device_id, NULL, NULL, &ret);
  
    /* コマンドキューの作成 */
    command_queue = clCreateCommandQueue(context, device_id, 0, &ret);
  
    /* メモリバッファの作成 */
    memobj = clCreateBuffer(context, CL_MEM_READ_WRITE, MEM_SIZE * sizeof(float), NULL, &ret);

    /* メモリバッファにデータを転送 */
    ret = clEnqueueWriteBuffer(command_queue, memobj, CL_TRUE, 0, MEM_SIZE * sizeof(float), mem, 0, NULL, NULL);

    /* 読み込んだバイナリからカーネルプログラムを作成 */
    program = clCreateProgramWithBinary(context, 1, &device_id, (const size_t *)&binary_size, 
                                        (const unsigned char **)&binary_buf, &binary_status, &ret);
    
    /* OpenCLカーネルの作成 */
    kernel = clCreateKernel(program, "vecAdd", &ret);
    printf("err:%d\n", ret);

    /* OpenCLカーネル引数の設定 */
    ret = clSetKernelArg(kernel, 0, sizeof(cl_mem), (void *)&memobj);

    size_t global_work_size[3] = {MEM_SIZE, 0, 0};
    size_t local_work_size[3]  = {MEM_SIZE, 0, 0};

    /* OpenCLカーネルを実行 */
    ret = clEnqueueNDRangeKernel(command_queue, kernel, 1, NULL, global_work_size, local_work_size, 0, NULL, NULL);

    /* メモリバッファから結果を取得 */
    ret = clEnqueueReadBuffer(command_queue, memobj, CL_TRUE, 0, MEM_SIZE * sizeof(float), mem, 0, NULL, NULL);

    /* 結果の表示 */
    for(i=0; i<MEM_SIZE; i++) {
        printf("mem[%d] : %f\n", i, mem[i]);
    }
  
    /* 終了処理 */
    ret = clFlush(command_queue);
    ret = clFinish(command_queue);
    ret = clReleaseKernel(kernel);
    ret = clReleaseProgram(program);
    ret = clReleaseMemObject(memobj);
    ret = clReleaseCommandQueue(command_queue);
    ret = clReleaseContext(context);

    free(binary_buf);

    return 0;
}

int main(void)
{
  hwloc_topology_t topology;
  cl_int clret;
  cl_platform_id *platform_ids;
  unsigned nrp, nrd, count, i, j;
  int err;

  hwloc_topology_init(&topology);
  hwloc_topology_set_flags(topology, HWLOC_TOPOLOGY_FLAG_IO_DEVICES);
  hwloc_topology_load(topology);

  clret = clGetPlatformIDs(0, NULL, &nrp);
  if (CL_SUCCESS != clret || !nrp)
    return 0;
  platform_ids = malloc(nrp * sizeof(*platform_ids));
  if (!platform_ids)
    return 0;
  clret = clGetPlatformIDs(nrp, platform_ids, &nrp);
  if (CL_SUCCESS != clret || !nrp)
    return 0;

  count = 0;
  for(i=0; i<nrp; i++) {
    cl_device_id *device_ids;

    clret = clGetDeviceIDs(platform_ids[i], CL_DEVICE_TYPE_ALL, 0, NULL, &nrd);
    if (CL_SUCCESS != clret || !nrd)
      continue;
    device_ids = malloc(nrd * sizeof(*device_ids));
    if (!device_ids)
      continue;
    clret = clGetDeviceIDs(platform_ids[i], CL_DEVICE_TYPE_ALL, nrd, device_ids, &nrd);
    if (CL_SUCCESS != clret || !nrd)
      continue;

    for(j=0; j<nrd; j++) {
      hwloc_bitmap_t set;
      hwloc_obj_t osdev, osdev2, ancestor;
      const char *value;

      osdev = hwloc_opencl_get_device_osdev(topology, device_ids[j]);
      osdev2 = hwloc_opencl_get_device_osdev_by_index(topology, i, j);
      assert(osdev == osdev2);
      if (!osdev) {
	printf("no osdev for platform %d device %d\n", i, j);
	continue;
      }

      ancestor = hwloc_get_non_io_ancestor_obj(topology, osdev);

      set = hwloc_bitmap_alloc();
      err = hwloc_opencl_get_device_cpuset(topology, device_ids[j], set);
      if (err < 0) {
	printf("no cpuset for platform %d device %d\n", i, j);
      } else {
	char *cpuset_string = NULL;
	hwloc_bitmap_asprintf(&cpuset_string, set);
	printf("got cpuset %s for platform %d device %d\n", cpuset_string, i, j);
	free(cpuset_string);
	assert(hwloc_bitmap_isequal(set, ancestor->cpuset));
      }
      hwloc_bitmap_free(set);

      printf("found OSDev %s\n", osdev->name);
      err = strncmp(osdev->name, "opencl", 6);
      assert(!err);
      assert(atoi(osdev->name+6) == (int) count);

      value = hwloc_obj_get_info_by_name(osdev, "Backend");
      err = strcmp(value, "OpenCL");
      assert(!err);

      value = hwloc_obj_get_info_by_name(osdev, "Name");
      printf("found OSDev name %s\n", value);

      count++;
    }
  }

  hwloc_topology_destroy(topology);

  return 0;
}

int
main(void)
{
    cl_int err;
    cl_platform_id platform = 0;
    cl_device_id device = 0;
    cl_context_properties props[3] = { CL_CONTEXT_PLATFORM, 0, 0 };
    cl_context ctx = 0;
    cl_command_queue queue = 0;
    cl_mem bufX, bufY;
    cl_event event = NULL;
    int ret = 0;
	int lenX = 1 + (N-1)*abs(incx);
	int lenY = 1 + (N-1)*abs(incy);

    /* Setup OpenCL environment. */
    err = clGetPlatformIDs(1, &platform, NULL);

    err = clGetDeviceIDs(platform, CL_DEVICE_TYPE_CPU, 1, &device, NULL);
    if (err != CL_SUCCESS) {
        printf( "clGetPlatformIDs() failed with %d\n", err );
        return 1;
    }

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

    if (err != CL_SUCCESS) {
        printf( "clGetDeviceIDs() failed with %d\n", err );
        return 1;
    }

    props[1] = (cl_context_properties)platform;
    ctx = clCreateContext(props, 1, &device, NULL, NULL, &err);
    if (err != CL_SUCCESS) {
        printf( "clCreateContext() failed with %d\n", err );
        return 1;
    }

    queue = clCreateCommandQueue(ctx, device, 0, &err);
    if (err != CL_SUCCESS) {
        printf( "clCreateCommandQueue() failed with %d\n", err );
        clReleaseContext(ctx);
        return 1;
    }

    /* Setup clblas. */
    err = clblasSetup();
    if (err != CL_SUCCESS) {
        printf("clblasSetup() failed with %d\n", err);
        clReleaseCommandQueue(queue);
        clReleaseContext(ctx);
        return 1;
    }

    /* Prepare OpenCL memory objects and place matrices inside them. */
    bufX = clCreateBuffer(ctx, CL_MEM_READ_WRITE, (lenX*sizeof(cl_float)), NULL, &err);
    bufY = clCreateBuffer(ctx, CL_MEM_READ_WRITE, (lenY*sizeof(cl_float)), NULL, &err);

    err = clEnqueueWriteBuffer(queue, bufX, CL_TRUE, 0, (lenX*sizeof(cl_float)), X, 0, NULL, NULL);
    err = clEnqueueWriteBuffer(queue, bufY, CL_TRUE, 0, (lenY*sizeof(cl_float)), Y, 0, NULL, NULL);

	printResult();

    /* Call clblas function. */
    err = clblasSrot(N, bufX, 0, incx, bufY, 0, incy, C, S, 1, &queue, 0, NULL, &event);
//	printf("here\n");
    if (err != CL_SUCCESS) {
        printf("clblasSrot() failed with %d\n", err);
        ret = 1;
    }
    else {
        /* Wait for calculations to be finished. */
        err = clWaitForEvents(1, &event);

        /* Fetch results of calculations from GPU memory. */
        err = clEnqueueReadBuffer(queue, bufY, CL_TRUE, 0, (lenY*sizeof(cl_float)),
                                    Y, 0, NULL, NULL);
        err = clEnqueueReadBuffer(queue, bufX, CL_TRUE, 0, (lenX*sizeof(cl_float)),
                                    X, 0, NULL, NULL);

        /* At this point you will get the result of SROT placed in vector Y. */
        printResult();
    }

    /* Release OpenCL events. */
    clReleaseEvent(event);

    /* Release OpenCL memory objects. */
    clReleaseMemObject(bufY);
    clReleaseMemObject(bufX);

    /* Finalize work with clblas. */
    clblasTeardown();

    /* Release OpenCL working objects. */
    clReleaseCommandQueue(queue);
    clReleaseContext(ctx);

    return ret;
}

int main(int argc, char** argv)
{
    int err;                            // error code returned from api calls
    
    float data[DATA_SIZE];              // original data set given to device
    float results[DATA_SIZE];           // results returned from device
    unsigned int correct;               // number of correct results returned
    
    size_t global;                      // global domain size for our calculation
    size_t local;                       // local domain size for our calculation
    
    cl_device_id device_id;             // compute device id
    cl_context context;                 // compute context
    cl_command_queue commands;          // compute command queue
    cl_program program;                 // compute program
    cl_kernel kernel;                   // compute kernel
    
    cl_mem input;                       // device memory used for the input array
    cl_mem output;                      // device memory used for the output array
    
    // Fill our data set with random float values
    //
    int i = 0;
    unsigned int count = DATA_SIZE;
    for(i = 0; i < count; i++)
        data[i] = rand() / (float)RAND_MAX;
    
    // Connect to a compute device
    //
    int gpu = 1;
    err = clGetDeviceIDs(NULL, gpu ? CL_DEVICE_TYPE_GPU : CL_DEVICE_TYPE_CPU, 1, &device_id, NULL);
    if (err != CL_SUCCESS)
    {
        printf("Error: Failed to create a device group!\n");
        return EXIT_FAILURE;
    }
    
    // Create a compute context
    //
    context = clCreateContext(0, 1, &device_id, NULL, NULL, &err);
    if (!context)
    {
        printf("Error: Failed to create a compute context!\n");
        return EXIT_FAILURE;
    }
    
    // Create a command commands
    //
    commands = clCreateCommandQueue(context, device_id, 0, &err);
    if (!commands)
    {
        printf("Error: Failed to create a command commands!\n");
        return EXIT_FAILURE;
    }
    
    // Create the compute program from the source buffer
    //
    program = clCreateProgramWithSource(context, 1, (const char **) & KernelSource, NULL, &err);
    if (!program)
    {
        printf("Error: Failed to create compute program!\n");
        return EXIT_FAILURE;
    }
    
    // Build the program executable
    //
    err = clBuildProgram(program, 0, NULL, NULL, NULL, NULL);
    if (err != CL_SUCCESS)
    {
        size_t len;
        char buffer[2048];
        
        printf("Error: Failed to build program executable!\n");
        clGetProgramBuildInfo(program, device_id, CL_PROGRAM_BUILD_LOG, sizeof(buffer), buffer, &len);
        printf("%s\n", buffer);
        exit(1);
    }
    
    // Create the compute kernel in the program we wish to run
    //
    kernel = clCreateKernel(program, "square", &err);
    if (!kernel || err != CL_SUCCESS)
    {
        printf("Error: Failed to create compute kernel!\n");
        exit(1);
    }
    
    // Create the input and output arrays in device memory for our calculation
    //
    input = clCreateBuffer(context,  CL_MEM_READ_ONLY,  sizeof(float) * count, NULL, NULL);
    output = clCreateBuffer(context, CL_MEM_WRITE_ONLY, sizeof(float) * count, NULL, NULL);
    if (!input || !output)
    {
        printf("Error: Failed to allocate device memory!\n");
        exit(1);
    }
    
    // Write our data set into the input array in device memory
    //
    err = clEnqueueWriteBuffer(commands, input, CL_TRUE, 0, sizeof(float) * count, data, 0, NULL, NULL);
//.........这里部分代码省略.........

int main() {

   // Set the image rotation (in degrees)
   float theta = 3.14159/6;
   float cos_theta = cosf(theta);
   float sin_theta = sinf(theta);
   printf("theta = %f (cos theta = %f, sin theta = %f)\n", theta, cos_theta, 
      sin_theta);

   // Rows and columns in the input image
   int imageHeight;
   int imageWidth;

   const char* inputFile = "input.bmp";
   const char* outputFile = "output.bmp";

   // Homegrown function to read a BMP from file
   float* inputImage = readImage(inputFile, &imageWidth,
      &imageHeight);

   // Size of the input and output images on the host
   int dataSize = imageHeight*imageWidth*sizeof(float);

   // Output image on the host
   float* outputImage = NULL;
   outputImage = (float*)malloc(dataSize);

   // Set up the OpenCL environment
   cl_int status;

   // Discovery platform
   cl_platform_id platforms[2];
   cl_platform_id platform;
   status = clGetPlatformIDs(2, platforms, NULL);
   chk(status, "clGetPlatformIDs");
   platform = platforms[PLATFORM_TO_USE];

   // Discover device
   cl_device_id device;
   clGetDeviceIDs(platform, CL_DEVICE_TYPE_CPU, 1, &device, NULL);
   chk(status, "clGetDeviceIDs");

   // Create context
   cl_context_properties props[3] = {CL_CONTEXT_PLATFORM,
       (cl_context_properties)(platform), 0};
   cl_context context;
   context = clCreateContext(props, 1, &device, NULL, NULL, &status);
   chk(status, "clCreateContext");

   // Create command queue
   cl_command_queue queue;
   queue = clCreateCommandQueue(context, device, 0, &status);
   chk(status, "clCreateCommandQueue");

   // Create the input and output buffers
   cl_mem d_input;
   d_input = clCreateBuffer(context, CL_MEM_READ_ONLY, dataSize, NULL,
       &status);
   chk(status, "clCreateBuffer");

   cl_mem d_output;
   d_output = clCreateBuffer(context, CL_MEM_WRITE_ONLY, dataSize, NULL,
       &status);
   chk(status, "clCreateBuffer");

   // Copy the input image to the device
   status = clEnqueueWriteBuffer(queue, d_input, CL_TRUE, 0, dataSize, 
         inputImage, 0, NULL, NULL);
   chk(status, "clEnqueueWriteBuffer");

   const char* source = readSource("rotation.cl");

   // Create a program object with source and build it
   cl_program program;
   program = clCreateProgramWithSource(context, 1, &source, NULL, NULL);
   chk(status, "clCreateProgramWithSource");
   status = clBuildProgram(program, 1, &device, NULL, NULL, NULL);
   chk(status, "clBuildProgram");
   
   // Create the kernel object
   cl_kernel kernel;
   kernel = clCreateKernel(program, "img_rotate", &status);
   chk(status, "clCreateKernel");

   // Set the kernel arguments
   status  = clSetKernelArg(kernel, 0, sizeof(cl_mem), &d_output);
   status |= clSetKernelArg(kernel, 1, sizeof(cl_mem), &d_input);
   status |= clSetKernelArg(kernel, 2, sizeof(int), &imageWidth);
   status |= clSetKernelArg(kernel, 3, sizeof(int), &imageHeight);
   status |= clSetKernelArg(kernel, 4, sizeof(float), &sin_theta);
   status |= clSetKernelArg(kernel, 5, sizeof(float), &cos_theta);
   chk(status, "clSetKernelArg");

   // Set the work item dimensions
   size_t globalSize[2] = {imageWidth, imageHeight};
   status = clEnqueueNDRangeKernel(queue, kernel, 2, NULL, globalSize, NULL, 0,
      NULL, NULL);
   chk(status, "clEnqueueNDRange");

   // Read the image back to the host
//.........这里部分代码省略.........

cl_context CreateContext()
{
	cl_int errNum;
	cl_uint numPlatforms;
	cl_platform_id * platformIDs;
	cl_uint numDevices;
	cl_device_id * deviceIDs;
	cl_context context = NULL;

	// First, select an OpenCL platform to run on. 
	errNum = clGetPlatformIDs(0, NULL, &numPlatforms);
	platformIDs = (cl_platform_id *)alloca(
		sizeof(cl_platform_id) * numPlatforms);
	errNum = clGetPlatformIDs(numPlatforms, platformIDs, NULL);
	if (errNum != CL_SUCCESS || numPlatforms <= 0)
	{
		std::cerr << "Failed to find any OpenCL platforms." << std::endl;
		return NULL;
	}
	else std::cout<<"number of platforms:"<<numPlatforms<<std::endl;

	deviceIDs = NULL;
	errNum = clGetDeviceIDs(
		platformIDs[0], 
		CL_DEVICE_TYPE_GPU, //寻找GPU device 的个数
		0,
		NULL,
		&numDevices);
	if (errNum != CL_SUCCESS && errNum != CL_DEVICE_NOT_FOUND)
	{
		checkErr(errNum, "clGetDeviceIDs");
	}
	else std::cout<<"number of devices:"<<numDevices<<std::endl;
	deviceIDs = (cl_device_id *)alloca(sizeof(cl_device_id) * numDevices);
	errNum = clGetDeviceIDs(
		platformIDs[0],
		CL_DEVICE_TYPE_GPU,
		numDevices, 
		&deviceIDs[0], 
		NULL);
	checkErr(errNum, "clGetDeviceIDs");
	// Next, create an OpenCL context on the platform.  Attempt to
	// create a GPU-based context, and if that fails, try to create
	// a CPU-based context.
	cl_context_properties contextProperties[] =
	{
		CL_CONTEXT_PLATFORM,
		(cl_context_properties)platformIDs[0],
		0
	};
	/*context = clCreateContext(
		contextProperties, 
		numDevices,
		deviceIDs, 
		NULL,
		NULL, 
		&errNum);
	checkErr(errNum, "clCreateContext");*/
	//如果有平台和设备类型，可以使用clCreateContextFromType()创建上下文。上下文较随意，可以包含不同设备类型
	context = clCreateContextFromType(contextProperties, CL_DEVICE_TYPE_GPU,
		NULL, NULL, &errNum);
	if (errNum != CL_SUCCESS)
	{
		std::cout << "Could not create GPU context, trying CPU..." << std::endl;
		context = clCreateContextFromType(contextProperties, CL_DEVICE_TYPE_CPU,
			NULL, NULL, &errNum);
		if (errNum != CL_SUCCESS)
		{
			std::cerr << "Failed to create an OpenCL GPU or CPU context." << std::endl;
			return NULL;
		}
	}

	return context;
}

	OpenCLDevice(DeviceInfo& info, Stats &stats, bool background_)
	  : Device(stats)
	{
		background = background_;
		cpPlatform = NULL;
		cxContext = NULL;
		cqCommandQueue = NULL;
		cpProgram = NULL;
		ckPathTraceKernel = NULL;
		ckFilmConvertKernel = NULL;
		null_mem = 0;
		device_initialized = false;

		/* setup platform */
		cl_uint num_platforms;

		ciErr = clGetPlatformIDs(0, NULL, &num_platforms);
		if(opencl_error(ciErr))
			return;

		if(num_platforms == 0) {
			opencl_error("OpenCL: no platforms found.");
			return;
		}

		ciErr = clGetPlatformIDs(1, &cpPlatform, NULL);
		if(opencl_error(ciErr))
			return;

		char name[256];
		clGetPlatformInfo(cpPlatform, CL_PLATFORM_NAME, sizeof(name), &name, NULL);
		platform_name = name;

		/* get devices */
		vector<cl_device_id> device_ids;
		cl_uint num_devices;

		if(opencl_error(clGetDeviceIDs(cpPlatform, opencl_device_type(), 0, NULL, &num_devices)))
			return;

		if(info.num > num_devices) {
			if(num_devices == 0)
				opencl_error("OpenCL: no devices found.");
			else
				opencl_error("OpenCL: specified device not found.");
			return;
		}

		device_ids.resize(num_devices);
		
		if(opencl_error(clGetDeviceIDs(cpPlatform, opencl_device_type(), num_devices, &device_ids[0], NULL)))
			return;

		cdDevice = device_ids[info.num];

		/* create context */
		cxContext = clCreateContext(0, 1, &cdDevice, NULL, NULL, &ciErr);
		if(opencl_error(ciErr))
			return;

		cqCommandQueue = clCreateCommandQueue(cxContext, cdDevice, 0, &ciErr);
		if(opencl_error(ciErr))
			return;

		null_mem = (device_ptr)clCreateBuffer(cxContext, CL_MEM_READ_ONLY, 1, NULL, &ciErr);
		device_initialized = true;
	}

int main(int argc, const char** argv) {
  // start logs
  printf("clDeviceQuery Starting...\n\n");
  bool bPassed = true;
  std::string sProfileString = "clDeviceQuery, Platform Name = ";

  // Get OpenCL platform ID for NVIDIA if avaiable, otherwise default
  char cBuffer[1024];
  cl_platform_id clSelectedPlatformID = NULL;
  cl_platform_id* clPlatformIDs;

  cl_uint num_platforms;
  cl_int ciErrNum = clGetPlatformIDs(0, NULL, &num_platforms);
  if (ciErrNum != CL_SUCCESS) {
    printf(" Error %i in clGetPlatformIDs Call!\n\n", ciErrNum);
    bPassed = false;
  } else {
    if (num_platforms == 0) {
      printf("No OpenCL platform found!\n\n");
      bPassed = false;
    } else {
      // if there's one platform or more, make space for ID's
      if ((clPlatformIDs = (cl_platform_id*)malloc(num_platforms * sizeof(cl_platform_id))) == NULL) {
	printf("Failed to allocate memory for cl_platform ID's!\n\n");
	bPassed = false;
      }

      printf("%d OpenCL Platforms found\n\n", num_platforms);
      // get platform info for each platform
      ciErrNum = clGetPlatformIDs (num_platforms, clPlatformIDs, NULL);
      for(cl_uint i = 0; i < num_platforms; ++i) {
	ciErrNum = clGetPlatformInfo (clPlatformIDs[i], CL_PLATFORM_NAME, 1024, &cBuffer, NULL);
	if(ciErrNum == CL_SUCCESS) {
	  clSelectedPlatformID = clPlatformIDs[i];
	  // Get OpenCL platform name and version
	  ciErrNum = clGetPlatformInfo (clSelectedPlatformID, CL_PLATFORM_NAME, sizeof(cBuffer), cBuffer, NULL);
	  if (ciErrNum == CL_SUCCESS) {
	    printf(" CL_PLATFORM_NAME: \t%s\n", cBuffer);
	    sProfileString += cBuffer;
	  } else {
	    printf(" Error %i in clGetPlatformInfo Call !!!\n\n", ciErrNum);
    bPassed = false;
  }
  sProfileString += ", Platform Version = ";

  ciErrNum = clGetPlatformInfo (clSelectedPlatformID, CL_PLATFORM_VERSION, sizeof(cBuffer), cBuffer, NULL);
  if (ciErrNum == CL_SUCCESS) {
    printf(" CL_PLATFORM_VERSION: \t%s\n", cBuffer);
    sProfileString += cBuffer;
  } else {
    printf(" Error %i in clGetPlatformInfo Call !!!\n\n", ciErrNum);
    bPassed = false;
  }

  // Log OpenCL SDK Version # (for convenience:  not specific to OpenCL)
  sProfileString += ", NumDevs = ";

  // Get and log OpenCL device info
  cl_uint ciDeviceCount;
  cl_device_id *devices;
  printf("OpenCL Device Info:\n\n");
  ciErrNum = clGetDeviceIDs (clSelectedPlatformID, CL_DEVICE_TYPE_ALL, 0, NULL, &ciDeviceCount);

  // check for 0 devices found or errors...
  if (ciDeviceCount == 0) {
    printf(" No devices found supporting OpenCL (return code %i)\n\n", ciErrNum);
    bPassed = false;
    sProfileString += "0";
  } else if (ciErrNum != CL_SUCCESS) {
    printf(" Error %i in clGetDeviceIDs call !!!\n\n", ciErrNum);
    bPassed = false;
  } else {
    // Get and log the OpenCL device ID's
    ciErrNum = clGetPlatformInfo (clSelectedPlatformID, CL_PLATFORM_NAME, sizeof(cBuffer), cBuffer, NULL);
    printf(" %u devices found supporting OpenCL on: %s\n\n", ciDeviceCount, cBuffer);
    char cTemp[2];
    sprintf(cTemp, "%u", ciDeviceCount);
    sProfileString += cTemp;
    if ((devices = (cl_device_id*)malloc(sizeof(cl_device_id) * ciDeviceCount)) == NULL) {
      printf(" Failed to allocate memory for devices !!!\n\n");
      bPassed = false;
    }
    ciErrNum = clGetDeviceIDs (clSelectedPlatformID, CL_DEVICE_TYPE_ALL, ciDeviceCount, devices, &ciDeviceCount);
    if (ciErrNum == CL_SUCCESS) {
      for(unsigned int i = 0; i < ciDeviceCount; ++i )  {
        printf(" ----------------------------------\n");
clGetDeviceInfo(devices[i], CL_DEVICE_NAME, sizeof(cBuffer), &cBuffer, NULL);
printf(" Device %s\n", cBuffer);
printf(" ---------------------------------\n");
clPrintDevInfo(devices[i]);
sProfileString += ", Device = ";
sProfileString += cBuffer;
      }
            } else {
      printf(" Error %i in clGetDeviceIDs call !!!\n\n", ciErrNum);
      bPassed = false;
    }
  }

  // masterlog info
//.........这里部分代码省略.........

int main( int argc, char* argv[] )
{
    // Length of vectors
    unsigned int n = 100000;
 
    // Host input vectors
    double *h_a;
    double *h_b;
    // Host output vector
    double *h_c;
 
    // Device input buffers
    cl_mem d_a;
    cl_mem d_b;
    // Device output buffer
    cl_mem d_c;
 
    cl_platform_id cpPlatform;        // OpenCL platform
    cl_device_id device_id;           // device ID
    cl_context context;               // context
    cl_command_queue queue;           // command queue
    cl_program program;               // program
    cl_kernel kernel;                 // kernel
 
    // Size, in bytes, of each vector
    size_t bytes = n*sizeof(double);
 
    // Allocate memory for each vector on host
    h_a = (double*)malloc(bytes);
    h_b = (double*)malloc(bytes);
    h_c = (double*)malloc(bytes);
 
    // Initialize vectors on host
    int i;
    for( i = 0; i < n; i++ )
    {
        h_a[i] = sinf(i)*sinf(i);
        h_b[i] = cosf(i)*cosf(i);
    }
 
    size_t globalSize, localSize;
    cl_int err;
 
    // Number of work items in each local work group
    localSize = 64;
 
    // Number of total work items - localSize must be devisor
    globalSize = ceil(n/(float)localSize)*localSize;
 
    // Bind to platform
    err = clGetPlatformIDs(1, &cpPlatform, NULL);
 
    // Get ID for the device
    err = clGetDeviceIDs(cpPlatform, CL_DEVICE_TYPE_GPU, 1, &device_id, NULL);
 
    // Create a context 
    context = clCreateContext(0, 1, &device_id, NULL, NULL, &err);
 
    // Create a command queue
    queue = clCreateCommandQueue(context, device_id, 0, &err);
 
    // Create the compute program from the source buffer
    program = clCreateProgramWithSource(context, 1,
                            (const char **) & kernelSource, NULL, &err);
 
    // Build the program executable
    clBuildProgram(program, 0, NULL, NULL, NULL, NULL);
 
    // Create the compute kernel in the program we wish to run
    kernel = clCreateKernel(program, "vecAdd", &err);
 
    // Create the input and output arrays in device memory for our calculation
    d_a = clCreateBuffer(context, CL_MEM_READ_ONLY, bytes, NULL, NULL);
    d_b = clCreateBuffer(context, CL_MEM_READ_ONLY, bytes, NULL, NULL);
    d_c = clCreateBuffer(context, CL_MEM_WRITE_ONLY, bytes, NULL, NULL);
 
    // Write our data set into the input array in device memory
    err = clEnqueueWriteBuffer(queue, d_a, CL_TRUE, 0,
                                   bytes, h_a, 0, NULL, NULL);
    err |= clEnqueueWriteBuffer(queue, d_b, CL_TRUE, 0,
                                   bytes, h_b, 0, NULL, NULL);
 
    // Set the arguments to our compute kernel
    err  = clSetKernelArg(kernel, 0, sizeof(cl_mem), &d_a);
    err |= clSetKernelArg(kernel, 1, sizeof(cl_mem), &d_b);
    err |= clSetKernelArg(kernel, 2, sizeof(cl_mem), &d_c);
    err |= clSetKernelArg(kernel, 3, sizeof(unsigned int), &n);
 
    // Execute the kernel over the entire range of the data set 
    err = clEnqueueNDRangeKernel(queue, kernel, 1, NULL, &globalSize, &localSize,
                                                              0, NULL, NULL);
 
    // Wait for the command queue to get serviced before reading back results
    clFinish(queue);
 
    // Read the results from the device
    clEnqueueReadBuffer(queue, d_c, CL_TRUE, 0,
                                bytes, h_c, 0, NULL, NULL );
 
    //Sum up vector c and print result divided by n, this should equal 1 within error
//.........这里部分代码省略.........