这个程序是一个简单的并行程序，它添加了2个向量的元素。 该程序没有错误，并且编译成功，但结果不正确
#include <stdio.h>
#include <stdlib.h>
#include <iostream>
#include <iomanip>
#include <array>
#include <fstream>
#include <sstream>
#include <string>
#include <algorithm>
#include <iterator>
#ifdef __APPLE__
#include <OpenCL/opencl.h>
#else
#include <CL/cl.h>
#include <time.h>
#endif
#define MAX_SOURCE_SIZE (0x100000)
// number of points in Both A and B files (number of rows)
const int number_of_points = 11;
// number of points axis in Both A and B files (number of Columns)
const int number_of_axis = 3;
using namespace std;
int main(int argc, char *argv[]) {
clock_t tStart = clock();
// Create the two input vectors
// working variables
int i;
ifstream input_fileA, input_fileB; // input files
string line; // transfer row from file to array
float x; // transfer word from file to array
int row = 0; // number of rows of file A,B (= array)
int col = 0; // number of rows of file A,B (= array)
// working arrays
// array contains file A data
float arrayA[number_of_points][number_of_axis]={{0}};
// array contains file B data
float arrayB[number_of_points][number_of_axis]={{0}};
// float X1[number_of_points]; // X values of file A points
float Y1[number_of_points]; // Y values of file A points
// float X2[number_of_points]; // X values of file B points
float Y2[number_of_points]; // Y values of file B points
float *X1 = (float*)malloc(sizeof(float)*number_of_points);
float *X2 = (float*)malloc(sizeof(float)*number_of_points);
// import input files
input_fileA.open(argv[1]);
input_fileB.open(argv[2]);
// transfer input files data to array
// input file A to arrayA
row = 0;
while (getline(input_fileA, line))
{
istringstream streamA(line);
col = 0;
while(streamA >> x){
arrayA[row][col] = x;
col++;
}
row++;
}
// input file B to arrayB
row = 0;
while (getline(input_fileB, line))
{
istringstream streamB(line);
col = 0;
while(streamB >> x){
arrayB[row][col] = x;
col++;
}
row++;
}
// put Xs of points in X vectors and Ys of points in Y vectors
// input file A
for (int i = 0; i<number_of_points; i++){
X1[i] = arrayA[i][1];
Y1[i] = arrayA[i][2];
}
// input file B
for (int i = 0; i<number_of_points; i++){
X2[i] = arrayB[i][1];
Y2[i] = arrayB[i][2];
}
// int i;
// const int LIST_SIZE = 50;
// int *A = (int*)malloc(sizeof(int)*number_of_points);
// int *B = (int*)malloc(sizeof(int)*number_of_points);
// for(i = 0; i < number_of_points; i++) {
// A[i] = X1[i];
// B[i] = X2[i];
// }
// Load the kernel source code into the array source_str
FILE *fp;
char *source_str;
size_t source_size;
fp = fopen("vector_add_kernel.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_ALL, 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 x1_mem_obj = clCreateBuffer(context, CL_MEM_READ_ONLY,
number_of_points * sizeof(float), NULL, &ret);
cl_mem x2_mem_obj = clCreateBuffer(context, CL_MEM_READ_ONLY,
number_of_points * sizeof(float), NULL, &ret);
cl_mem c_mem_obj = clCreateBuffer(context, CL_MEM_WRITE_ONLY,
number_of_points * sizeof(float), NULL, &ret);
// Copy the lists A and B to their respective memory buffers
ret = clEnqueueWriteBuffer(command_queue, x1_mem_obj, CL_TRUE, 0,
number_of_points * sizeof(float), X1, 0, NULL, NULL);
ret = clEnqueueWriteBuffer(command_queue, x2_mem_obj, CL_TRUE, 0,
number_of_points * sizeof(float), X2, 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);
// Build the program
ret = clBuildProgram(program, 1, &device_id, NULL, NULL, NULL);
// Create the OpenCL kernel
cl_kernel kernel = clCreateKernel(program, "vector_add", &ret);
// Set the arguments of the kernel
ret = clSetKernelArg(kernel, 0, sizeof(cl_mem), (void *)&x1_mem_obj);
ret = clSetKernelArg(kernel, 1, sizeof(cl_mem), (void *)&x2_mem_obj);
ret = clSetKernelArg(kernel, 2, sizeof(cl_mem), (void *)&c_mem_obj);
// Execute the OpenCL kernel on the list
size_t global_item_size = number_of_points; // Process the entire lists
size_t local_item_size = 64; // Process in groups of 64
ret = clEnqueueNDRangeKernel(command_queue, kernel, 1, NULL,
&global_item_size, &local_item_size, 0, NULL, NULL);
// Read the memory buffer C on the device to the local variable C
// int *C = (int*)malloc(sizeof(int)*number_of_points);
float *C = (float*)malloc(sizeof(float)*number_of_points);
ret = clEnqueueReadBuffer(command_queue, c_mem_obj, CL_TRUE, 0,
number_of_points * sizeof(float), C, 0, NULL, NULL);
// Display the result to the screen
for(i = 0; i < number_of_points; i++)
printf("%f + %f = %f\n", X1[i], X2[i], C[i]);
// Clean up
ret = clFlush(command_queue);
ret = clFinish(command_queue);
ret = clReleaseKernel(kernel);
ret = clReleaseProgram(program);
ret = clReleaseMemObject(x1_mem_obj);
ret = clReleaseMemObject(x2_mem_obj);
ret = clReleaseMemObject(c_mem_obj);
ret = clReleaseCommandQueue(command_queue);
ret = clReleaseContext(context);
free(X1);
free(X2);
free(C);
printf("ALL Time taken: %.2fs\n", (double)(clock() - tStart)/CLOCKS_PER_SEC);
return 0;
}
和内核文件
__kernel void vector_add(__global float *X1,
__global float *X2,
__global float *C) {
// Get the index of the current element
int i = get_global_id(0);
// Do the operation
C[i] = X1[i] + X2[i];
}
结果是
0.000000 + 0.000000 = 0.000000
1.000000 + 1.000000 = 0.000000
2.000000 + 2.000000 = 0.000000
3.000000 + 3.000000 = 0.000000
4.000000 + 4.000000 = 0.000000
5.000000 + 5.000000 = 0.000000
6.000000 + 6.000000 = 0.000000
7.000000 + 7.000000 = 0.000000
8.000000 + 8.000000 = 0.000000
9.000000 + 9.000000 = 0.000000
1.000000 + 1.000000 = 0.000000
ALL Time taken: 0.07s
答案 0 :(得分：1)
您已经承诺了OpenCL编程的一个主要罪行，因为您没有检查任何OpenCL API调用的错误代码！您应始终检查每个 OpenCL API调用的返回码。如果你这样做，它会很快指出你的问题。
问题出在你的内核入队调用中。如果您检查错误代码，则会看到您返回-54，这与CL_INVALID_WORK_GROUP_SIZE相对应。具体而言，内核调用要求工作组大小（本地大小）完全划分全局大小。您要求的工作组大小为64，全局大小为11，这不符合此要求。
您还可以将NULL作为工作组大小参数传递，OpenCL实现将选择一个肯定适合您的工作组大小。