#include "cuda_runtime.h"
#include "device_launch_parameters.h"#include <iomanip>
#include <iostream>
#include <stdio.h>using namespace std;//检测GPU
bool CheckCUDA(void){int count = 0;int i = 0;cudaGetDeviceCount(&count);if (count == 0) {printf("找不到支持CUDA的设备!\n");return false;}cudaDeviceProp prop;for (i = 0; i < count; i++) {if (cudaGetDeviceProperties(&prop, i) == cudaSuccess) {if (prop.major >= 1) {break;}}}if (i == count) {printf("找不到支持CUDA的设备!\n");return false;}cudaGetDeviceProperties(&prop, 0);printf("GPU is: %s\n", prop.name);cudaSetDevice(0);printf("CUDA initialized success.\n");return true;
}//使用指针相加
__global__ void addNumber(double a, double b, double *c);int main(){//检测GPUif (!CheckCUDA()){cout << "No CUDA device.";return 0;}cout << "*************************************************************************************************************" << endl;double h_a, h_b, h_c;       //在CPU上定义三个变量double *d_c;           //定义一个将指向GPU的指针
h_a = 2.2;h_b = 3.3;cudaMalloc((void **)&d_c, sizeof(double));     //为指针在GPU上分配内存空间　　 //调用核函数并启用一个线程块和一个线程addNumber<<<1, 1>>>(h_a, h_b, d_c);            //只是单独的两个数相加，不是两个数组相加，只需使用单线程，数组相加可以使用多线程cudaMemcpy(&h_c, d_c, sizeof(double), cudaMemcpyDeviceToHost);    //将GPU上计算好的结果返回到CPU上定义好的变量//setw(10)表示输出10个空格，需添加 #include <iomanip>cout << setw(10) << h_a << " + " << h_b << " = " << h_c << endl;  cout << endl << endl;system("pause");return 0;
}__global__ void addNumber(double a, double b, double *c){*c = a + b;
}