CUDA编程整理

#include <stdio.h>
#include <stdlib.h>
#include <cuda_runtime_api.h>
#include <iostream>

/* 核函数 */
__global__ void kernelFunc(float *a) {
    a[threadIdx.x] = 1;
}

int main(int argc, char **argv) {
    //设置显卡设备
    cudaSetDevice(0);
    //分配显存空间
    float *aGpu;
    cudaMalloc((void**)&aGpu, 16 * sizeof(float));
    //从内存到显存拷贝数据
    float a[16] = {0};
    cudaMemcpy(aGpu, a, 16 * sizeof(float), cudaMemcpyHostToDevice);
    //执行CUDA并行函数
    kernelFunc <<<1, 16>> >(aGpu);
    //CUDA函数结束后，将结果从显存拷贝回内存
    cudaMemcpy(a, aGpu, 16 * sizeof(float), cudaMemcpyDeviceToHost);
    for (int i = 0; i < 16; i++) {
        printf("%f", a[i]);
    }
    return 0;
}

nvcc main.cu -o main

1.0000001.0000001.0000001.0000001.0000001.0000001.0000001.0000001.0000001.0000001.0000001.0000001.0000001.0000001.0000001.000000