cuda_opencv 矩阵相加

实现矩阵相加

#include <stdlib.h>
#include <stdio.h>
#include <opencv/cv.h>
#include <opencv/highgui.h>
#include <opencv2/opencv.hpp>

#include "cuda_runtime.h"
#include "device_launch_parameters.h"
using namespace std;
using namespace cv;

__global__ void Add_kernel(const int2* d_A, const int2* d_B,int2*  d_C,int width, int height)
{
    int x = threadIdx.x + blockIdx.x * blockDim.x;
    int y = threadIdx.y + blockIdx.y * blockDim.y;

    if(x < width && y < height)
    {
        int offset = x + y*width;
        d_C[offset].x = d_A[offset].x + d_B[offset].x;
        d_C[offset].y = d_A[offset].y + d_B[offset].y;
    }
}
int main()
{
    Mat img(3, 4, CV_32S, Scalar_<int>(0));

    cout<<img<<endl;
    cout<<endl;


    for(int i = 0 ; i < img.rows; i++)
    {
        for(int j = 0 ; j < img.cols; j++)
        {
            img.at<int>(i,j)=i+j;
        }
    }
    cout<<endl;

    cout<<img<<endl;


    size_t memSize = img.step * img.rows;
    int2* d_A = NULL;
    int2* d_B = NULL;
    int2* d_C = NULL;
    cudaMalloc((void**)&d_A, memSize);
    cudaMalloc((void**)&d_B, memSize);
    cudaMalloc((void**)&d_C, memSize);

    cudaMemcpy(d_A,img.data,memSize, cudaMemcpyHostToDevice);
    cudaMemcpy(d_B,img.data,memSize, cudaMemcpyHostToDevice);

    dim3 threads(16, 16);
    dim3 grids((img.rows + threads.x - 1)/threads.x,(img.cols + threads.y - 1)/threads.y);
    Add_kernel<<<grids,threads>>>(d_A, d_B, d_C, img.rows, img.cols);

    cudaMemcpy(img.data, d_C,memSize,cudaMemcpyDeviceToHost);
    cout<<"GPU"<<endl;
    cout<<img<<endl;
    cudaFree(d_A);
    cudaFree(d_B);
    cudaFree(d_C);

    system("pause");
    return 0;
}