CUDA 一维卷积实现

简单实现了在CUDA中的一维卷积

//一维卷积实现
__global__ void convolution_1D_basic_kernel(int *N, int *M, int *P,
	int Mask_Width, int Width){
	int i = blockIdx.x*blockDim.x + threadIdx.x;

	float Pvalue = 0;
	int N_start_point = i - (Mask_Width / 2);
	for (int j = 0; j < Mask_Width; j++){
		if(N_start_point + j >= 0 && N_start_point + j < Width){
			Pvalue += N[N_start_point + j] * M[j];
		}
	}
	P[i] = Pvalue;
}

int main()
{
	const int M[5] = { 3, 4, 5, 4, 3 };
	const int N[7] = { 1, 2, 3, 4, 5, 6, 7 };
	int Mask_Width = 5;
	int Width = 7;
	int P[7] = { 0 };


	int *dev_M = 0;
	int *dev_N = 0;
	int *dev_P = 0;


	//申请内存
	cudaMalloc((void**)&dev_M, Mask_Width * sizeof(int));
	cudaMalloc((void**)&dev_N, Width * sizeof(int));
	cudaMalloc((void**)&dev_P, Width * sizeof(int));

	cudaMemcpy(dev_M, M, Mask_Width * sizeof(int), cudaMemcpyHostToDevice);
	cudaMemcpy(dev_N, N, Width * sizeof(int), cudaMemcpyHostToDevice);

	convolution_1D_basic_kernel <<<1, 7 >>>(dev_N, dev_M, dev_P, Mask_Width,Width);

	cudaMemcpy(P, dev_P, 7 * sizeof(int), cudaMemcpyDeviceToHost);
	for (int i = 0; i < 7; i++)
	{
		cout << P[i] << " ";
		if (i == 6)
		{
			cout << endl;
		}
	}

	cudaFree(dev_M);
	cudaFree(dev_N);
	cudaFree(dev_P);

    return 0;

}