CUDA矩阵加法出错

[sbaban]

#include<cuda_runtime_api.h>
#include<stdio.h>
__global__ void add(int a[2][2], int b[2][2], int c[2][2])
{
	int i = blockIdx.x*blockDim.x + threadIdx.x;
	int j = blockIdx.y*blockDim.y + threadIdx.y;
	if (i < 2 && j < 2)
	{	
			c[i][j] = a[i][j] + b[i][j];
	}
}
int main()
{
	int i, j, k;
	int a[2][2] = { 1, 2, 3, 4 };
	int b[2][2] = { 10, 20, 30, 40 };
	int c[2][2] = { 0 };

	cudaError_t error = cudaSuccess;

	int device_a[2][2], device_b[2][2], device_c[2][2];
	error = cudaMalloc((void **)&device_a, sizeof(int)* 4);
	error = cudaMalloc((void **)&device_b, sizeof(int)* 4);
	error = cudaMalloc((void **)&device_c, sizeof(int)*4);

	cudaMemcpy(device_a,a, sizeof(int)* 4, cudaMemcpyHostToDevice);
	cudaMemcpy(device_b,b, sizeof(int)* 4, cudaMemcpyHostToDevice);

//	dim3 threadsPerBlock(1, 1);
//	dim3 numBlocks(2 / threadsPerBlock.x, 2 / threadsPerBlock.y);

	add << <1, 4 >> >(device_a, device_b, device_c);

	cudaMemcpy(c, device_c, sizeof(int)* 4, cudaMemcpyDeviceToHost);

	for (i = 0; i < 2; i++){printf("\n");for (j = 0; j < 2; j++)printf("%4d", c[i][j]);}

	return 0;
}

请问为什么结果显示是0 ，哪里出问题了

[sbaban]

怎么改，能具体说下吗

[fgg1991]

读了一下 代码感觉没啥问题 你把参数改成int*在host里给他们分配空间再试？

[gundamzzj]

把 global函数 改成 __global__ void add(int*a int*b,int*c) 然后main 函数里你应该声明三个指针 然后为他们动态分配内存： int *device_a, *device_b, *device_c;