#include <cuda_runtime.h>
#include <helper_cuda.h>
#include <iostream>
#include <stdio.h>
#define N 5
__global__ void matrixAdd(int* A, int* B, int* C) {
	int i = blockIdx.x * blockDim.x + threadIdx.x;
	int j = blockIdx.y * blockDim.y + threadIdx.y;
	A[i * N + j] = B[i * N + j] + C[i * N + j];
}

int main() {
	int h_A[N][N], h_B[N][N], h_C[N][N];
	size_t size = sizeof(int) * N;
	int* d_A, * d_B, * d_C;
	cudaMalloc((void**)& d_A, size * N);
	cudaMalloc((void**)& d_B, size * N);
	cudaMalloc((void**)& d_C, size * N);
	for (int i = 0; i < N; i++) {
		for (int j = 0; j < N; j++) {
			h_A[i][j] = -1;
			h_B[i][j] = 1;
		}
	}
	cudaMemcpy(d_A, h_A, size * N, cudaMemcpyHostToDevice);
	cudaMemcpy(d_B, h_B, size * N, cudaMemcpyHostToDevice);
	cudaMemcpy(d_C, h_C, size * N, cudaMemcpyHostToDevice);
	dim3 blocks = { 2, 2 };
	dim3 threads = { 16,16 };
	matrixAdd << < blocks, threads >> > (d_A, d_B, d_C);
	cudaMemcpy(h_C, d_C, size * N, cudaMemcpyDeviceToHost);

	for (int i = 0; i < N; i++) {
		for (int j = 0; j < N; j++) {
			std::cout << h_A[i][j] << " + " << h_B[i][j] << " = " << h_C[i][j] << '\n';
		}
	}

	cudaFree(d_A);
	cudaFree(d_B);
	cudaFree(d_C);
	return 0;
}