#include "cuda_runtime.h"
#include "device_launch_parameters.h"

#include <stdio.h>
//changed into global variables
const int arraySizeX = 12;
const int arraySizeY = 12;
const int numThreadsPerBlock = 5;
//dividing into correct number of blocks (safe)
int numBlocks = arraySizeX * arraySizeY / numThreadsPerBlock + ((arraySizeX * arraySizeY) % numThreadsPerBlock == 0 ? 0 : 1);

cudaError_t addWithCuda(int **c, int **a, int **b, size_t sizeX, size_t sizeY);

__global__ void addKernel(int **c, int **a, int **b)
{
    int i = threadIdx.x + blockIdx.x*numThreadsPerBlock; //offset for number of blocks
	int j = threadIdx.y + blockIdx.y*numThreadsPerBlock; //offset for number of blocks
    c[i][j] = a[i][j] + b[i][j];
}

int main()
{
	int **a;
	int **b;
	int **c;
    //int a[arraySizeX][arraySizeY];
    //int b[arraySizeX][arraySizeY];
    //int c[arraySizeX][arraySizeY];
	int i;
	a = (int**)malloc(arraySizeX * sizeof(int*));
	for(i = 0; i < arraySizeX; i++)
		a[i] = (int*)malloc(arraySizeY * sizeof(int));

	b = (int**)malloc(arraySizeX * sizeof(int*));
	for(i = 0; i < arraySizeX; i++)
		b[i] = (int*)malloc(arraySizeY * sizeof(int));
	c = (int**)malloc(arraySizeX * sizeof(int*));
	for(i = 0; i < arraySizeX; i++)
		c[i] = (int*)malloc(arraySizeY * sizeof(int));

	//init arrays
	int j;
	int iterator = 1;
	for(i = 0; i < arraySizeX; i++)
		for(j = 0; j < arraySizeY; j++)
		{
			a[i][j] = iterator;
			b[i][j] = iterator * 10;
			c[i][j] = 0;
			iterator++;
		}

    // Add vectors in parallel.
    cudaError_t cudaStatus = addWithCuda(c, a, b, arraySizeX, arraySizeY);
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "addWithCuda failed!");
        return 1;
    }

	for(i = 0; i < arraySizeX; i++)
	{
		for(j = 0; j < arraySizeY; j++)
		{
			printf("%d ", c[i][j]);
		}
		printf("\n");
	}
    //printf("{1,2,3,4,5} + {10,20,30,40,50} = {%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d,%d}\n",
    //    c[0], c[1], c[2], c[3], c[4], c[5], c[6], c[7], c[8], c[9], c[10], c[11]);
	system("pause");
	// cudaThreadExit must be called before exiting in order for profiling and
	// tracing tools such as Parallel Nsight and Visual Profiler to show complete traces.
    cudaStatus = cudaThreadExit();
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaThreadExit failed!");
		system("pause");
        return 1;
    }

    return 0;
}

// Helper function for using CUDA to add vectors in parallel.
cudaError_t addWithCuda(int **c, int **a, int **b, size_t sizeX, size_t sizeY)
{
    int **dev_a = 0;
    int **dev_b = 0;
    int **dev_c = 0;
    cudaError_t cudaStatus;

	int i;
	cudaMalloc((void**)dev_a, sizeX * sizeof(int*));
	for(i = 0; i < sizeX; i++)
		cudaMalloc((void**)&dev_a[i], sizeY * sizeof(int));


	cudaMalloc((void**)dev_b, sizeX * sizeof(int*));
	for(i = 0; i < sizeX; i++)
		cudaMalloc((void**)&dev_b[i], sizeY * sizeof(int));

	cudaMalloc((void**)dev_c, sizeX * sizeof(int*));
	for(i = 0; i < sizeX; i++)
		cudaMalloc((void**)&dev_c[i], sizeY * sizeof(int));

	//dev_b = (int**)malloc(arraySizeX * sizeof(int*));
	//for(i = 0; i < arraySizeX; i++)
	//	dev_b[i] = (int*)malloc(arraySizeY * sizeof(int));
	//dev_c = (int**)malloc(arraySizeX * sizeof(int*));
	//for(i = 0; i < arraySizeX; i++)
	//	dev_c[i] = (int*)malloc(arraySizeY * sizeof(int));

    // Choose which GPU to run on, change this on a multi-GPU system.
    cudaStatus = cudaSetDevice(0);
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaSetDevice failed!  Do you have a CUDA-capable GPU installed?");
		system("pause");
        goto Error;
    }

    // Allocate GPU buffers for three vectors (two input, one output)    .
  //  cudaStatus = cudaMalloc((void**)&dev_c, size * sizeof(int*));
  //  if (cudaStatus != cudaSuccess) {
  //      fprintf(stderr, "cudaMalloc failed!");
		//system("pause");
  //      goto Error;
  //  }

  //  cudaStatus = cudaMalloc((void**)&dev_a, size * sizeof(int));
  //  if (cudaStatus != cudaSuccess) {
  //      fprintf(stderr, "cudaMalloc failed!");
		//system("pause");
  //      goto Error;
  //  }

  //  cudaStatus = cudaMalloc((void**)&dev_b, size * sizeof(int));
  //  if (cudaStatus != cudaSuccess) {
  //      fprintf(stderr, "cudaMalloc failed!");
		//system("pause");
  //      goto Error;
  //  }

    // Copy input vectors from host memory to GPU buffers.
    cudaStatus = cudaMemcpy(dev_a, a, sizeX * sizeY * sizeof(int), cudaMemcpyHostToDevice);
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaMemcpy failed!");
		system("pause");
        goto Error;
    }

    cudaStatus = cudaMemcpy(dev_b, b, sizeX * sizeY * sizeof(int), cudaMemcpyHostToDevice);
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaMemcpy failed!");
		system("pause");
        goto Error;
    }

    // Launch a kernel on the GPU with one thread for each element.
    addKernel<<<numBlocks, numThreadsPerBlock>>>(dev_c, dev_a, dev_b);

    // cudaThreadSynchronize waits for the kernel to finish, and returns
    // any errors encountered during the launch.
    cudaStatus = cudaThreadSynchronize();
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaThreadSynchronize returned error code %d after launching addKernel!\n", cudaStatus);
		system("pause");
        goto Error;
    }

    // Copy output vector from GPU buffer to host memory.
    cudaStatus = cudaMemcpy(c, dev_c, sizeX * sizeY * sizeof(int), cudaMemcpyDeviceToHost);
    if (cudaStatus != cudaSuccess) {
        fprintf(stderr, "cudaMemcpy failed!");
		system("pause");
        goto Error;
    }

Error:
    cudaFree(dev_c);
    cudaFree(dev_a);
    cudaFree(dev_b);

    return cudaStatus;
}