#include <stdio.h>

__global__ void fnKern(int *pA, int *pB, int *pSum)
{
    pSum[threadIdx.x] = pA[threadIdx.x] + pB[threadIdx.x];
}

int main()
{
    const int N = 512;
    int SIZE = N * sizeof(float);
    int a[N], b[N], sum[N], aa[N], bb[N];

    for (int i = 0; i < N; i++)
    {
        a[i] = b[i] = 100;
        //sum[i] = 0;
    }

    printf("A: %d\nB: %d\n", a[0], b[0]);

    int *pA, *pB, *pSum;
    cudaMalloc((void**)&pA, SIZE);
    cudaMalloc((void**)&pB, SIZE);
    cudaMalloc((void**)&pSum, SIZE);

    cudaMemcpy(pA, a, SIZE, cudaMemcpyHostToDevice);
    cudaMemcpy(pB, b, SIZE, cudaMemcpyHostToDevice);
    //cudaMemcpy(pSum, sum, SIZE, cudaMemcpyHostToDevice);

    fnKern<<<1, N>>>(pA, pB, pSum);

    cudaMemcpy(aa, pA, SIZE, cudaMemcpyDeviceToHost);
    cudaMemcpy(bb, pB, SIZE, cudaMemcpyDeviceToHost);
    cudaMemcpy(sum, pSum, SIZE, cudaMemcpyDeviceToHost);

    printf("A: %d\nB: %d\nSum: %d\n", aa[0], bb[0], sum[0]);
    return 0;
}


/* OUTPUT:
A: 100
B: 100
A: 0
B: 0
Sum: 0
*/