Cuda factorial value invalid

Question

I'm starting to learn CUDA. I wrote program to calculate factorial. Code is working but when i calculate factorial more what 12 i getting wrong values. Why CUDA getting invalid value? How to fix this? This is my code.

#include "cuda_runtime.h"
#include "device_launch_parameters.h"
#include <stdio.h>
#include <conio.h>
#define CUPRINTF cuPrintf

__device__ int silnia(int n)
{
    if (n<2)
        return 1; //silnia z 0 i 1 wynosi 1
    return n*silnia(n - 1);
}
__global__ void kernel(int *a)
{
    *a = silnia(15);
}

int main()
{
    cudaEvent_t start, stop;
    float elapsedTime;
    cudaEventCreate(&start);
    cudaEventRecord(start, 0);
    int *dev_a,a;
    cudaEventRecord(start);
    cudaMalloc((void**)&dev_a, sizeof(int));
    kernel << <1, 1 >> >(dev_a);
    cudaMemcpy(&a, dev_a, sizeof(int), cudaMemcpyDeviceToHost);
    cudaEventCreate(&stop);
    cudaEventRecord(stop, 0);
    cudaEventSynchronize(stop);
    cudaEventElapsedTime(&elapsedTime, start, stop);
    printf("Elapsed time : %f ms\n", elapsedTime);
    printf("%d", a);
    getch();
    return 0;
}

Thx for help.

Answer 1

12 factorial is 479001600 which fits in a int variable.

13 factorial is 6227020800 which does not fit in a int variable.

If you change int to unsigned long long (everywhere, including your printf format specifier) you'll get some more range, but eventually you will run out of range.

Answer 2

Simply because there is "overflow". If you continue to use "int" you cannot solve the problem ;)