Different Results on Two Different Cards..

Accelerated Computing CUDA CUDA Programming and Performance
#1

I have written a GPU application, which squares number 1-9 on the GPU. I have executed the CUDA program on two different machines; however, I get completely different results. I am using a Tesla C2075 device and a GeForce 9400M

Thanks,
Any help on this issue would be greatly appreciated.

#2

Likely a precision issue. Since the 9400M doesn’t support double precision, perhaps you are just comparing its single-precision results with the Tesla’s double-precision results?

#3

My results for the Tesla C2075 is

0 0.000000

1 1.000000

2 2.000000

3 3.000000

4 4.000000

5 5.000000

6 6.000000

7 7.000000

8 8.000000

9 9.000000

My results for the 9400M is

0 0.000000

1 1.000000

2 4.000000

3 9.000000

4 16.000000

5 25.000000

6 36.000000

7 49.000000

8 64.000000

9 81.000000

#4

What is the code leading to these results?

#5

Hrm, not a precision issue then. Posting your code would help.

#6

#include <stdio.h>
#include <cuda.h>

// Kernel that executes on the CUDA device
global void square_array(float *a, int N)
{
int idx = blockIdx.x * blockDim.x + threadIdx.x;
if (idx<N) a[idx] = a[idx] * a[idx];
}

// main routine that executes on the host
int main(void)
{
float *a_h, *a_d; // Pointer to host & device arrays
const int N = 10; // Number of elements in arrays
size_t size = N * sizeof(float);
a_h = (float *)malloc(size); // Allocate array on host
cudaMalloc((void **) &a_d, size); // Allocate array on device

// Initialize host array and copy it to CUDA device
for (int i=0; i<N; i++) a_h[i] = (float)i;
cudaMemcpy(a_d, a_h, size, cudaMemcpyHostToDevice);

// Do calculation on device:
int block_size = 4;
int n_blocks = N/block_size + (N%block_size == 0 ? 0:1);
square_array <<< n_blocks, block_size >>> (a_d, N);

// Retrieve result from device and store it in host array
cudaMemcpy(a_h, a_d, sizeof(float)*N, cudaMemcpyDeviceToHost);

// Print results
for (int i=0; i<N; i++) printf("%d %f\n", i, a_h[i]);

// Cleanup
free(a_h); cudaFree(a_d);
}

Thank you!

#7

So the code does not execute at all on the C2075. To notice when this happens, always check return codes of CUDA calls.
How do you compile your code?

#8

For the squaring number application, it seems not to be executing. I compile my code with nvcc application.cu -o application. Also, I executed a simple hello world application on the C2075, and I got Hello Hello, when it should be of course, Hello World.

The code for it is:

#include <stdio.h>

const int N = 16;

const int blocksize = 16;

global

void hello(char *a, int *b)

{

a[threadIdx.x] += b[threadIdx.x];

}

int main()

{

char a[N] = "Hello \0\0\0\0\0\0";

int b[N] = {15, 10, 6, 0, -11, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};

char *ad;

int *bd;

const int csize = N*sizeof(char);

const int isize = N*sizeof(int);

printf("%s", a);

cudaMalloc( (void**)&ad, csize ); 

cudaMalloc( (void**)&bd, isize ); 

cudaMemcpy( ad, a, csize, cudaMemcpyHostToDevice ); 

cudaMemcpy( bd, b, isize, cudaMemcpyHostToDevice ); 



dim3 dimBlock( blocksize, 1 );

dim3 dimGrid( 1, 1 );

hello<<<dimGrid, dimBlock>>>(ad, bd);

cudaMemcpy( a, ad, csize, cudaMemcpyDeviceToHost ); 

cudaFree( ad );



printf("%s\n", a);

return EXIT_SUCCESS;

}

Thanks

#9

I took your advice and did some error checking.

int main()

{

    char a[N] = "Hello \0\0\0\0\0\0";

    int b[N] = {15, 10, 6, 0, -11, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};

char *ad;

    int *bd;

    const int csize = N*sizeof(char);

    const int isize = N*sizeof(int);

printf("%s", a);

if (cudaSuccess != cudaMalloc( (void**)&ad, csize ) )

    printf("AD / Error!\n");

if (cudaSuccess != cudaMalloc( (void**)&bd, isize ))

    printf("BD / Error!\n");

cudaMemcpy( ad, a, csize, cudaMemcpyHostToDevice );

cudaMemcpy( bd, b, isize, cudaMemcpyHostToDevice );

dim3 dimBlock( blocksize, 1 );

    dim3 dimGrid( 1, 1 );

    hello<<<dimGrid, dimBlock>>>(ad, bd);

    if (cudaSuccess != cudaGetLastError())

    printf("Kernal - Error\n");

cudaMemcpy( a, ad, csize, cudaMemcpyDeviceToHost );

    cudaFree( ad );

printf("%s\n", a);

    return EXIT_SUCCESS;

}

My output for this is:

Hello AD / Error!

BD / Error!

Kernal - Error

Hello

Do you know what causes this?

Thanks a lot -

#10

The exact error code would be of interest here.
Anyway, I can guess what it is - try compiling with [font=“Courier New”]nvcc -arch sm_20 application.cu -o application[/font] .