Can anyone tell me why this kernel code crashes

Accelerated Computing CUDA CUDA Programming and Performance
1

Can anyone tell me why this kernel code crashes every time please:

int same =1;

for (i =0; i<32; i++){

if (toFind[i] != getting[i]) same = 0;

}

if (same == 1) {ptgot[0] = key[0]; ptgot[1] = key[1]; ptgot[2] = key[2]; ptgot[3] = key[3]; ptgot[4] = key[4]; ptgot[5] = key[5];}

same = 1;

If i remove:

if (toFind[i] != getting[i]) same = 0;

it runs ok.

Thanks in advance.

2

Directly with this snippet, no. But I’m sure using cuda-memcheck will prove insightful for you.

3

Heres the entire code attached. Line 184 is where problem is.

I’ll have a look at cuda-memcheck.

Thanks for your quick response.
counter.cu (8.72 KB)

4

Well, the two only things I see are:

    [*]The out of bound access to getting line 182

    [*]The format error in printf line 243

The first error having undefined behaviour, it might be the cause of your problem. But beside this, nothing obvious that I can spot immediately.

5

Ok, i’ll check that out.

Thanks for your help.