CUDA program returning incorrect results [duplicate] - cuda

Question

This question already has answers here:
What is the canonical way to check for errors using the CUDA runtime API?
(4 answers)
Closed 2 years ago.
I have copied a vector addition example from the book "CUDA By Example" and I am getting unexpected incorrect results. Here is my code
#define N (33*1024)
__global__
void add(int *a, int *b,int *c){
int tid = threadIdx.x+blockIdx.x*blockDim.x;
while (tid < N){
c[tid] = a[tid]+b[tid];
tid+=blockDim.x*gridDim.x;
}
}
int main()
{
int a[N], b[N], c[N];
int *dev_a, *dev_b, *dev_c;
cudaMalloc((void**)&dev_a,N*sizeof(int));
cudaMalloc((void**)&dev_b,N*sizeof(int));
cudaMalloc((void**)&dev_c,N*sizeof(int));
for(int i = 0 ; i<N;i++){
a[i]= -i;
b[i]= i*i;
}
cudaMemcpy(dev_a,a,N*sizeof(int),cudaMemcpyHostToDevice);
cudaMemcpy(dev_b,b,N*sizeof(int),cudaMemcpyHostToDevice);
cudaMemcpy(dev_c,c,N*sizeof(int),cudaMemcpyHostToDevice);
add<<<128,128>>>(dev_a,dev_b,dev_c);
cudaMemcpy(c,dev_c, N*sizeof(int), cudaMemcpyDeviceToHost);
bool success=true;
//print results
for(int i=0; i<N;i++){
if((a[i]+b[i])!=c[i]){
printf("Error: %d + %d != %d\n",a[i],b[i],c[i]);
success=false;
}
}
if(success) printf("we did it!\n");
cudaFree(dev_a);
cudaFree(dev_a);
cudaFree(dev_a);
printf("done");
return EXIT_SUCCESS;
}
and I am getting a bunch of incorrect addition results, here is just a few
Error: -33784 + 1141358656 != 255
Error: -33785 + 1141426225 != 0
Error: -33786 + 1141493796 != 0
Error: -33787 + 1141561369 != 0
Error: -33788 + 1141628944 != 4609792
Error: -33789 + 1141696521 != 0
Error: -33790 + 1141764100 != 4207408
and there are many many more. I am a complete CUDA beginner but my guess is I either
A) copied the code incorrectly from the book OR
B) the incorrect results come from the fact that I am using CUDA 10 which came out long after this book was written
EDIT: I restarted my computer and it worked

Answer 1

I happen to be able to repeat your error if I alter my configuration. Something is probably wrong with your configuration, too. When I used fitting CUDA and driver versions it worked after fixing a minor typo:
cudaFree(dev_a); //this line is copied three times in your code
Please wrap your cuda call with something like the following to check the return value. Must be one of the cuda functions failed.
#define CUDA_CHECK_RETURN(value) { \
cudaError_t _m_cudaStat = value; \
if (_m_cudaStat != cudaSuccess) { \
fprintf(stderr, "Error %s at line %d in file %s\n", \
cudaGetErrorString(_m_cudaStat), __LINE__, __FILE__); \
exit(1); \
} }
//for example
CUDA_CHECK_RETURN(cudaMemcpy(c,dev_c, N*sizeof(float), cudaMemcpyDeviceToHost));
It should tell you what might go wrong.