Skip to content

Instantly share code, notes, and snippets.

@nwh
Created July 12, 2014 01:08
Show Gist options
  • Save nwh/2bf2557437cbea55b3c8 to your computer and use it in GitHub Desktop.
Save nwh/2bf2557437cbea55b3c8 to your computer and use it in GitHub Desktop.
Take care with CUDA Unified Memory
#include <iostream>
#include <cstdio>
using std::cout;
using std::endl;
__global__
void print_a(int* a, const int n) {
for (int i=0; i<n; i++) {
printf("a[%d] = %d\n",i,a[i]);
}
}
int main() {
// allocate a
const int lena = 6;
int* a;
cudaMallocManaged(&a,lena*sizeof(*a));
// allocate b
const int lenb = 1;
int *b;
cudaMallocManaged(&b,lenb*sizeof(*b));
// set a
for (int i=0; i<lena; i++) {
a[i] = i;
}
// set b
*b = 55;
// print a from device
print_a<<<1,1>>>(a,lena);
// add this device sync to fix seg fault
//cudaDeviceSynchronize();
// print b from host
cout << "*b: " << *b << endl;
cudaFree(a);
cudaFree(b);
return 0;
}
go: go.cu
nvcc -gencode arch=compute_30,code=sm_30 $< -o $@
run: go
./go
clean:
$(RM) go
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment