Hi,
I came into a problem with a MacBook Pro with an NVIDIA 9400M card.
Specifically I have developed this routine in an application :
size_t GetGPUmemory(){
size_t memfree, memtot;
cudaError_t error = cudaMemGetInfo(&memfree, &memtot);
if (error != cudaSuccess)
return 0
return memfree;
}
Without checking the MacBook goes 0-5MB and sometimes to 2GBs (it has only 256MB VRAM).
So by putting the cudaSuccess it stays stable at 0M. I am wondering if anyone knows if setting to zero when cuda fails to retreive memory is correct. Also what happens and cuda does not succeed?
Cheers,
Alexander.