I’ve been searching through documentation, google and the forum, but I did not a satisfying answer to the following question:

What is the CUDA_SAFE_CALL macro doing and when do I need it? Do I need it at all?



P.S.: CUDA 3.0

CUDA_SAFE_CALL is a convenience macro that checks the return value from a cuda call. If it is not cudaSuccess, it prints an error message and aborts the program. It is part of the cutil package in the SDK, which (as we are repeatedly informed) should not be used in your programs. The behavior can change at any time.

I find something like CUDA_SAFE_CALL useful for programs I write and use myself, so I copy the definition of it into my own project headers.

Thank you, that solved my issue.