I am having problems with my current openCL program. I would like to debug the kernel, but I am using the GPU. I have a Nvidia GPU and I have it set up to use that. Am I able to make it so that it uses the CPU for debugging? I haven’t been able to find any tutorials or talk about that on Google. Let me know, thanks.
AMD or Apple’s OpenCL libraries support a CPU target. NVIDIA’s doesn’t.
Man I am gonna have trouble figuring out this problem now…
Regardless, even if there was a CPU implementation, memory and thread management is not exactly the same (due to obvious differences), so you could literally get away with murder on the CPU implementation, i.e., things that will work perfectly fine on the CPU could cause problems in the GPU. This was one of the limitations of running CUDA in emulation mode. It was generally helpful, but it did not necessarily uncover all the problems.