My PC has an onboard graphics processor. I can get a CUDA compatible card of compute capability 3 and not 3.5. I wonder if i can debug my CUDA application (using the onboard graphics processor for display) or should i buy another NVIDIA card? If the former option is possible, what setting should i make?
Many thanks
Hi,
Single-GPU Debugging in cuda-gdb is available on devices with compute capability 3.5 and above. This allows the same device to be used for both display and CUDA debugging.
Based on your setup description, though, if you are going to use onboard graphics to drive your display, then the card with compute capability 3.0 can be used for running and debugging CUDA. What type of application are you planning on debugging? Will it be using CUDA/OpenGL interop?