I am getting started with CUDA and the documentation says I need two CUDA enabled GPUs for local debugging. So my questions are:
- Do really both have to support CUDA or only one of them?
- Do they or should they be the same or can I use a low-end version for display and a high-end card as the target?
- Could I also get away with just one CUDA enabled GPU? If so what are the disadvantages?
Thanks.