how many gpu is necessary for debagging program cuda on Parallel Nsight?
In the two requirements listed GPU. Both the GPU should support CUDA? Where will occur debug: on the GPU or CPU? Will it be a place to debug a program if the GPU will be the same (eg 2 GPU GTX570)?
The latest version of Nsight supports local debugging, meaning you will need just a single CUDA-capable card.
The debugging will take place on the GPU you choose with cudaSetDevice