It is known that one single GPU card can run multiple kernels simultaneously as long as these kernels are from the same application process. Now suppose I installed two Fermi cards on a PC, is it possible that, when I run two distinct applications simultaneously, one kernel from one application run on one GPU card, and another kernel from another application run at the same time on another GPU card? Thanks.
cudasetsetdevice(0) in one prpgram and cudasetdevice(1) in the other.