My Linux system has two GPUs: a Tesla and a Quadro. I want to do the computation on the Tesla and then send the results to the Quadro to do the OpenGL rendering.
Do I need to specify that OpenGL commands are only sent to the Quadro device, and not the Tesla?
As far as I know, the WGL_NV_gpu_affinity extension does this on Windows system. Is there some sort of equivalent on Linux?