I have two video cards in my computer. Both are GTX280, only one is link to monitor. I want use one to calculate, and the other one for display. But i don’t know which one is used to display the desktop. Could you tell me how to know it!
With CUDA 2.1 you can find out if device is connected to monitor by calling cudaGetDeviceProperties() and checking kernelExecTimeoutEnabled field of cudaDeviceProp structure. If it is 1 then watchdog timer is active and device is connected to display.
AFAIK there’s no way to achieve this before CUDA 2.1