I have two graphic cards on my machine, one is connected to my monitor
and the other (GTX580) is not because I’d like to use it for CUDA programming.
So, it seems that GTX580 does not do anything when CUDA is not running.
But when I checked free memory of the GTX580,
via NVIDIA Control Panel -> Workstation -> View system topology menu,
only 1503 MB remains for free memory out of the total memory 2816 MB.
Since I want GTX580 to be dedicated to CUDA, I hope it to have usable memory as much as possible.
Anybody has an idea about why my GTX580’s memory is being used although it is never connected to the monitor ?