Running GPU on a Headless Ubuntu Machine
I'm running several machines headless using the "dummy" xorg driver and I remote VNC into them.
I've recently acquired a machine with an NVidia GPU card and set it up the same headless way. All works well in this dummy setup, until I install CUDA. When I install CUDA, I lose the VNC since I have no monitor connected. Is there a method to keep the NVidia driver and CUDA installed and available for GPU programming while still having the machine use the dummy driver so I can keep it headless?
Last fiddled with by EdH on 2020-09-16 at 14:34
|