Is there an official “Lambda” way to setup the GPU for ML only, and not use it at all for graphics?
I don’t want any graphics program or X itself to run on the GPU.
I am using a TensorBook with a 2080, running Ubuntu 20.04.
Thanks,
Victor
Hi Victor,
Likely the easiest way to accomplish this is by simply isolating the multi-user target, as right now your Tensorbook is running in the graphical target.
To do this, run:
systemctl isolate multi-user.target
This will require you to use your login credentials again and you’ll be limited to a TTY. If you find this works for you, to make the multi-user target the default target for when your machine boots, run:
systemctl set-default multi-user.target
Another option would be to completely reinstall Ubuntu Server, as opposed to Ubuntu Desktop. If you’d like to try this, here is our guide on how to do so. There is a link to another guide on how to install Ubuntu installation media, just be sure to choose the Ubuntu Server ISO.
Hopefully this helps and please do let us know should you have any further questions or concerns!
Regards,
Calvin Wallace
Linux Support Engineer
Calvin,
What do you mean by “limited to a TTY”? I still need to use X windows, I just don’t want it (or any GUI program) to use the GPU.
Thanks,
Victor