Graphics passthrough using as host

A quick question, I’m using my 1080Ti as a pass-through device on a windows VM. But one other reason I bought the 1080Ti is to use it for machine learning. Is there a way to use the GPU for this application. Because the way I understand it, the device itself is blacklisted for normal use. Is there a way to initialize the GPU after the system boots up?

Or would it be better to have a different vm (linux based) and use the same pass-trough method to use the GPU on a linux based system for ML.

What are your thoughts on this?

I think what you want is bind the GPU to the host on boot, and once your VM starts up bind it to the guest instead?
This is currently not possible as far as I’m aware because they don’t like being re-initialised. Some cards do, but most of them don’t.

1 Like