GPU Passthrough for Windows 10 VM


I am using a Lambda TensorBook to do 3D modeling for my company, running Ubuntu 18.04. Unfortunately a 3rd Party software requires Windows 10 or MacOS and needs to have OpenGL 4.3+ installed to run properly. I have set up a Windows 10 VM using KVM virt-Manager, that works pretty well, but I think I need to set up GPU pass-through for the software to properly visualize my data/models. I am wondering if there are any suggestions on how to do this. I have already installed mesa3D on my host but the software still doesn’t run properly on my guest VM. I have seen a few how-to’s, but they are generally more along the lines of running a VM for gaming purposes.


I need OpenGL 4.3+ to be accessible on a Windows 10 VM running through KVM virt-Manager. I have already installed Mesa3D on my host, Ubuntu 18.04, but still not able to run the 3rd party software on my windows 10 VM.

Any suggestions/help is appreciated

This is really quite tricky to get working and usually involves running a script that would prevent your system from actually displaying anything in your host OS which is probably a non-starter if you’re running Ubuntu. My suggestion would be to try to do a dual boot w/ Windows. I would highly recommend that you do that dual boot on two separate drives though!

Hey thanks for the reply. That is really helpful. I’m not familiar with how to set up dual-boot systems, and while there are lots of articles on the web about dual booting, could you recommend a few walk-throughs to help me get started? Thanks.

It’s actually really difficult to set up a dual boot like that — I don’t have a good tutorial off the top of my head that’s easy to follow. The ArchLinux wiki has a lot of info on it but it’s complicated.