Some questions arround virtualization and passing through a gpu

[dasBaum]

Hello guys and girls,

first of all im sorry for my bad english, im not a native speaker. At the moment im studying business informatics and doing some little programming stuff. Im using Windows 10, but for numerous reasons i want finally switch completely to arch linux. Now I ran into the old allknown problem, im also a gamer who want to play games. So I thought it would be a good idea to run a windows 10 vm inside of arch. But now I ran into a problem, I've only got a single 2080ti that i want to passthrough to my vm. After some recherche i discovered that i would need 2 gpu's to do this. But I don't have the money for a second 2080ti, so my question is if it is possible to use a second gpu that is not a 2080ti for the host system. And if this is the case, if it would work if i use a cheap amd gpu(i thought of something like a 5500xt) for the host and the 2080ti for the vm. And if this works, if it would also work that the host system use the 2080ti as long the vm dont run, so that i can do things with tensorflow with arch when im not gaming under windows. I know there is also the possibility of a dual boot system, but i dont like this for numerous reasons. Thanks for your time and help.

in case u need the exact specs:

CPU: Ryzen 3700x

Ram: 32gigs

GPU: MSI 2080ti Gaming X Trio

MB: MSI X570 Gaming Edge Wifi 

EDIT: I want to virtualize with KVM/Qemu

[Electronics Wizardy]

Yea a cheap gpu will work fine here, they don't have to be the same.

 

Yea normally you can use the gpu for the host os, but this cna be trickly to have it change drives while running.

 

Have you checked your board does the iommu groups correctly for gpu passthrough?

[dasBaum]

1 minute ago, Electronics Wizardy said:

Have you checked your board does the iommu groups correctly for gpu passthrough?

no but i will do it when im home again, thanks for the advice

[4L83R7]

GPU passthrough to a Windows guest OS on a Linux host is possible as long as you have another GPU, whether integrated or discrete, for the host OS. Up until a few weeks ago (if I remember correctly), you had to jump through hoops to get the guest OS to properly install drivers for a Nvidia GPU as GPU passthrough wasn't supported - the infamous error 43. I recently read somewhere that Nvidia is now officially supporting it in a beta driver. I am yet to test this new driver as I don't have Linux running yet. But I did successfully set up a passthrough last year after several hours of googling and typing commands in Terminal.