Nvidia has now officially enabled GPU passthrough support for Windows virtual machines on GeForce graphics cards.
In other words, this effectively means it’s possible to run a Linux machine and then run a virtual Windows machine within it, and hand that unfettered access to a graphics card. This is a big win for those wanting to run Windows games from within a virtual machine on your Linux desktop. They will be able to play Windows-based games using a virtual machine with GPU passthrough enabled.
GeForce desktop graphics cards that are based on the Kepler, Maxwell or Pascal architectures or newer will be able to take advantage of GPU passthrough when running a virtual machine on Windows.
Nvidia GeForce GPU Passthrough limitations
In order to start using GeForce GPU Passthrough on your system, you’ll need to ensure that you have the R465 driver installed.
GeForce GPU passthrough supports only 1 virtual machine. If you want to enable multiple virtual machines to have direct access to a single GPU or want the GPU to be able to assign virtual functions to multiple virtual machines, you will need to use NVIDIA Tesla, Quadro, or RTX enterprise GPUs.
You can’t leverage the same GPU being used by the host OS for virtualization. One GPU is required for the Linux host OS and one GPU is required for the Windows virtual machine. In other words, you will need 2 GPUs: 1 for Linux host, and 1 being passed through to Windows.