Using GPU after setting up Discrete Device Assignment (DDA) with Windows 10 VM

Question

I have setup a VM to allow it to use a graphics card through DDA per these instructions (and much other investigation):

https://docs.microsoft.com/en-us/windows-server/virtualization/hyper-v/deploy/deploying-graphics-devices-using-dda

I can login to VM and have assigned it an Nvidia Quadro graphics card.

However, where is the viewport acceleration occuring? Through the Hypervisor viewer? Remote desktop? Only through the video card output? Or all of the above?

I can't find an definitive answer for this anywhere.

Thanks for any help

m

Answer 1

It is being accelerated through Remote Desktop for certain. I have used Performance Monitor and watched an NVIDIA GPU metric and observed the GPU usage, memory, etc.

However, logging in through Hypervisor viewer did not, in my case, trigger use of the GPU. The NVIDIA GPU metrics were all 0. Not sure why this is.

However, when I

1. Logged in with RDP, observed GPU usage as expected, closed the window (but did not log out of the session)

2. Later connected with the same login account and Hypervisor, I observed the NVIDIA GPU metric and the GPU was still being used

RDP must trigger GPU usage in some way that Hypervisor viewer does not.