Tesla P4 - Testing

Hi there.

We have a server running vSphere (ESXi 6.7) and have one Tesla P4 pass-through to a single VM (Windows Server 2019 with NViDIA driver version: 425.25). Everything appears to be working fine, but need help figuring out a good and easy test that I can perform on the VM via RDP that would tell me the if GPU is working as intended and that applications, such as Google Earth Pro, Esri, and AutoDesk products are taking advantage of its power and not using the onboard GPU. Right now, nvidia-smi and Task Manager doesn’t indicate any GPU load or running processes.

After some digging, it appears the Tesla P4 is in Compute mode. Does that mean the card is useless for graphic applications? Is there any configuration that can be done to help with graphic processing?

Any guidance is much appreciated.

Thanks, J!

Hi WS,

you should enable the GPU for remote sessions (local policy). Afterwards you should see the GPU load :)

regards
Simon