I have a Tesla P4 that in ESXi 8.0 being passed through to a Windows 10 / Windows 11 / Ubuntu 22.04 VMs at different times to test the GPU hardware acceleration in Plex. Oddly enough in Ubuntu upon hypervisor and vm reboot, I get it working but when the vm reboot it stops and I have to reboot the hypervisor host.
I'm not trying to pass through a vgpu just the whole gpu so that I can use it with Plex for my homelab. I have all of the memory allocated all the time. I have tried many drivers in Ubuntu and Windows. I see the card show up in all of the OSes that I have tested but after the drivers are install in Windows, I get a Code 10 (not Code 43).
This device cannot start. (Code 10)
Insufficient system resources exist to complete the API.
I have not installed any Nvidia Grid VIB files because I don't have any and waiting on NVIDIA to respond so I can either get a trial or pay for a license.
I thought this was only supposed to be an issue with consumer level cards and not datacenter cards. I thought the NVIDIA licensing was to enable the creation of vGPUs so multiple VMs can use the same hardware which I also do not need.
Any thoughts will be great.
