Hi, I’m trying to set up GPU Passthrough on an AWS g4dn.metal instance running Windows Server 2019 with HyperV and guest OS is latest Ubuntu 20.04.
The GPU on those instances are Tesla T4.
The guidelines I’m following are from here:
https://docs.nvidia.com/grid/latest/grid-vgpu-user-guide/index.html#using-gpu-pass-through-windows-server-hyper-v
I can successfully disable and dismount one of the GPUs and assign them to the VM. I can verify the GPU is assigned properly via Get-VMAssignableDevice -VMName Ubuntu and everything seems to be working fine until it fails to start the VM and I’m getting the following error message:
Virtual Pci Express Port (Instance ID 02583E40-07CB-4C02-9CF4-B4D3A55868F9): Failed to Power on with Error 'The hypervisor could not perform the operation because the object or value was either already in use or being used for a purpose that would not permit completing the operation.'.
Could someone explain if this is a problem with AWS BIOS settings or whether I’m doing a something wrong during the setup? I can provide further information / steps if necessary.
P.S.: I also ran a script from Microsoft so check whether the T4 is available for DDA from here which gave good results for all 8 T4.