PCIE vGPU Passthrough Issue

[KBlast]

I am currently getting this error when I attempt to start my windows 10 VM. Context: I am trying to pass through a vGPU profile of my 2080 Ti.

kvm: -device vfio-pci,sysfsdev=/sys/bus/mdev/devices/00000000-0000-0000-0000-000000000106,id=hostpci0,bus=pci.0,addr=0x10,rombar=0: vfio 00000000-0000-0000-0000-000000000106: error getting device from group 70: Input/output error
Verify all devices in group 70 are bound to vfio-<bus> or pci-stub and not already in use
start failed: QEMU exited with code 1

Here is the VM conf:

args: -uuid 00000000-0000-0000-0000-000000000106
bios: ovmf
boot: order=ide0;ide2
cores: 4
efidisk0: local-lvm:vm-106-disk-0,efitype=4m,pre-enrolled-keys=1,size=4M
hostpci0: 0000:81:00.0,mdev=nvidia-257,rombar=0
ide0: Main_VMs:vm-106-disk-0,size=20G,
ide2: local:iso/win10.iso,media=cdrom,size=4492992K
machine: pc-q35-7.1
memory: 4096
meta: creation-qemu=7.1.0,ctime=1672798294
name: p99-01
net0: e1000=BE:0F:EE:3F:5E:9B,bridge=vmbr0,firewall=1
numa: 0
ostype: win11
scsihw: virtio-scsi-pci
smbios1: uuid=dd81e453-a8a1-41d7-ad6e-8816d6e6277e
sockets: 1
vmgenid: 51e1dbca-41a9-4223-a6b8-c50d55223f8d

I am on kernel 5.15.30-2-pve with NVIDIA 510.85.03 drivers.

Any ideas on how to fix this? Is there any other information I should share to help others that might have the same problem or help people understand the problem?

I did find other threads on GPU pass-through, but not the same problem and I no one with the same problem attempting vGPU.

 

[Chris]

Hi,
please correct me if I'm wrong here but I don't think the Nvidia 2080 Ti has vGPU support.

For PCIe passthrough of the GPU to the VM you can follow the guide in our wiki, see https://pve.proxmox.com/wiki/PCI(e)_Passthrough
In that case you have to make sure the PVE host is not using the GPU

 

[KBlast]

It's possible with consumer NVIDIA cards with the Maxwell, Pascal, Volta, Turing architecture but not yet Kepler Ampere. See: https://github.com/DualCoder/vgpu_unlock & https://github.com/mbilker/vgpu_unlock-rs

I've done GPU passthrough before on this system no problem (1 entire GPU shared directly & exclusively to 1 VM)

I've confirmed [ 0.000000] DMAR: IOMMU enabled

Any other suggestions?

 

[KBlast]

I solved the problem. My custom profile had "frl_enabled = 60" instead of 1 or 0. I set to 1 and now it works. I also had an incorrect amount of memory (not a clean 2GB) assigned and that was needed as well.

CraftComputing is great, but his text walkthrough had a number of issues, like this =-/

 

[icebreaker374]

I solved the problem. My custom profile had "frl_enabled = 60" instead of 1 or 0. I set to 1 and now it works. I also had an incorrect amount of memory (not a clean 2GB) assigned and that was needed as well.

CraftComputing is great, but his text walkthrough had a number of issues, like this =-/

Worked for me! Thanks mate!