I don't see GPU in Virtual Machine

[Cezar397]

Hello, I added the video card to the virtual machine and it does not appear in my system.
From nvidia-smi it shows me that the grid has been allocated.

I have another virtual machine where my video card appears.

 VM 005 witch is working
args: -uuid 00000000-0000-0000-0000-000000000105
balloon: 0
bios: ovmf
boot: order=ide0;ide2;net0
cores: 6
cpu: host
cpuunits: 1024
efidisk0: local-lvm:vm-105-disk-0,efitype=4m,pre-enrolled-keys=1,size=4M
hostpci0: 0000:3b:00.0,mdev=nvidia-49,pcie=1
ide0: local-lvm:vm-105-disk-1,cache=writeback,size=150G
ide2: local:iso/virtio-win-0.1.229.iso,media=cdrom,size=522284K
machine: pc-q35-7.1
memory: 16384
meta: creation-qemu=7.1.0,ctime=1679578186
name: VM005
net0: virtio=7E:25:E5:8C:3C:44,bridge=vmbr0,firewall=1
numa: 0
ostype: win10
scsihw: virtio-scsi-pci
smbios1: uuid=b354d1bc-a958-40b4-97fe-2b0ab5498f8f
sockets: 1
tags: ssd
vga: memory=256
vmgenid: aa4ba3db-a3a6-41de-ac68-cf8efcb1c275

tags: hdd
vmgenid: 0c25f078-ac83-418b-b130-bb4b43082da3

#GNU nano 5.4                                         /etc/pve/qemu-server/108.conf
agent: 1
balloon: 0
bios: ovmf
boot: order=ide0;ide2;net0
cores: 6
cpu: host
cpuunits: 1024
efidisk0: local-lvm:vm-108-disk-0,efitype=4m,size=4M
hostpci0: 0000:3b:00.0,mdev=nvidia-49,pcie=1
ide0: local-lvm:vm-108-disk-1,cache=writethrough,size=150G
ide2: local:iso/virtio-win-0.1.229.iso,media=cdrom,size=522284K
machine: pc-q35-7.2
memory: 16384
meta: creation-qemu=7.2.0,ctime=1680529223
name: VM008
net0: virtio=4E:A3:E4:7C:93:DC,bridge=vmbr0,firewall=1
numa: 0
ostype: win11
scsihw: virtio-scsi-pci
smbios1: uuid=fdc01204-99ad-41ec-ab09-54af65f0fda2
sockets: 1
tpmstate0: local-lvm:vm-108-disk-2,size=4M,version=v2.0
vmgenid: aa12fec4-3e83-48dc-85e2-0284f1b2eacc

I don't understand where the problem is.

Another thing is that after a while, the VM005 goes very hard. The video card limits the frames to 15 FPS.

 

[noel.]

Hi,

So, if I understand correctly, you are splitting your p40 into two vgpus. (we have a wiki entry on vgpus[1]. It might be of help to you) and it works in the win10 vm but not the win 11.

From what I can tell, Windows 11 vGPU support for the Ubuntu kernel (from which Proxmox gets its kernel) was only officially added with 15.1[2] and from your `nvidia-smi` attachment, it looks like you are running 14.X. If possible, I'd try upgrading.

2 other things I would try for testing purposes is seeing if reaching through the whole GPU (or the other p40) works (see [3] and [4]) and seeing if creating a second win10 vm and adding a vgpu works.

---

[1]: https://pve.proxmox.com/wiki/NVIDIA_vGPU_on_Proxmox_VE_7.x
[2]: https://docs.nvidia.com/grid/latest/product-support-matrix/index.html
[3]: https://pve.proxmox.com/wiki/PCI(e)_Passthrough
[4]: https://pve.proxmox.com/wiki/PCI_Passthrough