GPU passthrough and iGPU Usage

[agtwashington]

Good Afternoon All,

I followed the guide below and have my P2000 passing through working well to my Plex VM, however I was wondering about the iGPU. I'm currently running an i7 with a integrated (whatever is on the 8700K) GOU, and I'm wanting to run noVNC/SPICE in the web browser. I understand that it fails because the primary display is passed to a guest VM. In my BIOS I have both the integrated, and PCI GPU active, and running lspci shows the iGPU active and paired with a driver. Is it possible to set a setting somewhere to still allow console display in the web browser for guest VMs? (My BIOS is set to use the iGPU as primary)

root@[machine]:~# lspci -nnk
00:00.0 Host bridge [0600]: Intel Corporation Device [8086:3ec2] (rev 07)
        Subsystem: Intel Corporation Device [8086:3ec2]
00:01.0 PCI bridge [0604]: Intel Corporation Skylake PCIe Controller (x16) [8086:1901] (rev 07)
        Kernel driver in use: pcieport
        Kernel modules: shpchp
00:02.0 VGA compatible controller [0300]: Intel Corporation Device [8086:3e92]
        Subsystem: Intel Corporation Device [8086:3e92]
        Kernel driver in use: i915
        Kernel modules: i915
00:14.0 USB controller [0c03]: Intel Corporation Device [8086:a2af]
        Subsystem: Intel Corporation Device [8086:a2af]
        Kernel driver in use: xhci_hcd
00:14.2 Signal processing controller [1180]: Intel Corporation Device [8086:a2b1]
        Subsystem: Intel Corporation Device [8086:a2b1]
00:16.0 Communication controller [0780]: Intel Corporation Device [8086:a2ba]
        Subsystem: Intel Corporation Device [8086:a2ba]
        Kernel driver in use: mei_me
        Kernel modules: mei_me
00:17.0 SATA controller [0106]: Intel Corporation Device [8086:a282]
        Subsystem: Intel Corporation Device [8086:a282]
        Kernel driver in use: ahci
        Kernel modules: ahci
00:1f.0 ISA bridge [0601]: Intel Corporation Device [8086:a2c9]
        Subsystem: Intel Corporation Device [8086:a2c9]
00:1f.2 Memory controller [0580]: Intel Corporation Device [8086:a2a1]
        Subsystem: Intel Corporation Device [8086:a2a1]
00:1f.4 SMBus [0c05]: Intel Corporation Device [8086:a2a3]
        Subsystem: Intel Corporation Device [8086:a2a3]
        Kernel driver in use: i801_smbus
        Kernel modules: i2c_i801
00:1f.6 Ethernet controller [0200]: Intel Corporation Ethernet Connection (2) I219-V [8086:15b8]
        Subsystem: ASRock Incorporation Ethernet Connection (2) I219-V [1849:15b8]
        Kernel driver in use: e1000e
        Kernel modules: e1000e
01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GP106GL [10de:1c30] (rev a1)
        Subsystem: NVIDIA Corporation GP106GL [10de:11b3]
        Kernel driver in use: vfio-pci
        Kernel modules: nvidiafb, nouveau
01:00.1 Audio device [0403]: NVIDIA Corporation Device [10de:10f1] (rev a1)
        Subsystem: NVIDIA Corporation Device [10de:11b3]
        Kernel driver in use: vfio-pci
        Kernel modules: snd_hda_intel

Any thoughts on if this is even possible? Can the internal GPU handle the noVNC sessions? Thanks!

https://pve.proxmox.com/wiki/Pci_passthrough

 

[dcsapak]

Is it possible to set a setting somewhere to still allow console display in the web browser for guest VMs?

if you want to use the integrated vnc, you have to use a virtual display/gpu
the moment you pass through a gpu, that output is not seen via the integrated vnc

the thing that works is using a gpu as cuda/opencl device only, then you have 2 gpus a virtual one for display and vnc, and the passed through one for computing

 

[agtwashington]

if you want to use the integrated vnc, you have to use a virtual display/gpu
the moment you pass through a gpu, that output is not seen via the integrated vnc

the thing that works is using a gpu as cuda/opencl device only, then you have 2 gpus a virtual one for display and vnc, and the passed through one for computing

I'm not sure what the workload for plex transcoding from one codec to another would be classified as. I don't need the display output of the P2000 GPU at all. It's a computing workload.

That said, I have two GPUs in my system, the Intel one, and the Nvidia P2000.

I don't really care if I can see the console of the VM I have the GPU passed into, I'm more concerned about my other VMs and being able to see their console.

How I would go about implementing what you're referring to?

 

[agtwashington]

So after more poking around, I'm not sure if I can see other Guest VMs consoles, just not the one I'm passing the GPU to. I'm not sure if that is because it's being rendered on the integrated GPU or because it's being done in software. I would be interested in why, but I'm just happy that it works.