Arch Linux

aknarts

Member

Registered: 2013-10-07

Posts: 8

GPU passthrough with integrated GPU

Did anybody tried using the integrated Intel GPU on the guest machine while keeping the more powerful graphics for the host machine?

I did read wiki, googled around a lot and everybody is seemingly using the more powerful graphics for the passthrough. What I would like to to is do the opposite, keep the nvidia graphics for the physical machine and from time to time get some "native" performance from the Intel GPU in the virtual machine.

If it is not possible I will just buy second card, no big. Just wondering if this is even supported.

R00KIE

Forum Fellow

From: Between a computer and a chair

Registered: 2008-09-14

Posts: 4,734

Re: GPU passthrough with integrated GPU

The important question is to which card are the video outputs connected? Is it a laptop or desktop machine? You will have to provide more details otherwise people can't help you.

---

R00KIE
Tm90aGluZyB0byBzZWUgaGVyZSwgbW92ZSBhbG9uZy4K

aknarts

Member

Registered: 2013-10-07

Posts: 8

Re: GPU passthrough with integrated GPU

Ok, will do.

Desktop machine, two GPUs(technically)
1. Nvidia 980TI - main and only output to the screen(also in UEFI enabled as the default output so used from boot to shutdown)
2. Intel Skylake i7-6700 GPU - to be honest I do not see it in any output so cannot provide any information, but I do know it is there and functional as I used it during installation.

Want to use 2(Intel) as a dedicated graphics card for a guest VM machine while keeping the main output on the host through 1(Nvidia)

And while writing this figured out why I do not see the intel GPU anywhere:

/etc/modprobe.d/blacklist.conf

install i915 /usr/bin/false
install intel_agp /usr/bin/false

R00KIE

Forum Fellow

From: Between a computer and a chair

Registered: 2008-09-14

Posts: 4,734

Re: GPU passthrough with integrated GPU

In this case I suppose you might be able to do what you want, if you haven't done so yet do refer to the wiki page about passthrough [1]. There are a few conditions that must be met when you want to use passthrough, if I'm remembering correctly I've seen people come to the conclusion that in their case some of the needed features where disabled by the system's bios/firmware.

As a side note, the video output(s) of your integrated gpu, if there are any, should be on the motherboard near all the other stuff (usb/audio ports). Also your integrated gpu should still show up in the output of lspci, it doesn't disappear from the pci bus just because you don't load the driver.

[1] https://wiki.archlinux.org/index.php/PC … h_via_OVMF

---

R00KIE
Tm90aGluZyB0byBzZWUgaGVyZSwgbW92ZSBhbG9uZy4K