Hey there,
I’ve been trying to get the intel arc GPU working on almalinux, either for passthrough to a podman container or passthrough to a VM, and it just simply doesn’t work. When passing it through to a VM it just freezes up the VM and won’t boot. When trying to pass it through to a container, it sees the device but can’t use it. Seems like some capabilities for MSE are disabled in the kernel config that may be required to get this working?
Howdy tbelway
Can you please tell me the almalinux version in the host?
I only found nvidia gpu info in podman docs GPU container access | Podman Desktop
can you see the intel gpu in the base host right ?
which commands did you run? can you paste it input and output
Cheers
I’m afraid I currently can’t test it atm but should be able to again in a few weeks time. That hypervisor is currently running a dev environment that is in use…
But I tried it in:
Almalinux Kitten
Almalinux 9.5, 9.4, 9.3, 9.2, 9.1, 9.0
Almalinux 8.10, 8.9
The OS always SEES the card but can never use it. As in it’s present in /dev/dri/renderD128 or renderD129 (depending on the config at the time), but it can’t be used. I was able to get it working when I tested proxmox, but I prefer to not use debian based for this hypervisor.
That being said, I have easily managed to get both an AMD card and an NVIDIA card running HWA on the same hypervisor.
As I don’t have it operational atm, I am unable to paste the commands, but from all the troubleshooting I did it looked to be an issue passing through due to the MEI parameters not being enabled in the kernel config.
Hey tbelway
interesting. It will be nice to see the podman version too.
I can’t find links of podman with intel gpu arc passthrough.
In this link the user connected the intel gpu with lxc/podman in proxmox
https://fyksen.me/gpu-lxc-podman-jellyfin-passtrough/
It has to be a way. maybe the intel gpu was to new back then.
I’m curious to know what do you see in alma 9 or alma 10 .
Please report back when you have the host free of use.
Cheers