Passing a A380 through to Jellyfin App

deegee457

Cadet
Joined
Mar 13, 2024
Messages
4
Hi everyone, I’ve been struggling to set up hardware encoding using my intel A380.

I’ve got a post on Jellyfin forum: https://forum.jellyfin.org/t-transcoding-setup-issues?page=4

And it seems like the GPU isn’t getting passed through to the app. The app has been set up as a custom app (there are images of the setup in the above post but I can add here as well if necessary!).
When I select “Allocate 1 intel gpu” and enable hardware encoding within the app nothing happens and I get an error in the Jellyfin log (again attached in Jellyfin forum but can put here).
I’m running Truenas scale dragonfish (which I believe uses kerbal 6.6 so should be okay with the A380) and the latest Jellyfin docker image.

The guys have been super helpful over in that forum but it seems to have gone beyond them now as it seems more of a Truenas issue.

Any help would be greatly appreciated and if anymore information is needed I’d be happy to supply, I’m very new to all of this so please bare with me if I have to ask what you need!! Lol

My system is:
Intel Xeon E5-1650 v3
64GB DDR4
 

Davvo

MVP
Joined
Jul 12, 2022
Messages
3,222
Are you booting with that GPU? Iirc you cannot passthrough the GPU the system is using.

Also, people are not likely to go check your link to an external forum. Please post here all you can, it gives people less reasons to skip your thread.

Finally, how can you fit and Airbus into a NAS??

Edit: yeah, you totally need to post here more infos. Start from complete hardware list please, and make it clear if you are asking help about the App or the Docker image, as well as all the step you have done.

Hope someone will help, I'm not a SCALE user!
 
Last edited:

deegee457

Cadet
Joined
Mar 13, 2024
Messages
4
Hey - thanks for the reply!

Okay so hardware:
Intel Xeon E5-1650 v3
64GB DDR4
4 x 10tb HGST hdds
1 x 1tv Samsung 989 NVME
1 x 480gb Samsung ssd
1 x nvidia q2000 (used for truenas)
1 x intel A380 (for Jellyfin transcoding)

So I’ve been trying to get transcoding working on my Jellyfin app (from truecharts) for a while and I’ve not had any luck. If I enable hardware encoding in the app and try to watch a video on my phone nothing happens, if I turn hardware encoding off it works fine. I started by upgrading to the Beta as I was told the kernal (6.6) in dragonfish supported intel gpus. This didn’t seem to make much difference other than I could select to allocate 1 intel gpu in the app container settings.

I was then advised in the Jellyfin forum that I should download the docker image as it definitely has the drivers needed whereas the app wasn’t done by Jellyfin themselves and they couldn’t guarantee what changes had been made.

Now I believe TrueNAS doesn’t use docker anymore, but I managed to create a custom app and get it up and running. I did the same thing as with the truecharts app and selected to allocate 1 intel gpu but when I go into the app and select intel qsv and enable hardware encoding it doesn’t work.

I’ve added user 568 to the group render (107) as I was told this might help but unfortunately still nothing. The guys at Jellyfin can’t help anymore as we’re not using straight docker - if Truenas was they could just help with an amended yaml file and it should work but they can’t.

Any help would be appreciated and if you need anything else please let me know!
 

LarsR

Guru
Joined
Oct 23, 2020
Messages
719
Which version of scale do you use? Beecause i think the current scale kernel doesnt support the itel gpu's, or at least not fully. Support will come with the new 6.6 kernel in dragonfish
 

deegee457

Cadet
Joined
Mar 13, 2024
Messages
4
I'm using the Dragonfish beta :) - that was an earlier issue that i was using the stable but that kernal doesn't have the right support so i switched to Dragonfish.
 

Davvo

MVP
Joined
Jul 12, 2022
Messages
3,222
He wrote he's using Dragonfish.

 

deegee457

Cadet
Joined
Mar 13, 2024
Messages
4
I’m sure this is also some how related - I tried to create a Linux VM and pass the GPU through to it and it came up with an issue saying it couldn’t do it because of an IMMOU issue?
 
Top