Server Version#: linuxserver/latest Version 4.140.0 < docker image
In transcoder I get “Unknown Intel” and transcoding doesn’t seem to work. Im trying to move off my Arc A310 to the Intel 15th Gen iGPU or 265k Ultra. Just note other docker applications that I use for tv and cameras are now working find on this iGPU. Plex is the only one im having trouble with. linuxserver uses Ubuntu Nobel, kernel 6.12 24.04 LTS
I will submit to the transcoder Engineer to check the PCIE ID codes.
Strangely, we did just get an updated list from Intel. Curious why that’s not included.
The CPU was released only 6 weeks ago.
Given the HEVC effort in progress, it’s understandable he hasn’t gotten to adding this yet.
I do know the Ultra 265K does work.
Please do review your docker config.
Also, if possible, spin up a native ‘test’ server to confirm it’s not a configuration issue. ( A lot of folks have had docker config issues recently )
Hello,
Im not sure I can do a native host. i will say the following, if it helps:
I really didn’t have to change anything to get the other containers to work when I updated the Truenas kernel. renderD128 was already avaialbe for ARC so my configs are all the same, and with the kernel update was available for iGPU to pass through to docker device as it was now being seen by truenas in /dev/dri. They all run on the same VM except now I’m passing the iGPU instead of the ARC iommu. The other containers transcode just fine, plex has stopped being able to transcode.
i originally was using pmsdocker but they had an old kernel and was based on 20.04. I switched to linuxserver as it is built on 24.04 which I thought would be better.
I did see this error deep in the logs:
TPU: hardware transcoding: enabled, but no hardware decode accelerator found
The permissions seem right on docker for the uuid and guid for the /dev/dri device permissions, but again I didn’t have to do any perm changes with I was passing the ARC device.
**** adding /dev/dri/card0 to video group video with id 44 ****
**** permissions for /dev/dri/card1 are good ****
**** creating video group videoka62 with id 107 ****
**** adding /dev/dri/renderD128 to video group videoka62 with id 107 ****
Docker is used for versioning skip update check
[custom-init] No custom files found, skipping...
Starting Plex Media Server. . . (you can ignore the libusb_init error)
Connection to localhost (127.0.0.1) 32400 port [tcp/*] succeeded!
I also tried to set the HardwarePath directly in the Preferences.xml but that didn’t seem to trick it either also
If there is anything I can do ill try and help
if I may recap?
So far so good ?
Ill try to clear it up some:
where all 3 docker containers worked fine with the ARC card
You have only TrueNAS Scale on the 6.12 kernel?
Did you know you can take Ubuntu 24.04 and upgrade it to 6.12 ?
I’m thinking / suggesting based on what I see and hoping to find an easier way for you while solving a problem (which is likely cased by all those layers)
My ProxMox support brick is all LXC . Everything works perfectly. (tteck config)
I also have a Ubuntu 24.04 VM with the iGPU passed in.
(the end result would be separation of TrueNAS (the storage element) from the other elements; allowing them to run as peers and not layers on top of each other.
Q: What do you get for lspci and lshw -C display so I can track against what I have here
yeh my other portainer VMs run that currently.
I dont think the layers are the issue. Ive run this way for years. ARC works fine with these layers. I am trying to avoid using NFS shares as much as possible. I may move back to an Ubuntu VM to test but it breaks the other stuff I need local storage paths for ideally. I can’t recall what happened when I tried iGPU about a month ago on the Ubuntu 24.04. LTS. Lots of stuff didn’t work. Maybe I’ll give that.a go tonight.
I use VMs and LXCs all the time.
I mostly use LXC now (it’s easier and everything is on the same host kernel)
(NFS via the internal stack is blazing fast – as fast as loopback)
I really dont want to debate LXC vs VM. lol. We all have our reasons. I’ll never use LXC.
fair enough. Just trying to help you solve.
So what do you get with the lspci and lshw ?
everything pass through as expected ?
The iGPU should always be on 00:02.0
[chuck@lizum ~.2004]$ sudo lspci
00:00.0 Host bridge: Intel Corporation Device 4660 (rev 02)
00:01.0 PCI bridge: Intel Corporation 12th Gen Core Processor PCI Express x16 Controller #1 (rev 02)
00:02.0 Display controller: Intel Corporation AlderLake-S GT1 (rev 0c)
Just found this.
The Intel Arrow Lake Linux graphics driver support appears largely wrapped up following a patch for properly handling the necessary GSC firmware requirements and building off all the existing Meteor Lake Arc Graphics driver code paths. There are a number of Arrow Lake PCI device IDs already present for the graphics while a new one is being added now to the kernel drivers.
The Intel Linux kernel graphics driver has already supported Arrow Lake graphics PCI device IDs of 0x7D41, 0x7D51, 0x7D67, and 0x7DD1 but now another one is being added: 0xB640.
The patch simply adds the additional device ID for Arrow Lake without any context for this new device ID. It would appear though that the Arrow Lake family may be slightly larger than originally planned given this late addition of another device ID.
I’ll chat with the engineer and confirm the device ID status.
If not added, it might be until January before it gets added (most of us are taking forced holiday time)
cool thanks
the device is indeed 00:02.0. Ive verified that the renderD128 device is the iGPU. Im not passing the arc card at all, which is the goal. Again the other containers are using the iGPU same device just fine so I think that’s pretty sorted. Appreciate the assist and time.
Is there any way to verify what IDs mean which CPUs now? I would assume the k series CPUs are the ones added already and the new ones are their lower TDP CPUs coming early next year I believe. Will that solve the Unknown Intel device in the transcode selection box?
so my CPU is 7D67
Intel® Graphics
Xe-LPG
Arrow Lake
Kernel: 6.8* < * This driver requires the force-probe kernel parameter.
64/48/32
per Supported Hardware — Intel® software for general purpose GPU capabilities documentation
claims support since kernel 6.8 i guess? which just isn’t true from my experience. I’ve only been able to use it with 6.11 kernels or newer.
00:02.0 VGA compatible controller [0300]: Intel Corporation Arrow Lake-U [Intel Graphics] [8086:7d67] (rev 06)
maybe i’ll switch back to the ARC card for everything ahd try on ubuntu 24.04 again. If I recall I had to use 24.10 to get some support but can’t remember what worked and what didn’t on that VM
So I moved plex to the 24.10 Ubuntu vm running docker, with 6.12 kernel and I get the same behavior. Can’t get transcode to work. Switch the device to the ARC card and keep everything else the same, and it works fine.
Another update. I deciced to install the dpkg of plex on the ubuntu 24.10 kernel 6.12 host and it can detect the Arrow Lake iGPU, where the docker versions say “Unknown Intel”.
So for some reason the Docker images for both linuxserver and pmsdocker have issues, at least for me. to show the Arrow Lake iGPU correctly. Again, ARC is dected fine, the only variable is the iommu change to iGPU vs ARC and docker. Intel 14500 docker was fine also.
Another update. I rebuilt the pmsdocker plexinc image using 24.10 and this works showing the Arrow Lake iGPU in the transcode section, and transcoding now works in docker.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
