How to use Intel Arc A310 when Jellyfin is installed an app in TrueNAS Scale

[mehran]

Hi,

I have installed Jellyfin 10.10.6 in TrueNAS Scale (Electric Eel 24.10) as a docker with the option "Passthrough available (non-NVIDIA) GPUs" checked. And within Jellyfin, I went to the Dashboard -> Playback -> Transcoding and selected QSV with path ˋ/dev/dri/renderD128ˋ. But when I play a video on my computer, it still says Direct Playback. And my tablet with smaller capacity cannot play large videos at all.

Can someone please first confirm that this setup should be able to utilize Intel Arc A310? And second, if it is possible, how? What am I doing wrong?

Thanks

[Efficient_Good_5784]

It doesn't sound like you're doing anything wrong. Direct Playback just means that no transcoding is being done due to the client 100% understanding the original file.

If you want to force a transcode, you need to lower the bitrate to be lower than the video's original bitrate.

Also, the capacity of your tablet is irrelevant to streaming with Jellyfin. The client will not be storing the video.
At best, it will only be temporarily storing small chunks at a time for the cache/buffer until you exit the player.

[mehran]

Thanks for the reply and sorry for the confusion. When I said capacity, I meant processing capacity. In other words, the tablet struggles to play the video. There are other videos it cannot play at all (and of course there are those it can but there's no point taking about them). In my mind, it's the exact scenario for the transcoding to kick in and yet it doesn't.

[mehran]

Here's a more concrete example where the GPU should be used but it isn't. I have a video encoded with AV1 but my tablet does not play AV1 and I was expecting Jellyfin to transcode that I into another codec that my tablet does play but this does not happen. Isn't this how transcoding works?

Also, I can confirm that the hardware is properly setup and working since handbrake app on the same platform (TrueNAS Scale) can leverage it to encoded. This is verified by seeing hardware utilization on intel_gpu_top when I encoded a video.

[Efficient_Good_5784]

You need to make sure your user doesn't have transcoding privileges disabled in the user settings within the dashboard.