I have a Ugreen DXP4800 Plus so I have Intel quicksync but how do I enable it? Thanks for the help
7 replies
Chris wrote:
Two steps. First, in your SparkBox dashboard go to Settings - Media, set Hardware Transcoding to "intel" and save -- that passes /dev/dri into Jellyfin automatically and restarts it for you, no SSH needed (the DXP4800's N100 is fully supported, including AV1). Then inside Jellyfin itself: Dashboard - Playback - Transcoding, set Hardware acceleration to VAAPI, VA-API Device to /dev/dri/renderD128, enable H.264/HEVC/VP9/AV1 decode + H.264/HEVC encode, and Save. Play something and force a lower quality to trigger a transcode -- Jellyfin's dashboard should show "(HW)" next to the session if it worked.
Chris wrote:
That's actually a good sign in one way — hardware transcoding is working, and Dolby Vision is exactly the kind of file it tends to trip on. DV Profile 8.1 carries an HDR10-compatible base layer, but some hardware paths choke trying to handle the Dolby Vision layer itself rather than falling back cleanly, which shows up as a hard failure instead of just slow playback. Quick test: in Jellyfin's Dashboard - Playback - Transcoding, temporarily set Hardware acceleration back to None and try that same file — if it plays fine in software, that confirms it's the DV + hardware combo, and the fix is just leaving that title on software transcoding rather than forcing VAAPI on it.
Jonas3108 wrote:
I did that but now if I try to play something it says there was a fatal Playback error. Im trying to play this very high quality file 4K HEVC Dolby Vision Profile 8.1 (HDR10)
Jonas3108 wrote:
I mean it does play on 10mbps, it's just really slow and stuttery. If I go above those 10mbps it just takes forever to load. Also in docker it says that hardware acceleration is disabled for sb-jellyfin-media.
Chris wrote:
Good detail — that changes the read. If Jellyfin's own logs say hardware acceleration is disabled, the device isn't being used at all, which explains stuttering even at a modest bitrate (that's software decode, and the N100 struggles there). Usual culprit: a group ID mismatch. Our overlay grants /dev/dri via GIDs 44 (video) and 105 (render) by default, but some NAS builds use different numbers. Run stat -c %g /dev/dri/card0 and stat -c %g /dev/dri/renderD128 on the host — if either doesn't match 44/105, set JELLYFINVIDEOGID and JELLYFINRENDERGID in Settings - Media and save; it'll recreate the container for you.
pereze94 wrote:
I know on my old gaming pc that has a i5-9400f and gtx 1660 I had to activate hardware trancoding by going into the system through ssh and manually editing the evm file. As none of the gui setting were sticking. Only found out when I went into the logs how you did and it showed it was off. I used the AI for help doing it as I'm barely learning/getting into home servers.
Chris wrote:
Good data point, thanks for sharing! That shouldn't happen -- saving Hardware Transcoding in Settings - Media is supposed to update everything for you, no SSH or manual file editing needed. If your current box is doing the same thing (setting doesn't stick after you save), run sudo sparkbox doctor and check what it shows under Media Storage/Hardware Transcoding. If it says off even after a save, that's worth flagging so Tom can look into why the save isn't taking.