I’m taking down my big Supermicro server to save energy and moving Plex/Jellyfin/*arr to a spare 10th gen Intel NUC with SSDs. Performance is fine for DirectPlay media to my SHIELD and mobile devices, but the onboard GPU power is limited and struggles to even transcode some 1080p media – let alone 4K. Does anyone have experience using eGPUs in a Thunderbolt chassis with a NUC and can you share what worked or didn’t work for you in terms of hardware and configuration?
Edit: this is an i7-10710U with NVMe storage and 32GB of RAM, running Windows 10 with all the latest drivers directly from Intel.
Edit 2.0: (a year later, which nobody will ever read) I switched to Linux and all my issues disappeared. Really.
You must log in or register to comment.
The onboard gpu is likely more powerful than all but some workstation gpus you could add for transcoding, it’s more likely you don’t have hardware acceleration working properly.
That’s good to know. Other than enabling hardware acceleration and setting the transcoding quality, what else can I try tweaking in Plex?
I think you can check logs, but iirc you need x11 running for it to all work.
Also install vainfo and see what it says.
Iirc arch has a page with information on vaapi which might include details on how you enable plex.
I edited the post to mention that this is Windows 10, but I’m not at all against installing Arch instead if it improves performance.
Oh wtf, hmm, I don’t know, let me check online, never tried on windows.
https://support.plex.tv/articles/115002178853-using-hardware-accelerated-streaming/
Take a look, at the bottom there’s a way of seeing if it’s used.
You should be golden, you should be able to handle it trivially, especially with your chipset.
something is definitely wrong with the config if its failing at a single 1080p, I did a plex server test on an intel n95 nuc (one of the lowest end cpus in their current range) and it blasted through multiple 4k HDR transcodes simultaneously.
I agree something isn’t right, I have Plex on an HP Chromebook G2 with a Celeron 3865U, it’s a 1.8ghz dual core without HT, and I had it doing like 15-20 1080p streams during testing. Quick sync is amazing.
Does the n95 handle 4k to 4k Plex transcodes well? Or are you talking about 4k to 1080p?
It’s fine, it can do a few at once. I didn’t do a lot of testing since I never have to transcode these days anyway.
Okay then there’s no reason why a hyperthreaded hexacore i7-10710U should struggle unless, like you said, there’s a misconfiguration. That makes me feel better because I’m hemorrhaging cash at the moment.
That’s encouraging. In Plex, what did you configure other than to set the transcoding quality and enable acceleration?
your problem might be windows.
I think everyone here might be nerd enough to be assuming everyone is running Linux.
And it’s fine if that’s the case. It’s not like I actually needed another reason to finally start ditching Windows for good but this might just be the straw that broke the camel’s back. I’m just a lazy Windows sysadmin who uses what he knows, but not if it doesn’t make sense for the application.
Your CPU should be more than enough: link
Does AMD have something like this? I have a Ryzen 5 small PC with proxmox and I wanted to use Jellyfin and transcode video
The only thing I could find was this but looking further looks like jellyfin supports only AMD GPUs for HW transcoding. Give it as try, it still might be enough for you
Not true. It also does intel; https://jellyfin.org/docs/general/administration/hardware-acceleration/intel/. I got an nvidia card working with hacky terminal workaround as well. it’s been rock solid for almost a year now.
The person asked about a AMD CPU with no further Infos…
If you tell us exactly what hardware, we could give you better answers.
If you want to avoid transcoding, you could configure your stack to filter out unsupported codecs, or pre-convert them.
Can you elaborate on pre converting them? Do you mean with an outside application like handbrake or an automated one in the arr stack?
There’s tdarr, unmanic and fileflows as the popular choices.
Personally I use fileflows as it is extremely customisable and you can set up quite complex rules for how and what to convert.
I appreciate your reply! Thanks!
There are -arr programs for it, or you could script something. Plex has it built in.
