For the past 25 years of sailing the high seas I’ve always used my PC for watching whatever. But as this is not always practical, I am looking to connect a raspberry Pi to my TV to have a setup with smaller fingerprint and larger screen.
I briefly tried one a couple of eons ago (2010ish?), but sadly I don’t remember the name.
Requirements:
- Must be able to run from a raspberry pi
- Must be able to stream media over my network (protocols aren’t that important as I can probably spin up whatever is needed. Preferably I would just have it index a couple of NFS mounts and local drives)
Bonus question: Which Pi model would you recommend running this? I have a bunch of Zero W, and while everything “works” on them, it simply wasn’t powerful enough to decode video at a watchable rate.
Running the Jellyfin client on the Pi should be no problem. It’s a bit underpowered for running the server, though. If you have only one device, Kodi is probably your best bet. Otherwise any old PC you have lying around will do as a Jellyfin server.
Unless you want to be able to do on-the-fly transcoding for clients that are picky about codecs (e.g. because they’re only fast enough to decode ones with dedicated hardware support). In that case, you’re going to want your Jellyfin server to have a GPU with decent hardware encoding support.
(I learned this the hard way: my Jellyfin server is running in a VM on Proxmox and I haven’t figured out GPU passthrough yet, so some of my media fails to play on clients like Roku because the codec isn’t supported.)
OK maybe not any old PC. But I think pretty much any 8th gen Intel processor or higher will do it without a GPU. I run it on an N100 right now and it works very well for hardware transcoding.
An N100 isn’t “without a GPU;” it’s got Intel UHD Graphics built in – moreover, even though it’s a pretty cheap chip it’s also a pretty new one, and thus supports recent video codecs.
Meanwhile, my AMD Ryzen 5 3600 does not work for transcoding, even though it’s way faster than an N100, because it’s not an APU.
You’ve only proven my point that having a GPU with hardware transcoding support is what makes all the difference.
OK OK, I meant without a discrete GPU.
This is a good point, the client is an important consideration. My setup is the similar, with Jellyfin running in docker in an Ubuntu Proxmox VM (host system CPU is an i7-6700t), but the client is an Nvidia Shield Pro, which so far has been able to handle everything Jellyfin throws at it, with the exception of AV1.
This is a good point, the client is an important consideration. My setup is the similar, with Jellyfin running in docker in an Ubuntu Proxmox VM (host system CPU is an i7-6700t), but the client is an Nvidia Shield Pro, which so far has been able to handle everything Jellyfin throws at it, with the exception of AV1.