2024-02-20, 06:40 PM
(2024-02-20, 01:48 PM)Zachary Drake Wrote: What would be the ideal or recommended hardware for a heavy use case? I want to:
- Have something low power, that I can leave running all the time.
- Stream 4k Video at best possible quality.
- Run 10+ high quality streams at a time.
- Would a powerful GPU be recommended?
- Should it be a standalone machine or a VM?
- Any advantage for JellyFin between Linux, Windows, Mac etc?
- Use DLNA When possible
- Be prepared for JellyFin to Decode/Transcode whenever necessary without stuggling or running into issues.
- Remote access my home network with a VPN and stream even while not at home. Is there an advantage to having a VPN vs a Proxy server setup?
I am not very familiar with Docker, and right now I am running the direct JellyFin install on Ubuntu 22.04 LTS. Any advantage to either method?
And I suppose my main question is what is the most important hardware to a JellyFin deployment? modern CPU? GPU? What should I be looking for?
Low power is going to be pretty hard to achieve with 10+ users. Especially if we are specing the system to assume a worst case of 10+ 4K transcodes with tone mapping. And there are also issues of Internet bandwidth if most of those users are remote.
Here is how I would spec a system for this situation
- CPU/Motherboard - Anything recent-ish. At least 4 cores. Maybe 6, since audio transcodes occur on the CPU.
- GPU - Intel Arc A770 - Has enough VRAM for 8-10 4K transcoded tone mapped streams (tone mapping is very VRAM intensive). Also has AV1 encoding support when that feature is available in JF 10.9
- RAM - 32GB
- OS Storage - 1TB NVMe SSD for OS + jellyfin data + transcode directory
- OS - Debian 12 or Ubuntu 22.04 (must use kernel 6.2 or newer, required for Arc)
If Jellyfin is the only service you are running on this system, just install Jellyfin directly on the OS. No need to add the complexity of docker when you aren't benefiting from what docker brings to the table. I have 5 Jellyfin instances on my server. My "production" jellyfin is installed directly on Ubuntu and the other 4 test instances are in docker.