2023-09-04, 07:59 PM
You have a few options. I mean for like $400 total you can get a used OptiPlex 5090 Micro: https://www.ebay.com/itm/362460534458. It has an i5-11500T with a TDP of 35W and a UHD 750 though I don't know how many simultaneous transcodes that would translate to...the iGPU is capable, but the CPU might be underpowered for five at a time.
Granted, you have limited capabilities to expand on this guy, I'm not even sure if you can replace the RAM (you did say 32 GB). Just want to make sure you've thought about options. One of my servers is an old OptiPlex SFF PC that is a workhorse and I'll be damned if it's not awesome. I'm lucky enough to have a source to get these old machines free on occasion (my desktop is a very old OptiPlex 9020 with an i7-4770).
What you should be looking for is definitely the iGPU, not the processing power of the CPU. @TheDreadPirate has the right idea with building around a capable iGPU. Just know that you'll likely be limited in what other applications you can push on your server before you start experiencing latency or other downstream effects. I went the other route and now have a pretty beefy rig and, while I don't regret it, nothing I do taxes it whatsoever and it's still perfectly responsive even when I'm in the depths of running weeks-long encoding jobs while still having ~22 (micro-)services running. It's great, but it's insane overkill for a media server.
That little $400 box probably would've been more than enough.
Granted, you have limited capabilities to expand on this guy, I'm not even sure if you can replace the RAM (you did say 32 GB). Just want to make sure you've thought about options. One of my servers is an old OptiPlex SFF PC that is a workhorse and I'll be damned if it's not awesome. I'm lucky enough to have a source to get these old machines free on occasion (my desktop is a very old OptiPlex 9020 with an i7-4770).
What you should be looking for is definitely the iGPU, not the processing power of the CPU. @TheDreadPirate has the right idea with building around a capable iGPU. Just know that you'll likely be limited in what other applications you can push on your server before you start experiencing latency or other downstream effects. I went the other route and now have a pretty beefy rig and, while I don't regret it, nothing I do taxes it whatsoever and it's still perfectly responsive even when I'm in the depths of running weeks-long encoding jobs while still having ~22 (micro-)services running. It's great, but it's insane overkill for a media server.
That little $400 box probably would've been more than enough.
Jellyfin 10.10.7 LSIO Docker | Ubuntu 24.04 LTS | i7-13700K | Arc A380 6 GB | 64 GB RAM | 79 TB Storage
![[Image: AIL4fc84QG6uSnTDEZiCCtosg7uAA8x9j1myFaFs...qL0Q=w2400]](https://lh3.googleusercontent.com/pw/AIL4fc84QG6uSnTDEZiCCtosg7uAA8x9j1myFaFsx_0wM5mKgTleZb466r4NtOHIRePq_CxyhsB72qsd-xuYFkqCGogLltAMN62BXsm-yFsSZcDuryqL0Q=w2400)