As part of my Homelab, I have TrueNAS Community Edition running in a big case with a bunch of Frankenstein parts including two GTX 1080s. TrueNAS recently phased out support for these older cards and I've been having to go through and manually install working drivers, which seem to have gotten overwritten on the last update anyway. Since I'm not actually using anything for video output (and have onboard HDMI if needed). One of the cards I am passing through to a VM (where it has no problem being recognized and used). The other card was used by Truenas for my apps.

Do you have any recommendations for hardware to accelerate tasks like Immich, Jellfin transcoding, media fingerprinting for metadata, OCR on bulk-uploaded pdfs etc.?

I was never really a hardware guy, especially when it comes to self-hosting and servers.

  • I don't want to play games on my server.
  • I don't want to empty my accounts to put a 5090 TI in a server.
  • The above tasks are too slow on CPU alone, so I have to put something in it.
  • I could keep running the 1080s with manually installed drivers, but it's a PITA and I wouldn't mind not doing that.

Are there reasonable options that can take up the workload, but come in at a decent price point? I know with a server, I don't necessarily have to run the same stuff as gamers. Are there options outside of the typical NVIDIA and AMD gaming linups that would work for my purpose, perhaps acquirable via FB marketplace, Craigslist, etc.? Does your server have video acceleration and, if so, what hardware are you using?

I appreciate any guidance the community can give me.

you are viewing a single comment's thread
view the rest of the comments
[–] 1 point 12 hours ago

The 16xx (Turing) are the same chips as the 20xx big without Ray Tracing capability. The Quadro T600 may be found cheaper than the 1660Ti since the latter is still is very popular gaming card. I'm still not 100% sure what you really need a GSP for in your setup and I really don't know why you'd need any gpu to run Home Assistant unless you are running a local LLM as a speech assistant

  • source
  • parent