ebay server graphics card for self-hosted AI (ollama). 8GB Tesla P4 from 2016. $120. It takes one pcie x16 slot. Small enough to fit in a 1U server. Should be enough to get started.

Reply to this note

Please Login to reply.

Discussion

ollama is really sick. It looks like ChatGPT's UI.

That's with Open WebUI, right?

Looks like that.

It has a simple API as well to develop against: https://github.com/ollama/ollama/blob/main/docs/api.md

8GB is not much. Do you have reason for using old tesla rather then current gen consumer GPU?

I use amd's rx 7700xt and it's look like weak point is 12GB memory and not chip itself. At least for llama

Can't fit anything bigger in the 1U case. Planning to get a 2U server dedicated to AI so I can use bigger cards. Also these are 8GB, but they're small (1 pcie slot), so you could fit 4 in a single machine.

Oh that explain it. I use my gaming rig also for ollama so server use is not come to my mind 🀣