Ollama & ARC GPU

Jackal65

Expert Member
Joined
Feb 10, 2024
Messages
2,353
Reaction score
2,546
Running Ollama with just a CPU is very slow and training an AI model becomes a nightmare. Specially if your CPU is as old as mine. Now a GPU should speed things up. Thing is I got myself an ARC GPU and for all the pain and suffering this GPU has given me I would love to just rip it out and turn it into slag. But I am not buying an upgrade anytime soon.

My question can I somehow get Ollama to run with ARC? If so and it is on Github can someone please tell me how to compile I am new to this stuff.

My goal here is to take all my Diary docs and give it to the AI to process. If am no more my kids will be able to get to know me a bit better. Or that is the hope. Stupid I know but I have Diary docs dating back to 1990s. I figured they are not going to read all that so Training an AI might be a good way for them to interact with it all. I am using llama2-uncensored figured it is a good place to start.
 
I really doubt it’s going to run on that.
 
Sounds like it may work (it was just added last month).
 
Installed Linux Ubantu but I was to stupid to get the intel driver installed and didn't understand the instructions very well. It was this massive document. So I pulled the plug on the Ollama AI idea. Figured if they wanted to support Intel properly they would have. But I will not buy any Intel product anymore they are the worst.
 
Grabbed a GPU on Carbonite, will see how this goes when I get it. But Intel can go **** itself.
 
Replaced the GPU a few days ago, the AI is working really well now. Very fast.
 
Top
Sign up to the MyBroadband newsletter
X