News for hackers. Stuff that matters.

A comment on Llama.cpp: Full CUDA GPU Acceleration. Parent comment

Re:Llama.cpp: Full CUDA GPU Acceleration by born-jre on Tuesday June 13, 2023 @03:57AM (#36307033)
short answer is they have somewhat competent hardware but software sucks or you can watch george hotz rant about how amd driver sucks

https://www.youtube.com/watch?v=Mr0rWJhv9jU

[ Parent | on HN ]
Re:Llama.cpp: Full CUDA GPU Acceleration by alecco on Tuesday June 13, 2023 @04:04AM (#36307086)
He got a tarball fix for the driver after his rant got viral. Still not looking good IMHO.

https://geohot.github.io/blog/jekyll/update/2023/06/07/a-div...

[ Parent | on HN ]
Re:Llama.cpp: Full CUDA GPU Acceleration by xiphias2 on Tuesday June 13, 2023 @05:40AM (#36307778)
It looks like some great engineers inside the company fighting the burocracy.
[ Parent | on HN ]
Re:Llama.cpp: Full CUDA GPU Acceleration by jandrese on Tuesday June 13, 2023 @12:23PM (#36312550)
> So it fixed the main reported issue! Sadly, they asked me not to distribute it, and gave no more details on what the issue is.

I think they missed the thrust of the rant.

[ Parent | on HN ]
Re:Llama.cpp: Full CUDA GPU Acceleration by HellDunkel on Tuesday June 13, 2023 @07:28AM (#36308569)
he goes from building a ml rack out of ati graphics cards to coding some python to recommending reading the unabomber manifesto, from marx to saying he owns a rolls royce... lord, please have mercy!
[ Parent | on HN ]