Hacker News new | ask | show | jobs
by steinvakt2 314 days ago
And flash attention doesn't work on 5090 yet, right? So currently 4090 is probably faster, or?
2 comments
PeterStuer 314 days ago
I don't think the 4090 has native 4bit support, which will probably have a significant impact.
link
diggan 314 days ago
> And flash attention doesn't work on 5090 yet, right?

Flash attention works with GPT-OSS + llama.cpp (tested on 1d72c8418) and other Blackwell card (RTX Pro 6000) so I think it should work on 5090 as well, it's the same architecture after all.

link