Mastodon Feed
dysfun@treehouse.systems ("gaytabase") wrote:
i have gotten llama.cpp to build, hallelujah. in theory it is now optimised for this machine, thus giving me the best chance at getting useful performance out of it.
tomorrow i'll build a docker image with it. i was reminded that i don't have a new enough graphics card for passing through my graphics card while using it to render my desktop, so virtualisation with pci passthrough is out.
then i'll look at some of the things to plug into it. can't wait 😬