00:06 mhenning[d]: airlied[d]: I think we've already implemented that, if that's what you're saying
00:11 airlied[d]: oh I should have grepped before asking 🙂
00:11 airlied[d]: indeed we o
00:12 airlied[d]: do
02:41 gfxstrand[d]: There's even a crucible test somewhere
02:56 airlied[d]: I just noticed llama.cpp got code to use it
04:04 gfxstrand[d]: That scares me a little...
17:48 babblebones[d]: Is the likelihood of vulkan realtime compute queue improving any in the near-term? :frog_pray:
18:46 mhenning[d]: I'm not aware of anyone working on it right now
20:18 gfxstrand[d]: babblebones[d]: Once I get back I'm going to work with Ben on better device enumeration and context creation. That'll likely include doing context priorities for real. But it needs kernel work so 6 months to a year at the earliest before it'll be fully upstream.
20:21 babblebones[d]: gfxstrand[d]: Even hearing this is on the agenda within the year is MASSIVE news
20:21 babblebones[d]: Cannot wait for usable VR finally free from the blobs <a:catrave:1205472102963089428>
20:23 gfxstrand[d]: It's possible that priorities will turn out to be a bigger thorn bush than expected but I think they should be pretty straightforward once we agree on the API.
20:24 gfxstrand[d]: It should just be a bit we set in the GSP context state.