00:22redsheep[d]: Ok, yeah makes sense. I haven't tested yet with no_cbuf, maybe that helps a little
06:12tiredchiku[d]: took me embarrassingly long to figure it out
06:12tiredchiku[d]: but I got it fixed
08:55tiredchiku[d]: and another https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31518
09:16tiredchiku[d]: entirely possible I'm not doing something important there, am open to feedback
09:32valentineburley[d]: tiredchiku[d]: that CTS just tests if all the binds you report have an unbind reported too, and it passes for 0 of each too 😄
09:33tiredchiku[d]: yeah, I did get told that on LGD too
09:33tiredchiku[d]: which is why I marked it as draft
09:33tiredchiku[d]: gonna be on vacation for a week starting tomorrow, I'll get it up and going after I get back :)
18:14gfxstrand[d]: Sweet! I'll take a look after a bit. I'm working on my XDC talks right now
18:18tiredchiku[d]: :3all the best for that :D sadly I won't be able to watch it live because vacation, but I'll catch up on it when I can
18:27karolherbst[d]: "working on my XDC talks" yeah... though I only filed for a lightning talk 🙃 Which is weird, because I actually would like to talk about SVM
18:28karolherbst[d]: I should finish my slides, but like...
18:33redsheep[d]: Looking though the talks, Mary has one on mesh shaders in nvk 👀
18:33redsheep[d]: That's exciting
18:35karolherbst[d]: most hyped talk
18:35tiredchiku[d]: I'm _really_ curious to see what nvidia's lightning talk is gonna be
18:35tiredchiku[d]: what will they possibly cover in 5 minutes
18:37redsheep[d]: Oh and a Windows talk from Faith. Yeah xdc has like 3x more interesting talks than I expected
18:54tiredchiku[d]: gonna miss them :o
19:02gfxstrand[d]: Mine are going to be pre-recorded since I'm not going to be there in person this year
19:03gfxstrand[d]: So if you watch online you'll get the same experience as everyone else
19:31marysaka[d]: redsheep[d]: Hope everyone will love the 15 min of curseness :linatehe: `
19:32redsheep[d]: marysaka[d]: I haven't watched the MR recently, did you ever find a way around the max invocations thing?
19:33marysaka[d]: redsheep[d]: Well I had some plan with Faith on how to handle it but it's extremely hard to materialize with NIR sadly
19:34marysaka[d]: (And I'm working on coop matrix for now, will come back to it after that probably)
20:43djdeath3483[d]: gfxstrand[d]: I didn't know that was a possibility
21:04gfxstrand[d]: djdeath3483[d]: Pre-recording? Or presenting virtually?
21:04djdeath3483[d]: both 🙂
21:04gfxstrand[d]: We've done virtual presentations ever since COVID and never really stopped
21:05gfxstrand[d]: Pre-recording... IDK. Mark wanted us Collabora folks to pre-record and then Guy is going to edit everything together and make it look fancy. A few years ago, Microsoft did a couple talks that way. It's kinda nice because it lets the presenter relax and chat with people in matrix during the talk.
21:07gfxstrand[d]: But I think presenting via jitsy is still possible
21:32Lyude: skeggsb9778[d], airlied[d] - if either of your two are around, fifo would be the engine that would send an interrupt when a fence completes right?
21:32Lyude: *you two
22:48tiredchiku[d]: gfxstrand[d]: just saw your comments on the MR, I'll address them next weekend (after vacation)
22:48tiredchiku[d]: thanks for taking a look though!
22:59gfxstrand[d]: yw
23:03skeggsb9778[d]: Lyude: Yes. Nouveau uses a host (~aka. fifo) semaphore release (to wait for idle + write a sequence number), followed by a "non-stall" interrupt to notify the driver as they're hit