08:45 ayaka_: about v6 Add pixel formats used in Synatpics SoC, is its document the problem left? Any possible we could put the pixel format header part into pending list
08:45 ayaka_: then I could just need to send a v7 of the document
14:59 MrCooper: really sad that apps breaking if GL fails to churn out tens of frames per second is still considered state of the art in 2023 :(
15:01 daniels: MrCooper: yeah, it is somewhat disappointing, but otoh the entirety of EGL is hugely disappointing full stop, and in hindsight it was a failure to not have explicit signaling like ActivityManager to let apps dtrt, but instead to rely on implicit actions
15:03 MrCooper: definitely not how I imagined 2023 when I started working on GPU drivers around the millennium
15:05 daniels: don't worry, we'll still have shit clients in 2043 as well. some of them will still be Motif :)
16:17 bcheng: Is it possible to profile a compute only vk app with the RADV SQTT capturing features?
17:16 ishitatsuyuki: unfortunately, the triggers only work if you call vkQueuePresent
17:17 ishitatsuyuki: very ugly, but if you can make the app create a window and present after every iteration of computation, you might be able to get it to work
17:23 pixelcluster: there is an even more ugly option
17:24 pixelcluster: you can do some hacks to get around the requirement of making a window if you don't have a display server running
17:24 pixelcluster: but if that's not a problem it's probably better to just create a dummy window
17:56 DavidHeidelberg[m]: robclark: btw. talking about asan, it catched some minor issues in the fd/tools: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22015 if you have moment to check it make sense
17:59 robclark: r-b
18:00 DavidHeidelberg[m]: :) thx, way to new ASAN in Debian bookworm! 🦄
18:40 bcheng: quite ugly haha, thanks i'll give it a try
18:41 bcheng: i assume amdvlk also has this limitation? I know wrong chat and all but since RGP is an AMD thing figured you guys might know