#asahi-gpu · 2022-12-15 20:59
alyssa2. Read from write-combine memory (i.e. uncached) from the CPU. The framebuffer, as well as the staging resource used with the blit, are both write-combine which makes readback from the CPU very slow.#asahi-gpu · 2022-12-15 20:58
alyssa Solution: prefer blit-based texture transfer. Most texture transfers will use a staging blit internally anyway for compression, or worse tile on the CPU, so this is probably good in practice.#asahi-gpu · 2022-12-15 20:57
alyssa1. Format conversion. Because we don't prefer blit-based texture transfer*, mesa/st falls back to software conversion. This would be more efficiently done on the GPU, especially once we have framebuffer compression with the DCP (we don't yet).#asahi-gpu · 2022-12-15 20:56
alyssaThere are three slowdowns here:#asahi-gpu · 2022-12-15 20:56
alyssahandlerug: Okay, I've looked at the trace and reproduced similar issues locally#asahi-gpu · 2022-12-15 20:23
handlerugsent in private messages#asahi-gpu · 2022-12-15 20:17
alyssa(You can preview it with `sysprof screenshot.trace`)#asahi-gpu · 2022-12-15 20:17
alyssaCan you install `sysprof`, run `sysprof-cli screenshot.trace`, run grim (with Celeste or foot), ctrl-c out of `sysprof-cli` and send me the resulting screenshot.trace file?#asahi-gpu · 2022-12-15 20:16
alyssahandlerug: yes that's indeed excessive#asahi-gpu · 2022-12-15 19:55
handlerugthat's when screenshotting the game Celeste. when screenshotting the foot terminal emulator (which uses CPU for rendering), grim takes 1.7s, but it's still quite a lot imo#asahi-gpu · 2022-12-15 19:53
handlerugalyssa: "grim 1.00s user 0.01s system 36% cpu 2.787 total", and sway at 100% CPU#asahi-gpu · 2022-12-15 19:53
alyssaapparently I do have interesting things in agx/next that I forgot about#asahi-gpu · 2022-12-15 19:53
alyssaand 2a094ba5b8d ("asahi,agx,nir: Implement depth and stencil export") for good measure#asahi-gpu · 2022-12-15 19:52
alyssaand e643d9b6937 ("asahi: Identify XML for more flatshading controls")#asahi-gpu · 2022-12-15 19:52
alyssaI thought that was already merged upstream >.>#asahi-gpu · 2022-12-15 19:52
alyssathat'll probably fix Darwinia#asahi-gpu · 2022-12-15 19:52
alyssalina: asahi/mesa needs f5dd101ab69 ("asahi: Set flatshading controls appropriately")#asahi-gpu · 2022-12-15 19:51
alyssalina: shit!#asahi-gpu · 2022-12-15 19:51
alyssaI don't think I have anything interesting in agx/next right now#asahi-gpu · 2022-12-15 19:50
handleruglemme ssh in#asahi-gpu · 2022-12-15 19:50
alyssaahah right#asahi-gpu · 2022-12-15 19:50
handlerugthe compositor freezes so I'm not sure I will be able to see what top shows :P#asahi-gpu · 2022-12-15 19:50
alyssa100% cpu usage? 3% cpu? very different implications :p#asahi-gpu · 2022-12-15 19:50
alyssaWhat does top show during that 4s?#asahi-gpu · 2022-12-15 19:50
handlerughuh. my machine is MacBook M1 Max 16", same grim + sway setup#asahi-gpu · 2022-12-15 19:50
alyssa(granted this on a release build of agx/next but it should be similar on what lina ships)#asahi-gpu · 2022-12-15 19:50
alyssaon an M1 Mini at 4K, `grim` in sway takes 0.34s#asahi-gpu · 2022-12-15 19:49
alyssahandlerug: Implemented, yes. Optimized, no. But 4 seconds seems... excessive.#asahi-gpu · 2022-12-15 10:51
handlerughas glReadPixels been implemented or optimized? just wondering because I believe wlr-screencopy-unstable-v1 uses glReadPixels, and for some reason screenshots take around 4 seconds#asahi-gpu · 2022-12-15 08:05
linaYay! ^^#asahi-gpu · 2022-12-14 21:27
Tramtristgreat news yuka ! #asahi-gpu · 2022-12-14 21:25
yukathe last update fixed all the crashes in sway! this improves my experience so much, thank you <3#asahi-gpu · 2022-12-14 06:19
austenNow, can I train ml models pretty well?#asahi-gpu · 2022-12-13 22:31
eric_engestromhttps://gitlab.freedesktop.org/mesa/mesa/-/issues/7890#asahi-gpu · 2022-12-13 22:31
eric_engestromalyssa: ended up writing an issue listing the things I could think of; feel free to edit :)#asahi-gpu · 2022-12-13 21:36
alyssa:)#asahi-gpu · 2022-12-13 21:11
iaguisthanks!#asahi-gpu · 2022-12-13 20:49
alyssadisabled MRs on asahi/mesa to hopefully avoid that UX fail#asahi-gpu · 2022-12-13 20:25
alyssa:)#asahi-gpu · 2022-12-13 20:20
iaguisdone: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20306#asahi-gpu · 2022-12-13 20:06
alyssaheh, meta task, writing up docs on "how to contribute to asahi" ;-p#asahi-gpu · 2022-12-13 20:06
alyssadocumenting all this would be good#asahi-gpu · 2022-12-13 20:03
iaguisI'll raise it in mesa/mesa after dinner#asahi-gpu · 2022-12-13 20:02
iaguisI see, makes sense#asahi-gpu · 2022-12-13 20:02
povikor are the interesting things to test linux-specific?#asahi-gpu · 2022-12-13 20:01
povikdedicate a macos machine to run the tests, so they can run on mesa/mesa?#asahi-gpu · 2022-12-13 19:59
alyssaand then when it's merged upstream, rebase asahi/mesa to pick up the change#asahi-gpu · 2022-12-13 19:58
alyssagit rebase -i upstream/main makes this easy enough#asahi-gpu · 2022-12-13 19:58
alyssaWhat Lina and I do is do dev on top of asahi/mesa, and then when everything is working, rebase on mesa/mesa (dropping all the asahi/mesa specific changes) and MR that#asahi-gpu · 2022-12-13 19:58
alyssabut asahi/mesa is constantly rebased on mesa/mesa and dev should happen in mesa/mesa