#asahi-gpu · 2023-01-04 16:53
eric_engestromappear fully transparent#asahi-gpu · 2023-01-04 16:53
eric_engestromI'm getting leftover text from the linux console in the shape of an apple in the middle of my screen, and it's above everything else (including the cursor), but it's like it's rendered with a single value used for all channels (only grey, and the opacity seems to match the luminosity, black being fully transparent), so I'm guessing it might have been missed because if there's no text that happens to be right there, it will #asahi-gpu · 2023-01-04 16:53
eric_engestromlina: any chance that "apple logo" at boot is its own hardware plane and it's missing a clear in your driver's init?#asahi-gpu · 2023-01-04 16:47
marcanso in practice w40a is going to be close to ~2x the memory bandwidth of w18p or b3a8#asahi-gpu · 2023-01-04 16:45
marcanbi-planar formats are much nicer because the alpha is going to be all 0xff 99% of the time, which should compress down to ~nothing with framebuffer compression#asahi-gpu · 2023-01-04 16:45
marcanjannau: w40a is in there (#4), and I would expect it works but it's such a massive waste of memory bandwidth... not a good idea for the main framebuffer#asahi-gpu · 2023-01-04 16:05
danielssingle-planar also works fine with wl_shm, you just don’t get to choose your chroma offset or stride#asahi-gpu · 2023-01-04 14:36
jannaumulti-planar YUV works with dmabuf without issues#asahi-gpu · 2023-01-04 14:28
dottedmagdmabuf _seems_ not to hardcode assumptions: width/height/format plus a number of weird flags (y invert, interlaced, bottom-first-interlaced), so it might work with multiplanar if they are advertised.#asahi-gpu · 2023-01-04 14:22
jannauyes, 6 bit padding per component. I guess handling 40 bit per pixels is annoying on HW side as well. the kwin code in question uses zlinux_dmabuf#asahi-gpu · 2023-01-04 14:18
dottedmagAlso Wayland implicitly implies that all the buffers are single-planar in wl_shm constructor. Nobody uses this constructor for anything useful, but well.#asahi-gpu · 2023-01-04 14:13
dottedmagIs it 10-bit with 6-bit unused per channel?#asahi-gpu · 2023-01-04 14:10
jannauw40a works for dcp but is despite its name 64 bit per pixel#asahi-gpu · 2023-01-04 12:53
jannaudid you test "w40a" 40-bit single plane ARGB10? I think I saw that as supported as well but it was missing from dcp_surfaces.txt#asahi-gpu · 2023-01-04 12:49
marcanin particular kwin seems to prefer alpha formats over alphaless, but right now we only *actually* support X2R10G10B10 as a 10-bit format (we claim A2 also but it's a lie). so right now kwin picks the A2 format even though it has no real alpha. we should offer it a proper 2-plane 10-bit alpha format as the best choice.#asahi-gpu · 2023-01-04 12:47
marcanalyssa, lina: BTW, does asahi handle multiplanar formats yet? DCP supports a bunch of those (and we're going to need them for some use cases), and Apple use multiplanar RGB10+A8 by default for UI stuff (compressed) so I assume it's supported by the hardware somehow (or is that just effectively "lol MRT"?)#asahi-gpu · 2023-01-02 17:54
alyssalina: Are you planning to review !20446?#asahi-gpu · 2022-12-30 18:05
alyssanot images#asahi-gpu · 2022-12-30 18:05
alyssakarolherbst: I have patches for ssbo's#asahi-gpu · 2022-12-30 16:15
karolherbstI might be inclined to work on that.. maybe... if I got time...#asahi-gpu · 2022-12-30 16:15
karolherbstor well.. ssbos?#asahi-gpu · 2022-12-30 16:15
karolherbstanybody started to figure out images yet?#asahi-gpu · 2022-12-30 16:15
karolherbstcompute, yay!#asahi-gpu · 2022-12-30 15:39
linaBut yeah, the next big thing is compute which is going to change one big part of the UAPI, and if not coincidentally with that, sync/etc will come right after since it follows naturally#asahi-gpu · 2022-12-30 15:38
linaThough it's a pretty easy hack to just track some process names so if I ever run into unexplainable GPU usage (this one seems pretty obvious) it might be worth just doing anyway ^^#asahi-gpu · 2022-12-30 15:37
alyssa:+1:#asahi-gpu · 2022-12-30 15:36
linaSure ^^#asahi-gpu · 2022-12-30 15:26
alyssaimho getting the UAPI right (especially sync) is far more important right now#asahi-gpu · 2022-12-30 15:26
alyssaMaybe hold off on this until we figure out what we want to do for panfrost and then copy that some stream ;)#asahi-gpu · 2022-12-30 15:26
alyssahttps://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20215#asahi-gpu · 2022-12-30 15:25
alyssalina: we're working on similar stuff at $WORK for panfrost#asahi-gpu · 2022-12-30 15:25
alyssasounds good#asahi-gpu · 2022-12-30 15:02
linaNoting this down to investigate later#asahi-gpu · 2022-12-30 15:01
linaAnd opening it immediately uses that RAM.#asahi-gpu · 2022-12-30 15:01
linaOK, just closing Inochi2D cleared out half of that usage so yeah. That's not normal, Inochi2D with this model should be using a single 4Kx4K texture atlas plus mipmaps and a few intermediate framebuffers.#asahi-gpu · 2022-12-30 14:59
lina(The 4GB was constant as far as I can tell, so it's either not an outright leak or it's a leak caused by something I don't understand yet and which has a specific trigger, not leaking in the steady state)#asahi-gpu · 2022-12-30 14:58
linaThat should be enough info to at least begin to analyze/debug memory usage patterns... thoughts? ^^#asahi-gpu · 2022-12-30 14:58
linaAlso keeping the creation timestamp is a good idea I think#asahi-gpu · 2022-12-30 14:58
linaBut maybe mesa could also pass some kind of identifier for what each allocation is, at least some kind of class bits? Like what PIPE_USAGE flags were used#asahi-gpu · 2022-12-30 14:57
linaI was thinking we might want to add some GPU memory consumption tracking? At the very least the driver should track which process (PID/comm, surviving process termination) initially created each GEM object I think#asahi-gpu · 2022-12-30 14:56
linaalyssa: So one thing I noticed when setting up today's stream is I ended up with (I think) ~4GB of GPU memory usage, which seems a bit excessive considering it was mostly just Inochi2D and OBS#asahi-gpu · 2022-12-30 02:28
alyssaas usual#asahi-gpu · 2022-12-30 02:28
alyssai was wondering and then forgot#asahi-gpu · 2022-12-30 02:28
alyssathanks for checking that :)#asahi-gpu · 2022-12-30 02:27
linaYeah, I saw it with a compute shader with tons of params#asahi-gpu · 2022-12-30 02:27
alyssai wasn't sure what the max was#asahi-gpu · 2022-12-30 02:27
alyssawill fix that in the rewrite. thanks for letting me know#asahi-gpu · 2022-12-30 02:26
alyssalina: brrrr okay, that tracks.#asahi-gpu · 2022-12-30 02:26
linaYou need to split after that (or at least Metal does)