#
✕ nick:marcan
#asahi-gpu · 2023-01-23 05:04
marcansorry, I just want to merge the new UAPI into asahi-dev soon :p
#asahi-gpu · 2023-01-23 05:01
marcanI also need to fix the stupid mailbox situation, I get the feeling that's going to help
#asahi-gpu · 2023-01-23 05:00
marcan(and lina)
#asahi-gpu · 2023-01-23 05:00
marcanalyssa: That's why I wanted a meeting with you and jekstrand!
#asahi-gpu · 2023-01-22 02:42
marcanmoo?
#asahi-gpu · 2023-01-21 19:42
marcanalyssa: lol. I think we still need a spinlock for submission, but not like the 4 or whatever mailbox manages to use. also, might be able to play games with atomics to avoid the spinlock in the uncontended case.
#asahi-gpu · 2023-01-05 06:55
marcanno, that's bg_color in the DCP swap if you then reduce the plane size along with it :p
#asahi-gpu · 2023-01-04 17:25
marcanI should get some sleep :p
#asahi-gpu · 2023-01-04 17:25
marcanjannau: you're right, sorry... misread the patch
#asahi-gpu · 2023-01-04 17:16
marcannot that it really matters right now since we only support a single plane anyway...
#asahi-gpu · 2023-01-04 17:16
marcanAIUI if a bit is set there and the surface is NULL, the plane is removed
#asahi-gpu · 2023-01-04 17:16
marcanit's not really "enabled", more like "update/request"
#asahi-gpu · 2023-01-04 17:16
marcanjannau: btw, now that I noticed, I think there's one missing thing still. when layers are removed, they should have their bits set in swap_enabled
#asahi-gpu · 2023-01-04 16:58
marcansee https://gist.github.com/marcan/d93ebdffce5b36be9aaa06079a60503f for what I've been able to find about supported pixel formats
#asahi-gpu · 2023-01-04 16:57
marcan(still feels really weird though)
#asahi-gpu · 2023-01-04 16:57
marcan(also obviously the hardware supports it, it's just DCP firmware that doesn't because I guess Apple never needed it for macOS?)
#asahi-gpu · 2023-01-04 16:57
marcan*bottommost plane
#asahi-gpu · 2023-01-04 16:57
marcanwe can cheat that with premultiplication for the bottommost planes, but not any overlay planes
#asahi-gpu · 2023-01-04 16:56
marcanI'm still quite annoyed that there isn't a "turn off alpha" bit anywhere as far as I can tell, and no XRGB8888 support at all
#asahi-gpu · 2023-01-04 16:55
marcanit's all pretty cursed
#asahi-gpu · 2023-01-04 16:55
marcanwhich is why the iBoot framebuffer works as intended, but then the apple comes back later when we enable another plane
#asahi-gpu · 2023-01-04 16:55
marcanthe apple shape is the original iBoot alpha, which interestingly does not actually apply on boot because DCP does some optimization where fullscreen single planes end up ignoring alpha anyway
#asahi-gpu · 2023-01-04 16:54
marcanand yes, that happened when we switched to plane 1 because plane 0 seems to be special on t600x, combined with the lack of alphaless formats (which we were unknowingly cheating by marking them as premultiplied)
#asahi-gpu · 2023-01-04 16:54
marcanthat's fixed in the 6.2-rc2 rebase I just pushed
#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 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 · 2022-12-28 12:59
marcanlina: oh god I just saw your tweet, please don't make me have to package CEF... *groan*
#asahi-gpu · 2022-12-28 12:48
marcanit's not practical because it will break every macOS update and require reverse engineering the new ABI
#asahi-gpu · 2022-12-28 12:46
marcanalso it's okay if I start tagging timestamped releases, right? otherwise we lose prior state
#asahi-gpu · 2022-12-28 12:46
marcanalyssa: I'll make a build a bit later and push it to asahi-dev
#asahi-gpu · 2022-12-28 12:45
marcansven: wrong channel?
#asahi-gpu · 2022-12-28 12:43
marcanask alyssa and lina about the current state, but at the very least it probably will only work on 12.3
#asahi-gpu · 2022-12-28 12:42
marcantly: The macOS build is for reverse engineering mostly, it's not really expected to work ^^
#asahi-gpu · 2022-12-22 17:58
marcanif not though, I think we're going to have to use the matrix stuff to expand it back to 0..1024 for most software to not be confused (or at least offer that option)
#asahi-gpu · 2022-12-22 17:58
marcanhowever, the weirdo range bothers me. if it turns out it's that DisplayPort thing, Linux already actually has something for that
#asahi-gpu · 2022-12-22 17:57
marcanand then we should implement the proper colorimetry connector modes
#asahi-gpu · 2022-12-22 17:57
marcanthis also means we should probably be using colorimetry 12 for both 8-bit and 10-bit modes for now, as that gives full uncalibrated range (oversaturated for sRGB, but hey people like that)
#asahi-gpu · 2022-12-22 17:56
marcanthe colorimetry table we have says 9 is BT.2020 and 12 is "scRGBfixed" which sounds like a plausible code for "native whatever" or something abused as such
#asahi-gpu · 2022-12-22 17:55
marcanso I think the green primary is purer than P3
#asahi-gpu · 2022-12-22 17:54
marcanI see significant red clipping in BT.2020 mode, but only a bit of green clipping. if it were strictly P3 I'd expect more green clipping
#asahi-gpu · 2022-12-22 17:54
marcanjannau: I'm not sure native is actually P3. I think internally they use P3 (the rtings.com shows the display as almost perfectly matching P3, which is... unlikely) but the display can do more
#asahi-gpu · 2022-12-22 17:53
marcanI think 9 is BT.2020 and 12 is just native
#asahi-gpu · 2022-12-22 17:48
marcanjannau: updated with 8-bit mode observations
#asahi-gpu · 2022-12-22 17:22
marcanjannau: so the supported modes are pretty boring (but make some sense): https://gist.github.com/marcan/5c89b85518827bca419bcbbb4bdf6b62
#asahi-gpu · 2022-12-21 20:15
marcanjannau: I think only 0-2 are supposed to work?
#asahi-gpu · 2022-12-21 19:32
marcansleeps
#asahi-gpu · 2022-12-21 19:31
marcanI guess it's possible we're missing some swap flags to make the colorspace properly effective? dunno, maybe I should see if I can find the iBoot swap struct in RAM too...

plus ›