#asahi-gpu · 2021-01-18 00:50
bloom"the currnet prerelease version has native arm64 support"#asahi-gpu · 2021-01-18 00:50
bloomhttps://github.com/XQuartz/XQuartz/issues/26#asahi-gpu · 2021-01-18 00:45
bloom(IOGPU shows up in random places :p)#asahi-gpu · 2021-01-18 00:45
bloomLet's just say it's hard not to see the full list of symbols if you're debugging a Metal app :p#asahi-gpu · 2021-01-18 00:45
bloomI mean#asahi-gpu · 2021-01-18 00:44
davidrysk[m](or using symbols that exist for backwards compatibility)#asahi-gpu · 2021-01-18 00:44
davidrysk[m](basically finding exported symbols and using them)#asahi-gpu · 2021-01-18 00:44
bloomto which the answer is a big "i'm not a lawyer 🤷"#asahi-gpu · 2021-01-18 00:44
bloomdavidrysk[m]: That's definitely r/e, the better question is whether it meets the userspace 3d policy#asahi-gpu · 2021-01-18 00:44
davidrysk[m]also the "official" way to set up Xquartz native on ARM is via macports#asahi-gpu · 2021-01-18 00:43
davidrysk[m]bloom: I see you're not in -re -- is using nm or class-dump to dump header data considered RE?#asahi-gpu · 2021-01-18 00:42
bloomEasiest is probably setting up Xquartz tbh#asahi-gpu · 2021-01-18 00:41
bloomjust inconveniences me and my obscure use case :p#asahi-gpu · 2021-01-18 00:41
bloomit's a good technical decision#asahi-gpu · 2021-01-18 00:41
bloomdavidrysk[m]: right, but in this case the API had to be removed since the gfx stack was fundamentally rearchitectured at some point#asahi-gpu · 2021-01-18 00:39
davidrysk[m]bloom: typically when Apple removes stuff from headers but the symbols are there you can use old API headers and things *might* still work#asahi-gpu · 2021-01-17 23:33
bloombut even with AppKit there are.. hoops to jump#asahi-gpu · 2021-01-17 23:32
bloomyeah, the framebuffer will be tiled+compressed anyway#asahi-gpu · 2021-01-17 23:32
davidrysk[m]but as I recall it stopped returning anything useful#asahi-gpu · 2021-01-17 23:32
davidrysk[m]there are still symbols for it in CoreGraphics.framework#asahi-gpu · 2021-01-17 23:32
bloomblinks something about deja vu#asahi-gpu · 2021-01-17 23:31
davidrysk[m]bloom: well they used to have CGDisplayBaseAddress()#asahi-gpu · 2021-01-17 23:26
bloomShiz: and I've already been horrified by the whole "sgx" thing ;p#asahi-gpu · 2021-01-17 23:25
bloomShiz: I think ioreg has (most? all?) of that#asahi-gpu · 2021-01-17 23:12
Shizbloom: line 3705 of what was just posted in -re might be useful for you#asahi-gpu · 2021-01-17 23:02
ShizI see here in the devtree the vram is at phys 0x9e0df8000 with size 0x1500000(?)#asahi-gpu · 2021-01-17 22:14
bloomCALayer is maybe the closest?#asahi-gpu · 2021-01-17 22:07
bloomTangentially, why is there no public API on macOS to push pixels to the screen bypassing the GPU/#asahi-gpu · 2021-01-17 22:05
yrlftriangle let's go :D#asahi-gpu · 2021-01-17 22:01
jn__hey, it's a triangle! :)#asahi-gpu · 2021-01-17 21:52
bloom("Is the teaser that you're super colour blind?")#asahi-gpu · 2021-01-17 21:48
bloomhttps://rosenzweig.io/teaser.png#asahi-gpu · 2021-01-16 15:15
bloom(amxldzi and amxstzi)#asahi-gpu · 2021-01-16 15:15
bloom^^ Looks like there are instructions to accelerate tiling indeed :p#asahi-gpu · 2021-01-16 15:15
bloomhttps://gist.github.com/dougallj/7a75a3be1ec69ca550e7c36dc75e0d6f#asahi-gpu · 2021-01-15 03:04
davidrysk[m]night!#asahi-gpu · 2021-01-15 03:03
bloommarcan wins by default ig#asahi-gpu · 2021-01-15 03:03
bloomcalling it a night#asahi-gpu · 2021-01-15 03:03
bloom--yeah, not going to figure this out tonight#asahi-gpu · 2021-01-15 02:54
bloomoh, the kernel cares about this structure.. ugh..#asahi-gpu · 2021-01-15 02:51
davidrysk[m]HAH ok#asahi-gpu · 2021-01-15 02:50
bloomI'm kidding :p#asahi-gpu · 2021-01-15 02:50
davidrysk[m]where is kIOReturnDontWriteYourOwnDrivers?#asahi-gpu · 2021-01-15 02:50
bloomkIOReturnDontWriteYourOwnDrivers#asahi-gpu · 2021-01-15 02:49
davidrysk[m]😼