#asahi-gpu · 2025-04-24 18:31
alyssahttps://rosenzweig.io/0001-asahi-do-not-use-Null-layout.patch the fix#asahi-gpu · 2025-04-24 18:20
alyssawell, I'm glad there's a reason for all this craziness.#asahi-gpu · 2025-04-24 18:20
alyssaahahah#asahi-gpu · 2025-04-24 18:20
alyssahttps://developer.apple.com/forums/thread/727492#asahi-gpu · 2025-04-24 18:20
alyssa[GPUDebug] Null texture access executing kernel function "ColorCorrection" encoder: "Keyframing.SurfaceWarpFuser.InverseWarpKeyframe", dispatch: 2#asahi-gpu · 2025-04-24 18:18
alyssaI guess one of Apple's debug tools wants to check whether null textures are ever accessed#asahi-gpu · 2025-04-24 18:18
alyssaugh this is insane#asahi-gpu · 2025-04-24 18:17
alyssaugh and that's why the blob twiddles the null textures too#asahi-gpu · 2025-04-24 18:16
alyssabut apparently still some things left T_T#asahi-gpu · 2025-04-24 18:16
alyssar/e eureka momenta are rare these days for m1#asahi-gpu · 2025-04-24 18:16
alyssaahhh#asahi-gpu · 2025-04-24 18:16
alyssaand .. probably we should be using it, actually, but one problem at a time#asahi-gpu · 2025-04-24 18:16
alyssa...Oh shit and I bet I know what that legacy sparse layout is#asahi-gpu · 2025-04-24 18:10
alyssaI don't *think* it will, but, not impossible#asahi-gpu · 2025-04-24 18:09
alyssa.....Yeah, it does. I will do that too, thanks#asahi-gpu · 2025-04-24 18:07
jannaudoes it make sense to test if it fixes parallel CTS runs#asahi-gpu · 2025-04-24 18:06
alyssaOK#asahi-gpu · 2025-04-24 18:06
jannaualyssa: I have builds in my uapi copr but I wouldn't say too late#asahi-gpu · 2025-04-24 18:04
alyssa"random gpu memory corruption due to a misunderstanding of the hardware" seems like it could explain, some things#asahi-gpu · 2025-04-24 18:03
alyssathis is not a regression or anything but#asahi-gpu · 2025-04-24 18:03
alyssajannau: going to retag if it's not too late for today#asahi-gpu · 2025-04-24 18:03
alyssaonly by accident#asahi-gpu · 2025-04-24 18:02
alyssaI just discovered a serious bug affecting both our gl & vk drivers#asahi-gpu · 2025-04-24 18:02
alyssa Oh jesus#asahi-gpu · 2025-04-18 18:59
jannaunot sure if that's an issue in other VMMs than muvm. for muvm we can either ship patches or just not build venus#asahi-gpu · 2025-04-18 18:58
jannauI have fixes for the venus issue https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34613#asahi-gpu · 2025-04-18 15:03
alyssaso maybe rc2 + virtgpu is shippable#asahi-gpu · 2025-04-18 15:03
alyssaor I guess rc2 is actually only 5 days from now#asahi-gpu · 2025-04-18 15:03
alyssaand there are a bunch of fixes that didn't make it into rc1#asahi-gpu · 2025-04-18 15:03
alyssasince i don't want to wait for rc2#asahi-gpu · 2025-04-18 15:02
alyssayeah I figure we'll tag our own thing#asahi-gpu · 2025-04-18 15:02
alyssaah i see#asahi-gpu · 2025-04-18 15:02
jannaumy local is a random mesa checkout from yesterday + virtgpu + my fixes but rpm build uses 25.1.0-rc1 + patches#asahi-gpu · 2025-04-18 15:00
alyssaah yeah#asahi-gpu · 2025-04-18 15:00
jannauyes, it's just not in 25.1.0-rc1#asahi-gpu · 2025-04-18 14:59
alyssano?#asahi-gpu · 2025-04-18 14:59
alyssawait yes the cited commit should fix that#asahi-gpu · 2025-04-18 14:59
alyssajannau: I thought I merged the fix #asahi-gpu · 2025-04-18 14:44
jannaucommit 9b55451ea7b ("hk: fix underbinding scratch")#asahi-gpu · 2025-04-18 14:42
jannauhuh, a VM_BIND with range=0x400 in 25.1-rc1 I don't see with 6d70ec449ff (random commit from yesterday)#asahi-gpu · 2025-04-18 12:40
alyssaat least debug builds should've blown up#asahi-gpu · 2025-04-18 12:40
alyssachaos_princess: that.. is probably unrelated and IDK what that would be about#asahi-gpu · 2025-04-17 20:53
chaos_princessalyssa: someone did hit regalloc allocating more registers than exist, so. #asahi-gpu · 2025-04-17 20:47
alyssahttps://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34595#asahi-gpu · 2025-04-17 19:50
alyssai wonder how many folks have hit this in the wild#asahi-gpu · 2025-04-17 19:50
alyssai think i just fixed a bunch of games#asahi-gpu · 2025-04-10 18:03
alyssahmmmmmmm#asahi-gpu · 2025-04-10 17:57
jannau`vkmark -bvertex` has regressed slightly: ~5.9k (old uapi) vs. 5.4k (new uapi). Testing a different install though #asahi-gpu · 2025-04-10 17:26
jannauglmark2 raw value of 6.13 is lower than 6.14 old uapi but probably in the margin of error#asahi-gpu · 2025-04-10 17:23
alyssaalthough i thought i specifically benchmarked this