01:00 <alyssa> UBO series is up! :)

01:06 <alyssa> and multisampling branch rebased

02:10 <alyssa> and pushed. calling it a night

02:11 <urjaman> nn

08:13 <macc24> alyssa: anything to test?

13:54 <alyssa> macc24: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8973

13:55 <alyssa> and https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774 if you're feeling adventurous (depends on the above)

14:08 <robertfoss_> I saw the lca talk about panfrost + gles 3.2 compute

14:08 <robertfoss_> Which was really neat, and made me think of opencl support

14:09 <robertfoss_> Does anyone have a branch with early opencl support for panfrost?

14:11 <amonakov> robertfoss_: can't find it, which talk is that?

14:11 <robertfoss_> https://www.youtube.com/watch?v=vXZmx2aYXwQ

14:13 <macc24> alyssa: okay ill test it

14:20 <macc24> alyssa: i think i saw some "Purging" messages on my odroid go super

14:20 <macc24> on 5.11 + panfrost

14:33 <macc24> alyssa: i'm not sure what 8973 could break, but it didn't break xonotic, supertuxkart, firefox, chromium or sway

14:33 <macc24> on G72 of course

14:38 <alyssa> =)

14:56 <macc24> if i wasn't tired now i'd test on midgard too

18:27 <icecream95> alyssa: The CL flakes were my fault. When writing emit_load_kernel_input, I added a bugfix which I didn't also put in the UBO code, so that when moving kernel_input support back to emit_load_ubo, the fix got lost

18:29 <icecream95> alyssa: The scheduler broke 855e29cd78f, which caused a lot of INSTR_INVALID_ENC faults for dEQP-GLES31 compute tests

20:07 <alyssa> icecream95: I see.

20:07 <alyssa> I thought I backported that fix but maybe it got lost.

20:07 <alyssa> Er, forward-ported

20:07 <alyssa> I guess not, shoot.

20:13 <alyssa> Will spin up a fix, sorry about that.

20:19 <alyssa> I specifically remember typing out a fix uh

20:20 <alyssa> git reflog...

20:21 <icecream95> git bisect needs a mode for reflog

20:23 <alyssa> lol

20:24 <alyssa> icecream95: I need to run so this isn't even compile-tested, but https://rosenzweig.io/0001-pan-bi-Fix-jumps-to-terminal-block-again.patch should do what we want (I think)

21:10 <icecream95> alyssa: That patch doesn't seem to work…

21:17 <icecream95> alyssa: That patch looks at the block of the branch, not the block of the target

21:28 <icecream95> alyssa: Also, list_is_empty in bi_is_terminal_block returns false because `list->{prev,next} == 0` is not considered an empty list

22:16 <alyssa> told you it wasn't even compile tested, whoops

23:46 <macc24> :o

23:46 <macc24> minecraft bedrock edition runs on G72

23:52 <HdkR> whoa

23:52 <macc24> it doesn't like being touched

23:53 <HdkR> bpx86 I guess?

23:53 <macc24> nope

23:53 <macc24> mcpelauncher

23:53 <HdkR> huh

23:54 <macc24> https://i.imgur.com/F0KoGJ5.jpeg

23:55 <HdkR> Oh, is mcpelauncher a way to run Android apps?

23:55 <macc24> only minecraft runs on it

23:56 <HdkR> So an Android app launcher that only runs ANdroid? :D

23:56 <HdkR> er, Minecraft*

23:56 <macc24> minecraft launcher that only runs minecraft

23:57 <macc24> now to get it running on odroid go super when i get analogs working