13:32 <robmur01> Woo! You got a mention: https://community.arm.com/arm-community-blogs/b/architectures-and-processors-blog/posts/software-enablement-and-the-morello-technology-demonstrator

13:34 <robmur01> (well, "You" being "Panfrost")

13:35 <daniels> robmur01: do keep up!

13:37 <robmur01> hey, I don't have time to find and read everything as soon as it comes out :P

13:41 <HdkR> The Morello blog post was a great read :)

13:45 <alyssa> robmur01: that was on Mark's Twitter, like, two weeks ago! :-p

13:45 <robmur01> "What's a Twitter?" :P

14:06 <daniels> it's LinkedIn for people who can't bring themselves to post earnestly

14:13 <HdkR> Does LinkedIn have blogging functionality built in to it?

14:15 <HdkR> "LinkedIn Publishing Platform" How silly

15:23 <rasterman> daniels: twitter? it's a public world wide irc channel for those with short attention spans... as obviously anything > 140chars is TL;DR :)

15:23 <daniels> rasterman: I don't think either of us can judge others for length of attention span tbf :P

15:25 <rasterman> daniels: :P :)

20:23 <cphealy> In the past I worked with an old Adreno A200 and with this GPU, there was a performance benefit to using glInvalidate as it saved us performance by allowing us to skip the cost of glClear (which was done using fragment shader). Is there any benefit to doing glInvalidate in the same manner on Mali GPU or is clearing essentially free?

20:41 <macc24> cphealy: check if `perf` debug option complains if you do glClear

21:52 <alyssa> clear is essentially free

21:54 <alyssa> I am suuuper tempted to play with Midgard, just for fun

21:54 <alyssa> done with Bifrost/Valhall for the week, don't have energy to do M1 stuff, and idk

21:54 <alyssa> Midgard is so cute I just wanna squish it

22:02 <macc24> alyssa: what if u make panvk work on midgard

22:56 <robclark> cphealy: the benefit of glInvalidate() is more about avoiding re-loading the tile buffer if there is no clear

23:05 <cphealy> Based on what alyssa and robclark said, it sounds like glInvalidate will not help like it did with the old A200. This is because clear is free, I don't need to find a way to avoid the cost of clear.

23:06 <cphealy> tnx alyssa and robclark

23:10 <alyssa> i wonder what a midgard compiler in Rust would look like wait no bad alyssa

23:30 <anarsoul> cphealy: if I'm not mistaken glInvalidateFramebuffer() will help in case if you don't need preserve depth buffer contents

23:30 <anarsoul> of course, if it's wired in panfrost :)

23:30 <anarsoul> *wired up

23:31 <alyssa> it should be

23:31 <anarsoul> is there glInvalidateFramebuffer equivalent for ES2?

23:32 <anarsoul> ah, here it is: https://www.khronos.org/registry/OpenGL/extensions/EXT/EXT_discard_framebuffer.txt

23:43 <cphealy> anarsoul: your bringing back memories talking about EXT_discard_framebuffer ;-)