16:17fdobridge: <gfxstrand> PSA: Rebasing nvk/main today
16:17fdobridge: <gfxstrand> "nv50/codegen: Set lower_uniforms_to_ubo" is fun to rebase over. 😅
16:18fdobridge: <gfxstrand> It's a good patch. I'm very glad @Mr Fall🐧 did that because now the UBO indices we set in NIR are the ones that it tries to load from instead of codegen implicitly adding 1. But, also, I didn't know that had changed and had to debug a bit. 😅
16:43fdobridge: <![NVK Whacker] Echo (she) 🇱🇹> Welcome back (I'm still surprised codegen is being maintained)
16:53fdobridge: <karolherbst🐧🦀> oof
16:55fdobridge: <karolherbst🐧🦀> it's more about deleting code
17:17fdobridge: <gfxstrand> Actually, no, it's still a mess
17:17fdobridge: <gfxstrand> Now it's actually a slightly worse mess. 😕
17:21fdobridge: <gfxstrand> Now I need to set both cb0 and cb1 because sysvals still come from cb1
19:38fdobridge: <gfxstrand> Rebase complete
19:54fdobridge: <gfxstrand> Hopfully merging tess soon
21:03fdobridge: <![NVK Whacker] Echo (she) 🇱🇹> https://gitlab.freedesktop.org/mesa/mesa/-/issues/9067 🤔
22:20fdobridge: <gouz> I will try to fix clip tests tomorrow!
22:29fdobridge: <gfxstrand> - NAK: `Pass: 313148, Fail: 906, Crash: 1150, Skip: 1672133, Flake: 38, Duration: 1:01:43`
22:29fdobridge: <gfxstrand> - codegen: `Pass: 313758, Fail: 1282, Crash: 161, Skip: 1672133, Flake: 41, Duration: 28:26`
22:38fdobridge: <Mohamexiety> nice! almost parity now..
22:38fdobridge: <Mohamexiety> why does NAK take over twice as long though? due to the crashes?
22:41fdobridge: <gfxstrand> Crashes hurt a lot
22:42fdobridge: <karolherbst🐧🦀> especially GPU ones
22:42fdobridge: <gfxstrand> The deqp-vk executable has non-trivial start-up time.
22:47fdobridge: <Mohamexiety> I see yeah. thanks!
23:01fdobridge: <gfxstrand> There may also be some overhead from running Rust in debug mode. Release mode rust is usually way faster.