09:20 <Ashleee> geez they definitely did not make it easy to cross compile mesa did they :)

09:20 <Ashleee> "oh look new meson, you just specify --cross-file" but literally nothing is found

09:20 <Ashleee> (even though in the same env I compiled everything from kernel through glibc to curl, htop)

10:34 <HdkR> The most frustrating experience when cross-compiling meson and cmake projects is that nobody provides blessed cross compile sample files. Sure it's claimed to be "easy", but I want an example I can copy and paste, then change as I find breakages. Because I don't know what the format is.

12:36 <Ashleee> hehe and docs are silent as well

12:36 <Ashleee> I got libdrm compiled

12:36 <Ashleee> as a pre-requisite of mesa

12:37 <HdkR> and a new enough kernel would be a prereq of that :D

13:21 <Ashleee> no worries about that, running 5.10.1 :)

14:22 <alyssa> Ashleee: As I hope you know, panfrost does not support opencl (yet)

14:47 <Ashleee> well :)

14:47 <Ashleee> and Lima do?

14:50 <HdkR> Lima can't physically support CL

15:15 <Ashleee> oh I thought it had OpenCL 1.1

15:18 <alyssa> I do want to work towards OpenCL and Vulkan support for panfrost, but time tables are notoriously hard to predict.

15:18 <alyssa> Collabora's priority has been providing kickass OpenGL ES drivers.

15:18 <alyssa> As for !Collabora contributors, patches very welcome ;)

15:22 <Ashleee> oh you're right, for whatever reason I thought that Mali400 has got OpenCL 1.1 :)

15:23 <Ashleee> apparently I was very wrong

15:23 <Ashleee> oh I know why, Iw as looking at Allwinner H6 which has Mali Twhatever thinking A64 has the same

15:30 <HdkR> It's a confusing mess when Utgard GPUs are still releasing as Midgard CPUs are rolling out

15:30 <Werner> H6 is T720 IIRC

15:36 <Ashleee> yeah

15:37 <Ashleee> I have a bunch of H6 devices and a bunch of rk3399 devices

15:37 <Ashleee> and about two dozens of A64 devices :)

17:56 <macc24> HdkR: i think i have cross-file for mesa in my armhf-tools repo

17:59 <macc24> icecream95: i don'e think media.patch, parts of panfrost.patch and sound.patch are needed

17:59 <macc24> usb.patch is a revert of some mediatek commits that broke usb when usb sound card or mouse is connected

20:11 <Lyude> well, igt bringup is probably about to get a lot more complicated :|

20:11 <Lyude> oops

20:11 <Lyude> wrong channel

20:12 <macc24> first time i see you, Lyude lol

20:12 <Lyude> hi!

20:12 <macc24> hey!

20:13 <Lyude> yeah-I haven't had any time to work on panfrost for a while now :s, and haven't really had the energy to try getting back into it on my off time

20:14 <HdkR> Obviously Lyude is just waiting until I can yeet some Valhall hardware at them :P

20:14 <macc24> panfrostt on valhall when :-D

20:15 <macc24> Lyude: in the meantime a fancy bifrost laptop has dropped :D

20:15 <HdkR> We need an SBC that isn't running Android so it's not a pita :/

20:16 <macc24> HdkR: ;)

20:17 <HdkR> I'm surprised that there are still zero on the market

20:54 <alyssa> HdkR: Bifrost took a while too iirc

20:57 <macc24> i heard that mt8192 will likely have valhall

21:43 <alyssa> 👀

22:16 <macc24> alyssa: https://www.mediatek.com/products/tablets/mt8192

22:17 <macc24> G57 is valhall

22:20 <macc24> and i think that's a chromebook chip

22:28 <icecream95> 10-bit gamma seems to be broken on mt8183 - there are graphical artefacts that go away if I change LUT_10BIT_MASK from 0x03ff to 0x03fc

22:33 <alyssa> macc24: I see

22:34 <macc24> can't wait to get my hands on it, hopefully next year

22:35 <alyssa> probably won't be panfrost drivers ready :(

22:35 <alyssa> but you never know 🤷

22:36 <macc24> someone's gotta get linux running on it before any development happens

22:39 <macc24> icecream95: uhh, does that work on llvmpipe?

22:42 <icecream95> macc24: ?

22:43 <macc24> icecream95: 10-bit gamma

22:43 <urjaman> i guess that's not a mesa thing, so that's why the question is nonsensical?

22:44 <urjaman> (as in gamma done in the mt8183 drm driver...)

22:59 <alyssa> also why I don't know what it is

22:59 <alyssa> why would you only want 10-bits in your generalization of factorial