12:21 <alyssa> Dcow[m]: grin

12:25 <as400[m]> I've seen it yesterday - great presentation. Congrats to you both alyssa and lina.

12:27 <alyssa> thank u

12:38 <Dcow[m]> alyssa: wink

14:42 <lina> Thank you!

14:54 <fmstrat> I know I'm late to the party, but consider me one representative from many lurkers (just the ones I personally know, anyway) watching and waiting for for the phrase "It's stable enough" to make the transition on the M1s. Thank you for all the work being put into the GPU acceleration from us all.

14:59 <lina> <3

15:01 <fmstrat> lina: (and alyssa and the rest of the team) Seriously. There are many, many people appreciative, even if only a few of us join a network to say so ;)

15:02 <phire> I mean, "stable enough" really depends on your level of tollarace

15:03 <phire> I'm today looking into implementing t600x agx support in-between talks

15:06 <lina> Someone's already running Minecraft I hear

15:07 <i509VCB> I'm hoping the vulkan driver project comes around soon. Although I imagine I will probably be doing driver grunt work in it

15:08 <_jannau_> is the tlb invalidation / waiting for the gpu to sleep responsible for the long application startup times? I killed firefox the first time because I thought it was hanging

15:08 <lina> No, if you're running gnome-session manually, prefix it with dbus-run-session

15:08 <lina> Otherwise there is some dbus timeout thing

15:09 <_jannau_> it was weston in my case but could be still the same issue

15:11 <lina> Weston needs it too, yes!

15:18 <Dcow[m]> does GDM works?

15:36 <_jannau_> phire: I've hoped t600x would wait until lina can rebase onto (asahi-)v6.1-rc1. just to avoid dealing with the t600x devicetree changes I did multiple time. but go ahead this, I'll fix that tonight (CEST)

15:37 <phire> Oh, it will take me a while. Without access to a second laptop, I'll have to take stabs in the dark and reboot

15:39 <_jannau_> mainly a note that you can't use my tree with never dcp version and m1n1 support for filling the UAT reserved memory nodes

16:01 <phire> currently I'm working on m1n1. Haven't got to kernel

16:03 <_jannau_> ah, that's annoying without a second computer. Can you ask around if someone has a spare laptop?

16:06 <phire> if I get that far

16:06 <phire> alyssa?

16:50 <alyssa> phire: uhh

16:51 <alyssa> phire: what's the question?

16:52 <phire> Oh... Just wondering about borrowing a laptop

16:53 <alyssa> Don't have any apple hw with me

16:53 <phire> literally just needs libusb

16:53 <phire> I'm wondering if I could find a cheap chromebook in a store

17:59 <jannau> gpu driver faults in glmark2-es2 terrain with "Unknown event message: Fault"

18:00 <jannau> manages ~10 fps with full printk logging

18:11 <jannau> with most of the printk spam disabled mostly 600 - 700 fps, lowest value is 175fps for "[desktop] effect=shadow:windows=4"

19:23 <phire> does 19.551w seem sane for a GPU power limit on the M1, and 81.415w on my M1 max (unk_674 & unk_698 in AGXHWDataA)

19:31 <jannau> higher than the numbers apple claimed 15W/60W but not completely out of range if this is short peak power use

19:31 <jannau> I could check a 24 gpu core m1 max

19:35 <phire> I've found a few unknowns that seem to match the number of gpu cores

19:35 <phire> HWDataB unk_53c and unk_568

20:29 <phire> the good news is that all of the initdata structure layouts seem to be indentical

20:29 <phire> and they only things change are mostly power data

21:10 <phire> Maybe I spoke too soon. The end of RegionC might be diffrent

21:32 <alyssa> Weee

21:39 <Dcow[m]> phire: what firmware are you at?

21:40 <phire> This is a older dump of the 12.1 firmware that I'm lining up with the code in m1n1

21:41 <Dcow[m]> as far I remember, lina saw that at 13 beta 4 RegionC is also different.

21:41 <Dcow[m]> might be incorrect

21:42 <Dcow[m]> myself, not lina* (lina is goddess)

21:48 <phire> well, it seems to go off the rails at RegionC.unk_893c, which is 29520 on M1 12.1, 30000 on M1 13.0 beta 4, and 0 on my m1 max

21:49 <Dcow[m]> moved likely

21:50 <phire> yeah, stuff seems to be missing, and I can see that AGXHWDataShared1/2 have moved fowards to fill the gap

21:50 <phire> but then it seems to re-sync later? there might be some extra padding?

22:03 <jannau> I guess I have to look into brightness control, I fixed DPMS westong issue with internal displays but the displays seems to wake up with minimal brightness

22:55 <phire> lina: Any idea what an array of 32bit ints looking like `[0x0000ffff] * 16` might be?

23:44 <lina> phire: I think I saw that already in t8103, but I have no idea what it is...

23:45 <lina> jannau: Yeah, there is no fault handling yet. Right now you have to track down those faults manually with the hypervisor.

23:47 <lina> phire: Someone is testing on a 7-core Air and it seems to work, not sure if the driver is expected to get/forward the number of cores somewhere?

23:47 <lina> Or maybe it's the max number of physical cores, not actually enabled ones?

23:48 <lina> phire: Conjecture: there is a union of HW-related data for different chips, that's why it resyncs

23:49 <lina> jannau: The person testing on the Air also reported that quitting sway causes the brightness to step down every time, until it goes to 0