06:26 kode54: huh
06:26 kode54: compression was already merged, but ZCULL hasn't
06:53 dylanetaft: ds
07:06 dylanetaft: Sorry wrong window
07:36 misyltoad[d]: dylanetaft: no dlss either :(
07:36 misyltoad[d]: maybe next year :p
08:03 kode54: I tried to use 6.19 rc2 with the zcull patch set applied
08:03 kode54: all I got for my trouble was a NULL pointer dereference in the GSP firmware init code
08:04 kode54: get_maintainers.pl still lists Maarten Lankhorst :[
08:05 kode54: hope he finds suitable station for his skillset
09:47 mohamexiety[d]: misyltoad[d]: wait what's that one stalled on?
09:53 esdrastarsis[d]: kode54: And there's no nvdec engine patch too 🙁
09:59 marysaka[d]: mohamexiety[d]: I think it was more out of fear it would not behave well on certain games but I really think we should just merge it and see how it goes (maybe add a flag to enable it?)
10:49 kode54: I want PQ color output too, even just for desktop software
10:50 kode54: does anyone even know how to turn that on at the hardware level?
10:50 kode54: wait, there's an nvdec engine patch?
10:58 mohamexiety[d]: pq color output?
11:02 kode54: the mess that HDR monitors choke on if they don
11:02 kode54: 't also choke on scRGB
11:03 kode54: nouveau doesn't appear to be able to do anything but sRGB or gamma 22 or whatever the basic standard is this week
11:03 esdrastarsis[d]: kode54: kinda, in the vulkan video MR there's a PoC patch for the kernel
11:03 kode54: oh, interesting
11:03 kode54: if I could get the kernel to work
11:03 kode54: I posted a bug report to the mailing lists and the maintainers of the file I crashed in
11:04 kode54: and stupidly forgot to prefix the subject with [BUG]
11:08 kode54: https://lore.kernel.org/dri-devel/176682185563.8256.115798774340102079@copycat/T/#u looks like it made it to lore for dri-devel, but not nouveau
11:08 kode54: I probably completely screwed up that report, though
12:27 phomes_[d]: I did not get that crash when I recently tested ZCULL on top of 6.19-rc1
12:53 phomes_[d]: marysaka[d]: It was working well in most of my testing. Only a few modes in specific games are not.
12:53 phomes_[d]: Maybe I should extend the testing to also check if we get similar perf results as prop driver for each mode
12:54 marysaka[d]: phomes_[d]: It could be nice to have that data yes!
12:55 marysaka[d]: phomes_[d]: on another note, my RM backend for NVK is almost passing CTS (excluding WSI, sparse and external memory as I still have to dig those) hopefully will have something usable in January so we could start compare any perf difference between nouveau and RM kernel module
12:56 marysaka[d]: If there is not much difference, we could start gathering more data with context switches info collected by nsight and possibly prototype quickly some stuffs in userspace that way (like TSG, CE engine selection, ect)
14:27 kode54: does 6.19 require newer firmware for nvidia?
14:27 kode54: I also see it's mentioning r535 in the function names, probably a false positive to my brain though
14:28 kode54: yeah, false positive, misunderstanding on my part
14:49 misyltoad[d]: mohamexiety[d]: review from faith
16:33 mhenning[d]: misyltoad[d]: Faith isn't really reviewing nvk MRs right now. In my mind, that MR is blocked on erroring on some of the cases I mention in https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37898#note_3161559 and fixing the bugs that phomes has found during testing.
20:24 misyltoad[d]: mhenning[d]: i see, will check it in the new year