08:18karolherbst_work: RSpliet: the boost level 0 is a value where the vbios guarentees, that no matter what load, the fan can always keep the gpu cool at max speed (or something along those lines)
08:18karolherbst_work: and 1 "mostly" safe
09:31karolherbst_work: can somebody verify, that mmiotracing nouveau doesn't work anymore on recent kernels?
09:31karolherbst_work: and/or confirm
09:32karolherbst_work: I think my patch broke it, which is kind of odd, because it fixed tracing nvidia....
09:32karolherbst_work: no idea what the big differences is here
09:46t-ask: can I switch to stable_reclocking_kepler_v6 istead of v5 or is v5 still recommended?
09:54karolherbst_work: t-ask: doesn't matter
09:55t-ask: ok then I stick with v5 as v6 throws build/nouveau/drm/nouveau/nouveau_drm.c:317:6: error: implicit declaration of function ‘vga_switcheroo_client_probe_defer’ [-W error=implicit-function-declaration] if (vga_switcheroo_client_probe_defer(pdev))
09:59karolherbst_work: t-ask: both branches are based on 4.7 kernel
10:00t-ask: it is more a question if you like me to test v6, too. v5 is now working well. I got it working last night with updates folder
10:08karolherbst_work: you can test v6 if you want, but it shouldn't change a thing (in theory)
10:27karolherbst_work: pmoreau: I think I will put some patches into the staging repository. If you want anything specific I can add it. I currently plan to add the led stuff and other things I may find at home today
10:43RSpliet: karolherbst_work: I assumed so yes
11:04pmoreau: karolherbst_work: Nice! I didn’t had time to play with the staging repo and include that on the image, as I was busy with SPIR-V.
11:05karolherbst_work: no problems
14:38Tom^: karolherbst_work: what does v6 add? :p
14:39karolherbst_work: clean ups
14:39karolherbst_work: review stuff
14:39karolherbst_work: I don't add new features anymore .p
20:54karolherbst: imirkin: mhh, as it seems the issues I've found are indeed regressions :)
20:55imirkin_: the compiler issues?
20:55karolherbst: I had a src/glsl/glsl_compiler
20:55karolherbst: and no crash with that one
20:55karolherbst: and this is from before the compiler moves inside mesa
20:56karolherbst: did only check the #version 0 issue though
20:56karolherbst: but I guess the other thing could also be a regression
20:57karolherbst: nope, the other thing still crashes
21:16karolherbst: mupuf: what do you think about nouveau logo stickers? Big ones, like 30x30 cm
21:17mupuf: lol, could be fun. Why do you want them?
21:17karolherbst: once for my laptop, because it looks too plain currently
21:17karolherbst: the others for sticking at random locations of course!
21:17karolherbst: we need 200 by the way
21:18karolherbst: or mhh
21:18karolherbst: maybe 10 30x30cm ones and 50 10x10 cm ones?
21:18karolherbst: no idea how expensive that stuff is
21:25karolherbst: imirkin_: that glsl error was introduces like this month
21:25imirkin_: the version 0 one?
21:25imirkin_: probably me
21:25imirkin_: or ian
21:25karolherbst: currently running bisect run
21:25imirkin_: but most likely me.
21:25karolherbst: 29e1c4a8a9f26ce41aa53dc9bf39852a8530adc6 - b82de88008ddfef051eeccfbc4b36e0e7d47daf3
21:25karolherbst: somewhere in between there
21:26imirkin_: oh interesting. probably not me then.
21:27karolherbst: 29e1c4a8a9f26ce41aa53dc9bf39852a8530adc6 - a7d33315a76efaa8943dbe439f5538ce6d76ff41
21:28imirkin_: my money's on eda6349346616f3a45ca2d03e2c1a3da956df6b3
21:28karolherbst: you lost already :p
21:28karolherbst: ohh wait
21:28karolherbst: you didn't
21:28karolherbst: it is still there
21:29karolherbst: yeah, looks like it
21:29imirkin_: told you it'd be me or ian ;)
21:29karolherbst: bisect run is awesome :3
21:30karolherbst: stupid is, glsl_compiler used to return 1 on compile error
21:30karolherbst: and returns 0 now
21:31karolherbst: a bitty messy to workaround that
21:31karolherbst: imirkin_: I think you are right, 2 commits lets
21:31karolherbst: yep, it is eda6349346616f3a45ca2d03e2c1a3da956df6b3
21:32imirkin_: you can't do a straight up revert
21:32imirkin_: you also have to revert the patch that came after it
21:32karolherbst: I don't plan to revert it
21:34karolherbst: imirkin_: is it fine if I add ian as CC in that bug?
21:35imirkin_: yeah. he probably already is anyways.
21:35imirkin_: he's email@example.com
21:35karolherbst: I was thinking about using his email from the commit
21:36karolherbst: "did not match anything"
21:37karolherbst: mhh odd
21:37imirkin_: use the one i said
21:37karolherbst: ahh, I think I mistyped earlier
21:38karolherbst: the fdo one works
21:39karolherbst: now the question, how can that commit break it...
21:40karolherbst: I see
22:43mek42_laptop: for a single GTX 760 used as a gpu, am i reading the chart correctly that there is not much functionality difference between nouveau and proprietary driver?
22:53imirkin_: mek42_laptop: depends how you judge functionality, i suppose
22:54imirkin_: in terms of actual GL exts, blob supports GL_ARB_enhanced_layouts and GL_KHR_robustiness, and in turn GL 4.5 (which basically requires those exts)
22:54imirkin_: it also supports ES 3.2 while nouveau only gets you ES 3.1 for now
22:55imirkin_: and obviously they support a huge pile of NV_* extensions
22:58mek42_laptop: wow - a whole bunch more stuff to read about
23:05imirkin_: mek42_laptop: is there some underlying question you're trying to answer?
23:11mek42_laptop: just trying to decide which to use
23:11imirkin_: if you are concerned with closed-source blobs running in your kernel space, i think nouveau's the only option
23:11imirkin_: if you're not, blob is way better at many things
23:12mek42_laptop: the people that are so concerned, why do they have such concern?
23:12imirkin_: why do people use linux and not windows or macos?
23:13imirkin_: i can't speak for others, but for me it's about openness
23:13imirkin_: nvidia blob is anti-open
23:17mek42_laptop: hmmm... i like openness, but I'm also planning to purchase pianoteq (that I'll run on linux) - right now I'm going to try nouveau, as I don't know that I'll notice the performance difference; the open argument helps, as I'm (slowly) working on reading an open code to add something it doesn't do yet (musescore, to support quarter-tones on playback)
23:18imirkin_: if you're interested in having a well-supported open-source driver, you should definitely aim for intel or amd hw
23:18imirkin_: both of those companies actually put resources forward towards such efforts
23:18imirkin_: while nvidia tends to stymie them