00:05orbea:is stuck on 4.6.3 :/ karol's branch needs 4.7 and r8168 needs 4.6....
00:10Tom^: orbea: karols branch is 4.7
00:10Tom^: or well he has a 4.6 too
00:11Tom^: oh you are stuck on it because of r8168
00:11orbea: not network without it
00:11Tom^: so yea just use the stable 4.6 branch then
00:11orbea: that has the reclocking?
00:12orbea: i missed that, thanks
09:21karolherbst_work: mupuf: do you thinkg REing the new fan table is "easy" enough for a newbie task?
09:22karolherbst_work: I actually created the trello card already
09:37mupuf: Well, sure, it is pretty obvious when something changes ;(
09:37mupuf: you can ask the blob or read the regs
09:38mupuf: and you can easily make charts
09:38mupuf: so, not too bad
09:38karolherbst_work: I would have looked into it, but… a bit hard with my system
09:39karolherbst_work: I think I will target that sw overheat protection next myself, when I find some time
09:39karolherbst_work: this is somewhat important, allthough I should prefer the power budget stuff
09:40karolherbst_work: I guess you won't find time for either this year?
09:43mupuf: this year, yes
09:43mupuf: but not during the summer, too much going on and travelling
09:43karolherbst_work: I finally created this one: https://trello.com/c/Z3lSDA3m/148-adjust-clock-reading-code-to-changes-since-kepler2
09:43mupuf: but I definitely slowed down at work, not so many insane hours anymore
09:44mupuf: oh no, not again
09:44mupuf: anyway, bbl!
10:34inglor: karolherbst_work: latest kepler reclocking fails to build on me. is the branch still "stable" ?
10:35Tom^: inglor: whats the error?
10:38inglor: Tom^: https://gist.github.com/inglor/a837747004f566bed6dea7ccd4ea1da7
10:38Tom^: inglor: and you are building against kernel 4.7?
10:39Tom^: "'/usr/lib/modules/4.6.5-2-ck/build'" or well no you arent :p
10:39inglor: shoudl it be ?
10:39Tom^: you need the _v4.6 to build against 4.6
10:39RSpliet: karolherbst_work: you might want to specify the problem in greater detail... I had a GTX780 reclock and work fine w/ nouveau
10:41Tom^: huh what changed? i can read my clocks on the gk110b ?
10:42inglor: bah Tobias is lazy..
11:49karolherbst_work: RSpliet: what do you mean?
11:49RSpliet: I mean that your Trello card does not explain *what* the problem is in detail
11:50RSpliet: on reading it a second time I presume you mean "using the blob/official driver"... making this a problem in envytools?
11:51RSpliet: are you trying to parse the PLL values? or is this a perf counter problem?
12:08karolherbst_work: you mean the clock thing?
12:08karolherbst_work: mhh right, it is PLL related
12:08karolherbst_work: I have no idea what the issue is. nvkm_clk_read just returns wrong stuff
12:15karolherbst_work: RSpliet: in fact, it is just about reading out the clocks while nvidia is used
13:28drathir: Lekensteyn: notice and read ^^
13:35Lekensteyn: drathir: you are really testing my memory right :P Luckily there is /lastlog :)
15:35orbea: I am running into an issue with SDL2 + DRI3 with ArxLiberatis and one of their devs seems to think the root problem is in the graphics stack so maybe you guys might have an idea with this backtrace? http://pastebin.com/0ZmdWSgz And the related issue (Video input freezes when playing fullscreen) https://bugs.arx-libertatis.org/arx/issues/873
17:53karolherbst: I will have time too look at the P.0x50 table today. This is how far I've got yet: https://gist.github.com/karolherbst/7532bf9ebb0298fe3ca69b24802585bc
17:54karolherbst: it does affect the temperature where nvidia starts to downclock the gpu on the driver side, that's what I am pretty sure of, and the speed/step size can be somehow affected with the other bytes
17:55karolherbst: maybe somebody of you has any ideas
18:28karolherbst: mupuf: I thought writing a new tool for REing stuff might be helpful: https://gist.github.com/karolherbst/ba17baf8add94acee471a25b99a92f23
18:31karolherbst: much better now: https://gist.github.com/karolherbst/ba17baf8add94acee471a25b99a92f23
18:31karolherbst: a pattern :O
18:32karolherbst: every 200µs it clocks down by one step I think
18:32karolherbst: two steps
18:32karolherbst: unknown step size
18:33karolherbst: I guess there is a formular
18:33karolherbst: like every 200µs decrease by 50MHz
18:33karolherbst: and it just checks which cstates fits best
18:38karolherbst: 200: c8
18:38karolherbst: k, found
18:43karolherbst: k, doubled the c8 and now I got this: https://gist.github.com/karolherbst/4f044d9ca59a2366d9c8a57f952ba282
18:43karolherbst: very good
19:04karolherbst: the 30 is odd
19:04karolherbst: it does change the step size somewhat
19:04karolherbst: but it is a flag byte
19:04karolherbst: odd thing
19:37karolherbst: this table is annoying :/
19:37karolherbst: there are flags and then other fields with meanings depending on those flags
19:41Tom^: proper late night food, http://i.imgur.com/kPNyHQK.jpg =D
19:42karolherbst: how sloppy that looks
19:42Tom^: haha it was leftovers, and i was hungry
19:43karolherbst: the cucumber pieces aren't even equally thin
19:43Tom^: nop, sliced with a big knife in the air
19:43karolherbst: I figured
22:19ahmed751995: excuse me , Iam using trisquel , how can i install bumblebee with nouveau
22:20NanoSector: nouveau shouldn't need bumblebee
22:21NanoSector: nouvea has optimus support built in, use "DRI_PRIME=1 program" to use it
22:23ahmed751995: NanoSector: why did fedora mention about nouveau and bumblebee in wiki (https://fedoraproject.org/wiki/Bumblebee)
22:25karolherbst: because they aren't up to date
22:26karolherbst: also, bumblebee has worse perf compared to prime
22:26karolherbst: ahmed751995: https://nouveau.freedesktop.org/wiki/Optimus/
22:26NanoSector: nouveau with prime is actually faster than bumblebee on my laptop...for the 5 seconds it works
22:27karolherbst: well, bumblebee won't be more stable
22:27NanoSector: but hey, it's heading in the right direction :)
22:27NanoSector: it is, actually
22:27NanoSector: well, not with nouveau
22:27karolherbst: no shit
22:28NanoSector: i see no point in running bumblebee with nouveau
22:31karolherbst: it actually result in worse perf as I said already
22:32karolherbst: ahmed751995: well, the wiki also says that Prime is better than bumblebee by the way
22:34ahmed751995: karolherbst: :D iam sorry i didn't read wiki well
22:34ahmed751995: thanks all alot for helping
22:39inglor: Do you have any recomendation on small bugs to pick up so I learn the code?