00:28mupuf: ok, I managed to overflow the period :D
00:28mupuf: the period is stored on a u8
00:40mupuf: playing with the PWM frequency is really painful for the ears
00:44mupuf: anyway, good data points ... and it is fucked up .... again
00:45mupuf: seriously, one would design this insanity!?
01:17mupuf: ok, my bad, I was storing a value too big for the frequency field
02:29mupuf: well, enough for today. I got a model that works in theory but not in practice :D Collecting more data to test my model on.
02:29mupuf: this is ... a big WTF
08:27mupuf: karolherbst: http://fs.mupuf.org/mupuf/nvidia/6b13_sawtooth-5.png when setting the PWM frequency to 65535Hz, it is pretty clear what is going on
08:27mupuf: I have a model and I have traces, so I am trying to make them match for both of them
08:28mupuf: sorry, it is 16384Hz
08:29mupuf: starting to collect data for 1kHz
08:29mupuf: and I'm out of here cause the sound of the PWM is unbearable
08:30Dezponia: mupuf: I just want to say that I read your nick as MuPDF all the time
08:40mupuf: Dezponia: I am not surprised ;)
08:41mupuf: my nickname is pretty old though, older than mupfd
08:42Dezponia: mupuf: Sue them! :P
08:43mupuf: It snowed quite a decent amount overnight, nice!
08:43mupuf: we are at more than 10cm it would seem
08:45Dezponia: mupuf: Hush you. I'm getting the tier changed on the work car today. Its been an experience staying on the road on summer tiers for the last few days
09:14karolherbst: mupuf: well, I had this yesterday already :3
09:15mupuf: you had what?
09:16karolherbst: the cap is funny though in the data
09:17karolherbst: anyway, good that you figured out why the data was so silly before
09:18mupuf: the cap is simply the divider
09:19mupuf: and no matter what the divider is, the slope is always the same
09:19mupuf: looks like 0 -> 0, 4096 -> $div
10:52RSpliet: mupuf: would you say that Hel froze over? #trump2016
10:53karolherbst: ohh, I am actually quite happy about it, no war with russia at least
10:53karolherbst: and assange has now friends in washington :D
10:53karolherbst: well "happy" is the wrong word
10:54karolherbst: both options were real bad
10:56karolherbst: ohh by the way
10:56karolherbst: imirkin: have fun over there
10:57Jaga: RSpliet: yea, been frozen over for a week now :p
10:58karolherbst: one week already
11:00Jaga: iirc yes
11:01karolherbst: I have snow since yesterday :)
11:02Jaga: playing catch up
11:07karolherbst: yeah, but we got new snow today
11:08karolherbst: at least
11:08karolherbst: now we have even snow in hamburg and not just outside the city :D
13:52mlankhorst: we had some snow on the roof this morning, think it melted
14:00mupuf: mlankhorst: hehe
14:00mupuf: RSpliet: yeah, I made this joke to myself when walking in the snow this morning :D
14:01mupuf: one dog had to jump to make strides because the snow was almost as high as him :p
14:03karolherbst: what a tiny dog
15:28karolherbst: mlankhorst: somebody told me you have some knowledge about display stuff?
16:00imirkin_: karolherbst: probably intel display stuff :)
16:01karolherbst: yeah, exactly what I need :p
16:01imirkin_: also Lyude has done a bunch of work around SKL watermarks, but perhaps is also more generally knowledgeable
16:01imirkin_: either way, #intel-gfx
16:07karolherbst: I already opened a bug.. but this isn't the kind of bug you just fix, because it is a stupid one
16:08karolherbst: need to find the right one who can deal with that
16:32Celmor: anyone know how I could get a different driver from nvidia included in my initramfs for ubuntu?
16:33Celmor: use case: loading that driver for a specific PCI device instead of nvidia but allowing nvidia to bind to the other PCI device
16:34imirkin_: sounds like a question for ubuntu, not nouveau.
18:45mlankhorst: karolherbst: a little
18:56karolherbst: mlankhorst: I have a silly issue with my intel HDMI port
18:56karolherbst: mlankhorst: https://bugs.freedesktop.org/show_bug.cgi?id=98491
18:58karolherbst: mlankhorst: vsyrjala already tired to help me with the issue, but afaik to fix that issue it is actually required to find the right people and ask the right question, I hoped that maybe you may be able to help
18:59karolherbst: the issue is basically, that currently drm doesn't know a way to read out the max clock supported by my on board DP -> HDMI adapter and the spec doesn't require anything
19:03foxtrot: Hey guys, quick question: If using the nouveau driver and I want hardware acceleration, I install libvdpau, correct?
19:03foxtrot: just like how I would on nvidia drivers
19:27mlankhorst: karolherbst: not sure about that :)
19:28karolherbst: foxtrot: vdpau is for video acceleration, like h.264 videos
19:32mlankhorst: karolherbst: seems like an issue that should be corrected during hotplug detect, if possible
19:32karolherbst: so to rather use the information from the edid instead of the adapter itself?
19:33mlankhorst: no idea
19:33karolherbst: well, the adapter is hard wired
19:33mlankhorst: but filter out modes not allowed
19:33karolherbst: that's on display connection time, right
19:34karolherbst: basically, drm falls back to 165MHz if the adapter doesn't follow this advanced spec where the adapter reports the max pixel clock supported
19:34karolherbst: I think TYPE_1_HDMI in drm or something like that
19:38foxtrot: karolherbst: Thanks, im pretty lost atm
19:38foxtrot: constant screen tearing and lag on both nvidia and nouveau driver
19:47karolherbst: foxtrot: what window manager?
19:48foxtrot: same in gnome
19:49karolherbst: did you check if you have hw accell for X at all? might want to check glxinfo and/or X log
19:50foxtrot: i didnt, let me do that
19:57foxtrot: karolherbst: i dont have glxinfo, but ive used it in the past
19:57foxtrot: would that program be available if i had the right stuff installed?
19:57karolherbst: mesa demso
19:58foxtrot: gotcha, grabbing it now
19:59foxtrot: glxinfo | grep "direct rendering" shows "direct rendering: Yes"
19:59karolherbst: check the vendor
20:00foxtrot: all say NVIDIA Corp.
20:00foxtrot: want me to switch to nouveau and try that too?
20:00karolherbst: k, then you run nvidia
20:00karolherbst: it is your choice
20:00foxtrot: I mean, i kinda just want to be able to watch youtube videos and move windows around
20:00karolherbst: usually you prevent tearing by enabling the tearing prevention within your window manager
20:02foxtrot: ive noticed if i drag windows around while a youtube video is playing, its realllly laggy
20:02foxtrot: and ofcourse the youtube videos always have lines through them
20:08karolherbst: well regarding stability you are better of with nvidia, so if you don't care about running an open driver I would go for that
20:08karolherbst: you may ask in #nvidia for help regarding vsync
20:13foxtrot: Thanks for the help bud, i appreciate it
21:00karolherbst: mupuf: I think I would rather have a constant within the subdev struct to store the max power values until we actually know more and how to get it if that field isn't set. I will move the bios interface to a header/entry splited thing to fix that memory leak and the subdev would only read that one entry out for now
21:01karolherbst: maybe it is always a constant value and we won't need a function to read it out anyway, on the other hand, a simple getter function only makes sense if we would have private members for our structs
21:01karolherbst: but I guess nouveau doesn't use that const field in struct const-cast hack?
21:01karolherbst: oh well, it isn't a library anyway, so we won't need to care about this
21:02karolherbst: (refering to your comment about modifying those values, but they are constant values)