18:24DragonsTear: hello, https://bugs.freedesktop.org/show_bug.cgi?id=94990 (Geforce GTX 970 not booting on kernel > 4.6), is there any way to help with this bug? seems "stalled"
18:27imirkin_: DragonsTear: try to get gnurou's attention - he's the one who contributed the secure boot code
18:29imirkin_: DragonsTear: ultimately a little late now, but if you're looking for a GPU with open-source support, definitely go with AMD.
18:31imirkin_: with the latest generations, nvidia holds all the keys and aren't particularly interested in letting anyone else play
18:31DragonsTear: imirkin_: yes... :( it's so fucking frigging sad...
18:31Yoshimo: the worst is that a channel of people dedicated to making a driver for cards from a certain vendor reccomends the competitors products
18:33imirkin_: well, i certainly can't recommend people getting nvidia ... it may be fun to hack on the driver, but that has nothing to do with end-user usefulness
18:33Yoshimo: which leads me to the good old question when will we be able to reclock a 980 gnurou?
18:33imirkin_: [and in any case, i've avoided making any significant improvements for maxwell+]
18:34DragonsTear: well... at least he's honest.
18:35imirkin_: note that i also don't speak for everyone here
18:35DragonsTear: i just had high hope as it sounded like could barely work
18:35DragonsTear: notice the barely. XD
18:36orbea:is glad kepler is still being worked on :)
18:37Yoshimo: bugs will get solved eventually, just a matter of time :P
18:38imirkin_: eh, dunno. a handful of bugs i've kinda lost hope on...
18:40orbea: imirkin_: you ever have a chance to try tracing that video segement I sent you with the glitched graphics?
18:40orbea: no worries, was just curious :)
18:42orbea: at least for videos like that I just switch over to mpv with software decoding...
18:43imirkin_: i really need to complete my locking thing
18:43imirkin_: including vdpau, not just the graphics entrypoints
18:47orbea: btw, somewhere in my change from dri3 and karolherbst's reclocking branch a lot of of the slow parts with pcsx2 + xenosage vanished, but others were only minorly affected...maybe there is a clue in that?...
18:48orbea: I will boot an older kernel to test later...
18:49orbea: however other games work great...
18:49orbea: gta san andreas for instance
22:27netz: NV110's EXA, XRender, Xv², and '3D tesselation programs' are currently marked as 'WIP'; where could I look up the specifics of this WIP state?
22:28imirkin: netz: https://github.com/imirkin/xf86-video-nouveau/commit/abf0933a236b6069f42f86ad5b0bf5bbab28e0d6
22:28imirkin: that's for the ddx
22:28imirkin: for 3d tess... hakzsam's been trying to make some sense of it, but it's slow-going
22:29netz: in laymans terms, what does this mean ? games with 3d graphics, or something like 3d tv stuff?
22:30imirkin: in layman's terms, buy AMD if you want open source support
22:30netz: I know that bit :)
22:30imirkin: tessellation is in reference to GL_ARB_tessellation_shader, an opengl extension that was made core in OpenGL 4.0
22:31netz: see, I'm a layman as far as nvidia/nouveau is concerned, but I'm not totally ignorant :)
22:31imirkin: EXA/Xrender/Xv are in reference to the relevant X11 extensions, which means that you basically can't use the nouveau ddx. but you can use the modesetting ddx instead.
22:31netz: how does nouveau fit into this wayland/mir setup?
22:31imirkin: (which in turn relies on nouveau's GL library to implement X11 acceleration primitives)
22:32imirkin: well, wayland is just a protocol afaik
22:32imirkin: however weston is a compositor that uses GL, and would thus use mesa's GL library
22:32imirkin: however it definitely does not make use of tessellation, nor will it ever
22:32netz: do you guys have any allegiance either way? Intel put its money on wayland over mir.
22:33imirkin: i have no plans on using anything other than X11 until an event occurs that requires me to move away from it. i don't anticipate that happening for another decade.
22:33imirkin: [as always, i speak for myself]
22:34netz: yeah, I see x being around for a goodly long while after today :)
22:34imirkin: i'm also the guy who hasn't changed anything of significance in his desktop setup for over 15 years.
22:35imirkin: so you might not want to get your "what's the fanciest new desktop" advice from me
22:35netz: oh, I'm not interested in fancy. I use i3 myself.
22:36netz: would like to use sway, but apparently the blob drivers can't do it yet.
22:36netz:has no idea what that is :)
22:36imirkin: i used to use AfterStep, but then they broke it in like version 1.6 or so =/
22:38netz: so, supposing the most graphically demanding game I play regularly is minecraft (and no, I'm not some tween :P) would you say nouveau on a EVGA GTX 750 ti is good enough for me?
22:38imirkin: are you a pre-tween? :p
22:39netz: pushing thirty, actually :/
22:39imirkin: anyways, right now there's no reclocking on GM107, so chances are at the lowest perf level, it won't run particularly well
22:39imirkin: if you're looking for good support from nouveau but still want to buy nvidia, i'd recommend a kepler
22:39imirkin: if you're looking for the best open-source support, definitely stick with AMD
22:40netz: ok, and kepler is before/after GM107?
22:40imirkin: kepler is before
22:40imirkin: (M = maxwell)
22:40imirkin: also, the GM20x chips come with signature verification
22:40netz: I actually still have a G650 I think.
22:40netz: imirkin: meaning? won't run on nouveau at all?
22:41imirkin: which means that we're basically at nvidia's mercy for acceleration, reclocking, etc
22:41netz: I'm getting really sick of this signed and locked firmware shit :/
22:41imirkin: they have, after ~1 yr of the cards being released, provided some firmware for 3d accel, but not for reclocking, so the GPUs are still glorified heatsinks
22:41imirkin: er, s/heatsinks/heat-producers/
22:41imirkin: which is also why i have little to no interest in doing anything for maxwells
22:42imirkin: unfortunately GM10x's get swept up with that even though they have none of the signature bs
22:42imirkin: well - i don't really have a problem with signed/locked/whatever firmware. i just have a problem with it not being made available in a redistributable form.
22:42netz: I swear... we need to file a lawsuit or something XD
22:43imirkin: go for it.
22:43netz: eh. I work for just a bit above minimum wage. I aint got the cash for that :/
22:44imirkin: even if you were considerably above that, your funds would still pale in comparison to nvidia's
22:44imirkin: and as we know, the american justice system is the best system that money can buy...
22:44netz: heh, yep.
22:44imirkin: so the best thing to do is just take your business elsewhere
22:45imirkin: the GTX 650 should be moderately well supported though.
22:45netz: wallet voting eh?
22:46netz: I'll keep that in mind on my next build.
22:48netz: thanks for the Q&A
22:48netz: one thing: this firmware you mention, is it like embedded in libGL.so or sommat?
22:49imirkin: the usual definition of firmware is software that runs on not-your-primary-CPU
22:49imirkin: (unless it's CPU firmware, which is a fun little corner case for that definition)
22:50netz: yeah. microcode
22:50netz: µcode, as it were :)
22:50imirkin: sure, but often-times these are full-fledged cpu's...
22:51imirkin: running a little RTOS and whatnot
22:51imirkin: anyways, that firmware is uploaded by the kernel module, and tends to live in separate files (nouveau), or embedded in the kernel module (nvidia)
22:51netz: gotcha. I've ran into that before.
22:51imirkin: it enables key gpu functions, like switching between hw gpu contexts
22:52imirkin: as well as control of the fans, memory reclocking, other clock control, etc
22:54netz: so, if they were to release reclocking firmware, it would be a separate bit from the portion you already have for nouveau?
22:56imirkin: right, well we'd have to add code to interact with it
22:56netz: obviously :)
22:56imirkin: but that firmware would run on one of the GPU's internal CPU processors
22:57imirkin: [it has multiple ones]
22:57imirkin: [for simplicity]
22:57netz: I suppose that if someone were to extract said firmware from the blobs ya'll wouldn't be able to accept it for fear of taint?
22:58imirkin: well, we def couldn't redistribute it
22:58imirkin: i did, earlier, write a script to auto-extract firmware
22:58imirkin: back then it was easily locatable
22:58imirkin: however they changed things around now so that it's harder to find.
22:58netz: not so easy nowadays eh?
22:58imirkin: [it's unlikely that the change was in order to make it more difficult to find, just turned out that way]
22:59imirkin: but e.g. video decoding (i.e. h264, etc) works with that auto-extracted firmware
22:59imirkin: since i just ship a script, and the end-user runs it against the blob that they download separately, pretty sure we're on solid legal ground
23:01netz: cool stuffs.