01:12Echelon9: Thanks for the merge imirkin_
07:44mwk: well, that's a lot of pull requests...
09:23karolherbst: that's the unimportant bits
09:23karolherbst: but it at least may show what certain DXIL instruction are for
15:00imirkin: nyef: http://www.fiercecable.com/cable/3d-tv-officially-dead-as-sony-and-lg-stop-making-sets
15:07nyef: Hunh. Yeah, I can sortof see that happening.
15:11nyef: Let's see, mid-to-late-'80s, early-'10s... we might see another surge of interest in stereography when the 32-bit time_t's wrap around.
15:13nyef: There was basically a short run of time when they were making PS3 games for 3D, but it wasn't long. Maybe it's a once-per-generation thing?
15:19imirkin: it's the sort of thing that sounds cool, but is incredibly difficult to execute on
15:19imirkin: and i think people have moved onto VR
15:28nyef: ... which is also the sort of thing that sounds cool, but is incredibly difficult to execute on.
15:31imirkin: it's easier to get people to put glasses on for vr than for 3d tv
15:32imirkin: i guarantee you that 3d tv without glasses would have been pretty successful
15:32karolherbst: I thought there were some without glasses
15:33imirkin: i'm not aware of any such technology
15:33karolherbst: no idea how much of a hoax that is: http://arstechnica.com/gaming/2016/02/the-glasses-free-technology-that-made-me-believe-in-3d-tv-again/
15:35karolherbst: mhh, but only with one image
15:35Tom^: i thought i saw some 3d tvs on youtube from china, prototypes that is without the need for glasses
15:35Tom^: https://www.youtube.com/watch?v=9STBsPHIEPA was this one i was thinking of
15:36imirkin: if those things really work
15:36imirkin: and not some weak effect
15:36imirkin: then that could be huge
15:36Tom^: *shrug* whish i could look at one in person :D
15:36karolherbst: well you have to start somewhere
15:36karolherbst: even if with cheap tricks :D
15:37imirkin: but that's the thing - until it's the real thing, no one cares.
15:37Tom^: no idea how it works tho
15:37Tom^: isnt 3d just 2 different frames on each eye. how do you achieve that without glasses? =D
15:38karolherbst: Tom^: how do you achive that without 2 frames :p
15:39karolherbst: you can actually turn a 2d image into a 3d one, as long as both display the same, cause you can just filter the angle and create two 3d able images
15:40karolherbst: and for the distance we are talking about, any object between both eyes don't matter that much anyway
15:41karolherbst: so I would guess some kind of bending lenses or other ways of distorting may apply here
15:41Tom^: and funnily enough perhaps this is one of those 3D effects i might be able to actually watch
15:42imirkin: i didn't say i had a solution... just pointing out that until there is one, no one will care.
15:42Tom^: since i sort of require normal glasses on one of my eyes since birth. its half as bad as the right one hence my brain sort of mostly just use the left one
15:42Tom^: and im to lazy to wear normal glasses, and thus 3d movies with 3d glasses my brains kinda filters it out
16:58nyef: Can a monitor EDID change between dual-link and single-link DVI cables?
17:00karolherbst: nyef: I am sure it can, but if it against the spec? I doubt it
17:00karolherbst: we had a situation here, where the EDID changed between different connectors, but they didn't update the checksum....
17:01nyef: Okay, so I still need to get a dual-link DVI cable to work with.
17:10karolherbst: CTS is open source now
17:10karolherbst: uhh it is based on dEQP?
17:11hakzsam: imirkin: this is for you! ;)
17:11karolherbst: here is the opengl stuff: https://github.com/KhronosGroup/VK-GL-CTS/tree/master/external/openglcts
17:14karolherbst: but the other question remains valid: are we allowed to claim OpenGL 4.4/4.5 support?
17:14karolherbst: at least everybody can fix issues now, which is a good thing
17:17imirkin_: nyef: i wouldn't be surprised. it definitely changes between dvi-d, vga, hdmi, and dp.
17:17imirkin_: hakzsam: i already found it yesterday
17:17imirkin_: hakzsam: played around with it a bunch. we have bugs :(
17:17hakzsam: imirkin: hehe
17:19karolherbst: imirkin_: serious bugs? or just random trivial stuff
17:20imirkin_: karolherbst: if you hit them, they're serious. if you don't, they're trivial :)
17:20imirkin_: stupid bgra4 being a thorn in my side again.
17:20imirkin_: but this time it's my fault :)
17:21imirkin_: [hmmm... need to double-check the spec. i may be able to weasel my way out of it...]
17:25mupuf: Karol, i am in contact with Khronos to handle for the x'org foundation the submission of results
17:26mupuf: Nvidia validates each arch (kepler 1, 2)
17:26imirkin_: makes sense.
17:27imirkin_: the arches are different
17:27mupuf: I have the hw to run the tests and generate results
17:27imirkin_: very easy to have a dumb bug on one and not another
17:28mupuf: Will contact Khronos again.
17:28imirkin_: we're nowhere close to passing all the tests
17:28imirkin_: and it will take a substantial effort to get to passing.
17:28karolherbst: mupuf: nice :)
17:28karolherbst: well, we can concentrate on 4.4 first
17:29imirkin_: my guess is that it'll take me a year if no one steps up to help.
17:29imirkin_: [not a year of full time work... a year of calendar time]
17:31karolherbst: gsoc project: make nouveau pass the CTS
18:12imirkin_: btw, that's not the whole cts test suite
18:12imirkin_: it doesn't include the GL3 tests, for example
18:12imirkin_: (i think)
18:32karolherbst: why does the CTS tries to use WGL
18:51karolherbst: imirkin: what is the best way to run those tests? I either get hardly understandable output or silly crashes :/
19:22karolherbst: this is like super odd...
19:36karolherbst: even calling the fini of secboot doesn
19:36karolherbst: 't help to reset the pmu
19:37karolherbst: gnurou: any ideas? I was able to load the nouveau pmu code after performing secboot after I put the machine into suspend and waking it up, but I don't get it to work without suspending, any ideas?
20:17whompy: imirkin: bummer on gl33. I saw stubs I think the Readme about it, so I was going to try killing my laptop with CTS.
21:10dmj_s76: skeggsb: Hey, do you know anything about the mouse cursor graphic being stuck in the corner of the screen on Pascal GPUs?
21:12airlied: ctx might run on 3.3 if you use --deqp-case=GL33* or something like that, however it'll fail badly, it's really only designed for latest GL
21:37imirkin_: dmj_s76: should be fixed in 4.10
21:37dmj_s76: imirkin_: Are there any cherry pickable commits that fix that?
21:37imirkin_: dmj_s76: probably... sec
21:38dmj_s76: Looking to see about possible backports for a distribution using 4.8.
21:38imirkin_: dmj_s76: https://github.com/skeggsb/nouveau/commit/b94985176af536ab752a63baecdee263ffcce63f
21:38imirkin_: and the 2 commits that came prior
21:40imirkin_: [note that's not a kernel tree, but you can find those commits upstream as well, just diff commit hashes]
21:41imirkin_: karolherbst: you have to force GL 4.5
21:56dmj_s76: imirkin_: Thanks.
22:22Echelon9: imirkin: Reviewing nvabios, I noticed there are two bios opcodes that aren't included in envytools' nvabios which are tracked in kernel
22:22Echelon9: Is adding the two opcodes a worthwhile task?
22:23imirkin_: Echelon9: which ones? new or ancient?
22:23imirkin_: Echelon9: either way, you can go ahead and add them
22:23imirkin_: it's not an immensely important task... skeggsb wanted to have some unification between the two codebases, but ... it hasn't happened
22:24Echelon9: It's INIT_ANDN_REG (opcode 0x47) and INIT_OR_REG (opcode 0x48). Added to the kernel around 2014 it appears by skeggsb
22:25imirkin_: yeah, definitely add those :)
22:25Echelon9: OK. Great I've found something useful and decently discrete to start with.
22:31Echelon9: And opcode 0x3a should be renamed from dp_condition to generic_condition to match the kernel
22:31Echelon9: Was changed by skeggsb at the start of 2016
22:37imirkin_: yeah, that'd be fine to do as well