01:11imirkin: skeggsb: any opinion about my question re clock arb code in nouveau?
01:21imirkin: [thanks for taking the initial fix]
09:02jossifer193: listen guys actually i am very sorry, i think you actually later understood what i meant, i have yeah too many nervous breakdowns continuously
09:04jossifer193: the isas are very intelligently composed actually, so there are so many algorithms possible, that all this is lost in traslation, when trying to explain it
09:08jossifer193: bit nvidia and amd have several ways close to in it's ISA to call the needed things which do very interesting performance stuff
09:08jossifer193: there is like multitude of ways for acheicing the same goal
09:09jossifer193: around 50 or 100 permutations to do generally the same theory, so it is hard to explain it in detail
09:12jossifer193: so we need to roughly agree with eahother and try to have a minor convo, which ones would be like to maintain, which types of them, in some timeline things can change and we'd end up knowing which in the end suites the best
09:16jossifer193: yeah in particular in scheduling, there is multiple ways to do almost in the end the same thing, using somewhat different flow each time, minorly changed, but it does the same, but maybe some methods work only faster
09:18jossifer193: hence i wanted to explain a bit of this world that i have been studying, so you'd be able to also think in the same level, and are capable to try to help filtering out the fastest method, i'd really need to do a speech about that somewhere, but i do not have resources at the moment to travel
09:21jossifer193: if you want to make it up, or so called bury the hatchet i am always agreering to do that, well shake hands somewhere in the future, and we'll decide who tries and how we do it, it needs some sorrt of agreement among programmers where we are heading
09:23jossifer193: i could try to use a webcam, and post the tutorial on youtube somehow also but...
09:23jossifer193: i'd prefer talking with programmers live about that stuff
09:28jossifer193: especially where this kind of sched code, which i am trying to author has some incorrect instruction, there are many correct calling ways, but if something is wrong, it messes up all and everything
09:33jossifer193: so listen mwk i am sorry allright, yes i was humiliated lot here, and this crew has had it's own drawbacks, because, they did nasty things to fill out their days, and just i had worrying and annoying fears where it leads, that normal people who are connected to this pointless crews, will take major drawback, and it was just what happened, one guy was no
09:33jossifer193: longer capable of going that way, and committed....
09:33jossifer193: suicide that is, and very tragic , since the nasty blokes did not learn how to live around him
09:35jossifer193: he was a strong fellow, and i have faught with similar things, i just could not allow such behavior around me everywhere
09:39jossifer193: in toher words all this crew truely haded the outcome was going nowhere but there unfotutnently that the only normal person in this crew just ....
09:42jossifer193: moral of the story, please have a filter on your friends, and stay distant of life burners, who just waste lives
09:46jossifer193: some people have no visions and do not feel strength in them, and they specilise in hanging out nastily exploiting every other normal person, i really did connect into one of such crews and shortly after they pushed up in my family before, well shortly after i was so shocked that i had to get out of there, or i end up dead , cause they eploited that bad
09:46jossifer193: being thsemselved absolute no point human beings
09:49jossifer193: because their initiative to stalk , harrass and all that was major, i lost evertyhing doing such a mistake in life, and now i only have mental illness papers and programming
09:49jossifer193: maybe it helps someone
09:49jossifer193: really just do not take every relations on if the persons do not pass tests
09:56jossifer193: we all envision and have hopes around beatiful people, it is oftenly happening that this is not the criteria to start real relations in life, if the person is rotten entirely, no matter how beatiful one is, it wil have fatal effect relations wise
10:00jossifer193: well the crew somewhere around the world those days searching other men stars to exploit and ruin their lives, really nasty guys entirely rotten stuff
10:03jossifer193: why i had to turn my back on them, was i did not see where those people could progress into some of a good spot position in life way with such activty
10:04jossifer193: i did not find any smaller chanche whatsover , that of a pointless folks mainly due to bloody activity which was not allowing progress to be made
10:10jossifer193: you can make work to be done in elegant way, but very crucial is who are your friends effecting your daily basis attitude, if those are normal or above things will work out hence too, for those who have internal content, they can occationaly then allready write any sort of code, instead of wasting time to stupidities
10:11jossifer193: you do yourself a daily plan, i do some programming, i do some sports, i meet nice people etc. and not let poor activity doing persons near close range
10:16jossifer193: i do not even need to practice much compared to other athles, i feel no joy in sports, i overtrained an generally am very good player in several events, they call me lazy, cause i have not gotten the engines and mentality sorted out, due the above things, and such energy conflits where i have continously suffered and only suffered
10:18jossifer193: allthough i felt all the needed strenght as young man to go to do professional sports too like a normal man, then the energy needed was sucked away from my body, constant persistent leak was actual
10:20jossifer193: it just starte as my dads visions, that sports athlete has no rights in life, terroristic vision of his own kid, i was not able to control the energey leak
10:24jossifer193: analysis show that i was because he is an endless sports soul, but born as an handicap, and terrorises people with abitlities
10:26jossifer193: this shocking detail came out later, when i inspected his leg structure, it was around the
10:27jossifer193: his bone above the toes is roughly around 2times thicker then normal
10:27jossifer193: and he has no way to work it out, underleg is not flexible enough to to professional sports
10:28jossifer193: throughout the life especially where there was allready anger towards me at other areas, he specialies in milking and terrorising sports athletes
10:32jossifer193: one anomaly in body has major effect on all the structure, it appareantly is the cause of his issues, so with no amount of surgeries this endless spots soul is not getting any results out unfortunently
10:35jossifer193: i dunno maybe 50percent of people have such screw ups, that sports isn't possible, but he is not able to handle it morally that situation
10:40jossifer193: such an issues were mainly present to erase my quite a noticable talent, if imirkin occationally tended to have sort of suspcion that something went wrong with me, and it was not my fault, yes that is the case
10:45jossifer193: agd5f: i was working out a shader sort of logic to interleave different wave with 1clock cycle delay each, this like other in that subject has several methods possible to do that
12:56gnarface: wtf was that all about?
12:58mwk: gnarface: beats us, but it's been a known occurence for years.
13:00mwk: some people have suggested that's what you get when you choose your ai training corpus poorly.
13:11gnarface: a fascinating notion
13:11gnarface: it almost seems like its trying to warn us about something...
13:35mwk: in retrospect, treating every color as A8R10G10B10Y8Y16 was a horrible idea
14:03tacchinotacchi: if anybody is italian here... that guy was a nice supercazzola generator
18:46stefanoz: Hello guys
19:10karolherbst: imirkin: I will look into the use nouveaus PMU code after secboot issues over the month
19:11karolherbst: mangix: clevo laptop?
19:11imirkin: karolherbst: awesome
19:12karolherbst: imirkin: I think gnurou last patches fixed a lot of things so maybe it already works
19:12karolherbst: have to investigate but the infrastructure seems to be there already
19:12karolherbst: and then the next step would be to add this one flag to enable reclocking on those
19:13karolherbst: I won't enable stuff like that automatically in _any_ case though, because hardware thermal throttling is most likely not set up on those boards
19:13karolherbst: so I want to implement the software throttling first before enabling anything by default
19:13imirkin: definitely would be good to get some testing before flipping it on for everyone
19:13imirkin: otoh, reclocking is a manual process anyways...
19:13karolherbst: yeah sure
19:14karolherbst: but we overheat the GPU
19:14karolherbst: under certain workloads
19:14karolherbst: I got to 102°C with nouveau with mine at full speed fans
19:14imirkin: nothing against automatic sw throttling
19:14karolherbst: yeah, only problem is, that the vbios table is stupid :(
19:14nyef: What happens if the GPU overheats? Does it "just" fry the chip, or is there a safety cutout that protects the hardware?
19:15karolherbst: nyef: there is a hw throttle mechanism
19:15karolherbst: it needs to be setup first
19:15karolherbst: it is done by vbios scripts on GPUs prior maxwell2
19:15karolherbst: we don't know if it is done on maxwell2 GPUs
19:15nyef: So it's probably not what causes the blob to complain that the GPU "fell off the bus"?
19:15karolherbst: there is a three staged throttle
19:15karolherbst: 1. cut clock signal by half
19:16karolherbst: 2. cut it by 7/8
19:16karolherbst: 3. remove GPU from bus
19:16nyef: Ah, okay, it possibly IS what causes that.
19:16nyef: Good to know. Thank you. (-:
19:16karolherbst: the factors are configureable though
19:16karolherbst: it just happens to be 1/2 and 1/8 of the original clock on my gpu
19:16karolherbst: + nvidia also clocks down to lowest states
19:16karolherbst: what nouveau doesn't do
19:18karolherbst: cutting the clock signal to 1/8 reduces power consumption by a half, which isn't much
19:18karolherbst: but enough most of the time
19:19nyef: I now half expect to find that, for one of my systems, either there's not enough thermal grease/padding/whatever between the GPU and heatsink, the heatsink itself is broken somehow, or the GPU fan is broken/disconnected.
19:20karolherbst: nyef: did you check the temperature of the gpu?
19:20nyef: Is that something that I can check after it falls off the bus?
19:20karolherbst: but usually nvidia is well enough in throttling the GPU, so that it won't fall of the bus
19:20karolherbst: if the GPU is off the bus, it is basically gone
19:22nyef: Right. I basically made a point of not stressing that GPU at all, and not using it on a day-to-day basis, and it's been fine since, but it'd be nice to know what's going wrong with it and if it can be fixed.
19:22karolherbst: you can monitor it with nvidia-settings
19:22nyef: But not a priority by any stretch of the imagination.
19:22karolherbst: but as I said: nvidia is good at throttling
19:22karolherbst: either you have a borked vbios, which disables all that stuff, or your GPU is basically broken
19:23nyef: But the GPU could be fine, and the cooling setup be broken.
19:23karolherbst: it doesn't matter
19:24karolherbst: the GPU won't overheat, if it is setup just fine
19:24nyef: I expect that I'll be tearing the entire system down this week, so I might find out what's going on then.
19:30stefanoz: Hey guys, I have a GTX 1050ti, can i ask you when nouveau will support it?
19:30mupuf: stefanoz: when nvidia finally releases its signed firmwares for it
19:31nyef: Crazy question, but... Is that firmware something that could be extracted from the windows driver, or is that not possible / too volatile / too much work for whatever reason?
19:32karolherbst: it is possible
19:32stefanoz: I am asking because I was using nvidia 375.26 and I have a "small" problem: when xorg shuts down I just see black screen, even when rebooting the computer, I can't see the motherboard logo. The unique way to get it to work is to reinsert the card on its pcie slot
19:33stefanoz: I would exclude h/w problem as with windows it is working great
19:34imirkin: stefanoz: define 'support'
19:34nyef: So, possible, but we're not doing it, "because reasons"?
19:34karolherbst: if anybody wonders, this was my week basically: https://i.imgur.com/1RHRNhg.jpg https://i.imgur.com/UuGIdxN.jpg :D
19:35imirkin: stefanoz: if you're looking for plain modesetting with no acceleration, it should either already work, or can be made to work with a very small tweak
19:35nyef: ... Is *that* where all our snow and cold weather got to?
19:35karolherbst: nyef: yeah
19:35imirkin: stefanoz: if you're looking for acceleration, i'd guess "never" due to nvidia's refusal to provide redistributable signed firmware
19:36karolherbst: nyef: was for a sking/snowboarding week in the Czech Republic
19:37stefanoz: Fuck them
19:37stefanoz: Anyway I am also looking for a generic driver with a 1360x768 resolution
19:37mupuf: karolherbst: lucky you!
19:37stefanoz: At least
19:37karolherbst: mupuf: :D
19:38karolherbst: so much snow and so cheap eating/drinking
19:38mupuf: snow is not a problem here, but mountains are ENOTFOUND
19:38imirkin: stefanoz: if you don't want accel, it should either already work with kernel 4.10, or if it's a GP107, should be easy enough to support by copying the GP106 logic
19:38karolherbst: mupuf: your 10 cm doesn't count :p
19:38mupuf: karolherbst: and fun people too ;)
19:38mupuf: 10cm at sea level :p
19:38mupuf: at (frozen) sea level :D
19:39stefanoz: imirkin: I'll try with 4.10
19:39imirkin: stefanoz: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/nouveau?id=1fe487d7d2858265e23f10fa6b4565112f9a17fe
19:39stefanoz: I would never say nVidia doing this
19:39imirkin: basically do that same thing, but for GP107 (0x137)
19:39karolherbst: mupuf: well, we had to much snow, nobody had any idea where to put it, so severel meter high snow "hills" were created
19:39stefanoz: I heard that was better than amd
19:39stefanoz: In fact of open source
19:39imirkin: if you're looking for well-supported open-source friendly hw, stick to intel and amd.
19:39mupuf: karolherbst: Helsinki has plenty of them ;)
19:40karolherbst: mupuf: nice :D
19:40mupuf: that's wide streets are so wide
19:40mupuf: that's why
19:40karolherbst: mupuf: but helsinki is much colder I guess, we had temperature around -5°C at day
19:40stefanoz: imirkin: maybe I will send back the gpu to amazon
19:40mupuf: no, it is quite warm again
19:40karolherbst: ohh I see
19:40karolherbst: yeah well
19:40karolherbst: there now too
19:40karolherbst: the day we drove home, the snow begin to melt
19:40karolherbst: perfect timing :p
19:41karolherbst: but now my entire body hurts, so I guess I won't do much today and tomorrow
19:43stefanoz: :( the gtx 1050ti has GP107 chipset, in this commit support is provided for gp106
19:43stefanoz: Anyway thanks
19:43stefanoz: I'll wait in the worst way
19:43imirkin: stefanoz: yes, my point is - you can just make a copy of that commit for GP107 instead of GP!06
19:43imirkin: should be the same
19:44imirkin: the chipset id is 0x137 (instead of 0x136)
19:44imirkin: [unless you're not a developer, in which case... yeah. just wait around until skeggsb gets his hands on one so he can test it out. could be a while though.]
19:45nyef: ... Or buy one to be delivered to skeggsb?
19:45imirkin: well, it's the sort of thing RH can probably afford
19:45nyef: (Probably wise to check before ordering one for that, yes.)
19:45imirkin: but yeah, i'm sure making one appear on his desk is also not a bad way to get things to be supported
19:46stefanoz: I'm not a kernel developer but I know various languages and interested in lower level
19:46stefanoz: i'll try editing those lines
19:47stefanoz: Hey guys
19:47stefanoz: Sorry for the rage quit
19:48imirkin: ... and stay out! :p
19:48stefanoz: Android application called onDestroy() randomly
19:48stefanoz: Anyway thanks for the help
19:48imirkin: that's convenient.
19:48imirkin: good luck, let us know if it works and/or submit a patch
19:48stefanoz: I have to go now, byee
19:49stefanoz: Yes of course
22:37imirkin: bonbons: you had a NV1A right?
23:17bonbons: imirkin: I do (IGP, in nforce chipset)
23:18imirkin: bonbons: did you ever have flickering issues?
23:19imirkin: i just sent a patch fixing the clock for nv1f, and simultaneously noticed that it was done wrong for nv1a as well, so i touched it up
23:20imirkin: in case you want to test it out, you can find the patch at https://github.com/skeggsb/nouveau/commit/883bf734d0c30dab7471520530c9a03fdee448ff.patch
23:20bonbons: not that I remember (I don't power it on that often). The biggest issue I have with it is missing DDC (proper I2C channel not found/associated/whatever)
23:20imirkin: ah =/
23:22bonbons: most recent kernel run on it: 3.19, quite some time ago!
23:23imirkin: well you don't have to test it if you don't want to - not exactly a ton of nv1a (or nv1f for that matter) users out there
23:29bonbons: would the flicker be visible under console or only under X (and maybe starting with which kernel version if known)
23:30imirkin: bonbons: tbh, i'm not sure. people were just reporting flickering. i think it'd be resolution-based.
23:30imirkin: it'd be present under all kernel versions
23:30imirkin: but i only ever heard about it from nv1f people. so really, if it doesn't *cause* flickering (or other unwanted effects) on nv1a, i'm happy.
23:33branau: Hi everyone, I'm sure most of the people who drop in here are just looking for support but I'd like to say thanks! I couldn't even get the official nvidia drivers to work on my laptop but the nouveau ones work just fine :)
23:34imirkin: branau: surprising, but great to hear
23:34imirkin: branau: what GPU do you have out of curiousity?
23:35branau: imirkin yeah, I'm a bit surprised myself. I've got a GeForece GTX 950M
23:35imirkin: cool. i think that's a GM108M or something?
23:35branau: I'm on Fedora 25 so it might have something to do with wayland perhaps?
23:35branau: Not sure, how would I check it?
23:35imirkin: er no, GM107M
23:36imirkin: either way, if you're looking to play video games where performance matters, i'd highly recommend grabbing linux 4.10-rcN as well as mesa 17.0.0-rc1
23:36imirkin: branau: lspci -nn -d 10de:
23:36branau: GM107M it is
23:36branau: But I'm not a huge gamer, just casual
23:37branau: I understand this is like a low to mid-level chip, right?
23:41nyef: I imagine that it's a better chip than my GK107M.
23:43branau: I wouldn't know :P I don't think it's high end though, I only paid around $600 USD for the laptop and got it new
23:47nyef: GK107M is one of the Kepler family devices, GM107M is a Maxwell, and Maxwell is the family after Kepler.
23:47imirkin: branau: ok - well with the boot-time configuration, that GPU will be slower than the intel gpu
23:47imirkin: branau: however with reclocking, it will likely be faster
23:48imirkin: 4.10 gives you the ability to reclock it manually
23:48imirkin: and mesa 17.0 includes some critical performance improvements for maxwell-family gpu's
23:50branau: imirkin is it a pretty big difference? This laptop is mostly for play, I have a second laptop I use for coding and work that I consider my stable laptop
23:51imirkin: branau: well, the difference from reclocking is probably be on the order of 5-10x, the difference from the mesa change should be another 2x.
23:51imirkin: nyef: memory type dominates those changes... e.g. a GM107 with DDR3 won't compete against a GK107 with GDDR5
23:53nyef: Ah, okay.
23:58nyef: eBay, 660M 2GB DDR5, for my hardware: $130. 980M, 8GB DDR5, $780.
23:58nyef: Clearly, RAM is expensive. d-:
23:58branau: imirkin whoa, that's pretty crazy. 4.10 isn't in the stable branch though for fedora 25, right?
23:59branau: I think I just updated to 4.9
23:59imirkin: that seems pricey
23:59nyef: 4.10-rc5 is expected "tomorrow", right?
23:59imirkin: branau: no, it's hardly stable software. no clue what the fedora update policy is either.