02:46 imirkin: ok... so the issue with no sampler for a texture is very cleanly reproduced inside a modified quad-tex. at least it's consistent.
02:51 imirkin: interesting. it seems to use sampler bound to slot 0 no matter what...
02:56 imirkin: oh. that's because i tell it to use sampler slot 0 apparently...
02:56 imirkin: that sure is naughty.
02:56 imirkin: i probably need to make use of the "special" sampler slots for this? hm...
18:25 pmoreau: skeggsb: I can confirm the MMU patch at https://github.com/skeggsb/nouveau/pull/1 fixes graphical artefacts and various PT_NOT_PRESENT/NULL_DMAOBJ error messages on my MCP79.
18:26 pmoreau: Waiting for some feedback for https://bugs.freedesktop.org/show_bug.cgi?id=105173 and https://bugs.freedesktop.org/show_bug.cgi?id=105174, to see if it helps them as well or not, but if it does, you might want to grab it for 4.16 and 4.15.
20:01 Rucha: Hello! I am Rucha and I am interested in this question "reduce performance when not all power connectors are plugged". Which GPU should I start this task with? I am new to hardware and GPU
20:02 imirkin_: mupuf: --^ people come in and ask about that every so often. i'm not sure it's even possible...
20:02 imirkin_: Rucha: unless you're desperate to do that task, i'd recommend picking something else
20:03 karolherbst: imirkin_: there should be a GPIO
20:03 karolherbst: but I am not sure if the machine even boots
20:05 Rucha: imirkin_ which task do you recommend for a beginner?
20:05 imirkin_: well ... let's start with what you have
20:05 imirkin_: (a) what are your capabilities
20:05 imirkin_: and (b) what hardware do you have
20:06 imirkin_: and perhaps most important (c) what's your time commitment
20:06 adya: I am also studying about this. I found that GeForce 7600GS does boot in reduced performance! But its quite an old GPU
20:07 Rucha: I can get all the hardware. It's a college project
20:08 Rucha: I wish to complete one task till Mid March.
20:10 Rucha: I wish to complete one task by Mid March. Then I will continue
20:11 imirkin_: ok, well mid-march isn't a long time away
20:11 Rucha: My capabilities are C, OOP and Python
20:11 imirkin_: 2 weeks, as it were
20:11 imirkin_: what's your level of knowledge of GL, or alternatively, linux kernel?
20:12 Rucha: I have basic knowledge about the linux kernel, but I am willing to learn about it!
20:13 imirkin_: well, there's a long list of projects on a trello board. pretty much every time i tell someone to take a specific project, they disappear never to be heard from again
20:14 Rucha: I don't think that will be the case!
20:14 imirkin_: that's what everyone says :p
20:14 imirkin_: just before they vanish into the darkness...
20:14 imirkin_: ok. so here's one.
20:14 imirkin_: it should be pretty easy to do.
20:14 Rucha: Yes please
20:14 imirkin_: https://trello.com/c/YnhHNblO/100-nv50-nvc0-fix-fbo-blending-formats-glrgb10
20:15 imirkin_: (and when i say "pretty easy" i mean "should only take a beginner like a week to do", not "it will take 30 seconds")
20:16 imirkin_: only do it for nvc0, not for nv50
20:17 Rucha: Could you provide with some links to start studying with and start pointers?
20:18 imirkin_: right, so you'll need some fermi or later piece of hardware to test
20:18 imirkin_: mesa is available at https://cgit.freedesktop.org/mesa/mesa
20:18 imirkin_: if you have any other questions, feel free to ask here
20:19 imirkin_: there's no such thing as a stupid question
20:19 Rucha: Sure will use it!
20:19 Rucha: Yeah since I am just beginning, going to come up with pretty basic questions
20:19 karolherbst: imirkin_: are there piglit tests for that?
20:19 imirkin_: yeah, it's mentioned in the subject :p
20:20 Rucha: Also, any particular GPU I should start with?
20:20 karolherbst: ahh
20:21 imirkin_: Rucha: any fermi or later gpu should do
20:22 Rucha: @imirkin_ Thanks! I'll start to study
20:24 Rucha: Could you please guide a little about the flow of the project? As this is the first time I am working with hardware. It would help a lot if you told me the step1
20:25 imirkin_: step 1: install mesa from your own build and run glxgears to make sure it works.
20:26 Rucha: Thank you so much! Will update my progress!
20:26 pmoreau: imirkin_: BTW, do we have some kind of contribution guide on the wiki (for example, where to send the patches, etc.)? I had a quick look around but didn’t find one.
20:26 imirkin_: feel free to write one.
20:27 imirkin_: most people who are in a position to write patches can work this sort of thing out.
20:28 pmoreau: Okay, so we have none, I wasn’t sure about that.
20:28 imirkin_: (or can quickly ask here, the number of times i've had to field that question has been ... never. so it doesn't come up too often)
20:29 pmoreau: Right. I was thinking of linking to it for the guy who opened a pull request on Ben’s repo.
20:29 imirkin_: i have little concept of what makes it "difficult" to contribute to a project beyond the practical difficulty of the task at hand
20:29 imirkin_: if there are more concrete things that have nothing to do with the actual tasks being done, feel free to write a guide.
20:29 imirkin_: yeah, that's why i generally don't like hosting things on github
20:29 imirkin_: people start thinking that they can send pull requests...
20:30 imirkin_: (and it's also why i don't like contributing to projects that are solely hosted on github... i have to send pull requests.)
20:30 pmoreau: I don’t know if one can disable that or not. At least you can for issues and wikis.
20:32 pmoreau: And even if there was a document saying how things work, that doesn’t mean that people are going to read it. But on the other hand, we can’t expect people to guess that we don’t do pull requests but send everything through the ML.
20:33 imirkin_: when you're opening the very first pull request
20:33 imirkin_: there's gotta be a thought going through your head
20:33 imirkin_: thinking "huh, maybe linux kernel is developed differently"
20:33 pmoreau: True, though, maybe the repo is brand new? :-D
21:47 mupuf: imirkin_: why wouldn't it be possible? The easiest option would be to check the gpio's status (if present) and disable 3D acceleration
21:47 imirkin_: mupuf: how to test?
21:47 mupuf: imirkin_: tesla GPUs boot just fine without the power connector
21:47 mupuf: but the blob refuses to boot in this case
21:48 mupuf: we could disable acceleration instead
21:48 imirkin_: ok
21:48 imirkin_: can you specify which GPIO
21:48 imirkin_: and some examples of tesla boards?
21:48 imirkin_: i have a FX 3700 (G92) with a power connector
21:48 imirkin_: i can try not plugging that in and seeing what happens
21:48 mupuf: sure, you'lll see it boot just fine
21:49 mupuf: I'll take a quick shower and be back
23:13 mupuf: imirkin_: GPIO function 16 = Thermal and External Power Detect: Sense bit when there’s a thermal event or the external power connector is connected or removed from the board.
23:13 mupuf: http://download.nvidia.com/open-gpu-doc/DCB/2/DCB-4.x-Specification.html#_gpio_assignment_table
23:21 imirkin_: is that in the trello item? :)
23:39 mupuf: let me check
23:40 mupuf: now it is
23:40 imirkin_: =]
23:44 mupuf: imirkin_: thx for the gentle kick in the butt
23:49 Booti3869: imirkin: What other task would you would recommend? And how long do you think it would take?
23:50 imirkin_: Booti3869: you should fix your own problems :p
23:50 imirkin_: (or have you already?)
23:50 Booti3869: Meh :c
23:51 Booti3869: Well, I don't know, at least it didn't crash for a long time
23:53 imirkin_: did you send a patch?
23:53 imirkin_: tbh i don't remember the details
23:54 Booti3869: It couldn't really be a proper patch
23:56 Booti3869: Mh, blitting to a y-reversed fb exceeding its dimensions => y is negative => read fault
23:56 imirkin_: right
23:56 Booti3869: Or x is negative
23:56 Booti3869: Ot both
23:56 imirkin_: well, x isn't touched.
23:56 imirkin_: but yeah, it's the y flip
23:56 Booti3869: ...
23:56 Booti3869: Yes.
23:56 Booti3869: y-flipped.
23:56 Booti3869: Silly me.
23:58 Booti3869: And, according to "another dev in #dri-devel", passing a negative y from the common code to the drivers is perfectly legal
23:58 Booti3869: So nouveau should handle the case
23:59 imirkin_: mmm ... negative y is totally legal for glBlitFramebuffer
23:59 imirkin_: iirc
23:59 imirkin_: it indicates a flip
23:59 imirkin_: but that's not how the gallium blit api
23:59 imirkin_: works
23:59 Booti3869: Yes