05:34 mangix: so apparently i'm part of the nouveau project on freedesktop now
05:34 mangix: interesting...
09:54 l4mRh4X0r[m]: Heya, I'm trying to get nouveau to work with my laptop (Lenovo ThinkPad P50, GM107GLM [Quadro M1000M] according to lcpci), but when modprobing nouveau I get the following in my dmesg:
09:55 l4mRh4X0r[m]:sent a long message: l4mRh4X0r[m]_2017-08-01_09:55:14.txt <https://matrix.org/_matrix/media/v1/download/matrix.org/iiRPiJoWaeqDJTTimnpenCaS>
09:56 l4mRh4X0r[m]: Any idea what's wrong?
09:57 l4mRh4X0r[m]: I'm running kernel 4.12.3 by the way
10:31 l4mRh4X0r[m]: Looks like this issue: https://github.com/Gnurou/nouveau/issues/21
13:01 awalls: I have the "Blank monitor, flicker, snow, or other random live image corruption" problem mentioned here: https://nouveau.freedesktop.org/wiki/TroubleShooting/#index8h3
13:01 awalls: with a Lenovo W520 with a Quatro 1000M graphics card
13:02 awalls: How can I adjust the card performance to get rid of the symptom?
13:03 karolherbst: awalls: you can't, because it's a fermi
13:03 awalls: FWIW, these two lines are reported in dmesg:
13:04 awalls: 2.493938] nouveau 0000:01:00.0: NVIDIA GF108 (0c1d00a1)
13:04 awalls: [ 2.506807] nouveau 0000:01:00.0: bios: version
13:04 awalls: Hmmm. Too bad.
13:04 karolherbst: yeah, it's quite sad
13:04 karolherbst: we support reclocking for the prior family and the both after...
13:05 awalls: Right now to have dual monitor, I've set the BIOS for NVidia Optimus mode, I'm using the integrated gfx for the primary gfx.
13:05 karolherbst: yeah
13:06 karolherbst: you should always prefer the intel
13:06 awalls: Probably good enough, I guess.
13:06 karolherbst: because the intel driver is usually more stable than nouveau
13:06 karolherbst: and it'S faster
13:06 karolherbst: without reclocking those nvidia GPUs are quite slow
13:07 awalls: At times I notice some weird diagonal tearing in windows with the intel driver & gfx.
13:07 awalls: But nothing too terrible.
13:09 awalls: I assume the proprietary NVidia driver doesn't support optimus properly under linux (Fedora 26 in my case).
14:29 tobijk: hey mupuf, would it be possible to get me access to the nouveau shader-db instance (karolherbst pointed out that there are some more shaders available in comparison to the fd version)
14:29 karolherbst: "some more"
14:30 mupuf: tobijk: sure, let's PM about that
14:37 karolherbst: I still need to update the readme
14:37 karolherbst: because I want to have timestamped subfolders in the game folders, so it's simplier to do a second run later and remove duplicates
14:50 tobijk: karolherbst: yeah you forgot civ in there as well ;-)
14:52 karolherbst: doubt it
14:53 tobijk: in the readme only
14:54 karolherbst: ohh yeah
14:54 karolherbst: doesn't really matter with the timestamp scheme
15:06 tobijk: mh, why can nvapeek probe my device :/
15:10 karolherbst: because of reasons
15:10 tobijk: yeah because silly reasons
15:10 karolherbst: tobijk: boot with iomem=relaxed
15:11 tobijk: nah, just use root :o
15:11 karolherbst: ohh wait, or this
15:11 karolherbst: but even with root you wouldn't be able if it's set to strict
15:11 tobijk: mh? anyway, got it now
16:10 rhyskidd_: any other relevant parties done RE work on VP6 / Pascal's PVDEC engine?
16:35 karolherbst_: rhyskidd_: nobody afaik
18:02 tobijk: uhm, does hitman cause a infinite shader compile run, or does it run _really_ long until compilation is finished?
18:12 karolherbst: tobijk: really long
18:13 karolherbst: tobijk: around 3-4 minutes
18:13 karolherbst: shader-cache helps a lot
18:18 tobijk: karolherbst: oh yeah, would be good to know what does take that long
18:18 tobijk: another thing on the todo list
18:19 karolherbst: bad compiler perf
18:20 tobijk: i still have an old optimization around, that cuts a bit of compiletime
18:23 tobijk: karolherbst: new approach for the civ6 fix: https://hastebin.com/oruheqetid.hs
18:25 karolherbst: looks better
18:42 tobijk: karolherbst: ah and the difference is actually _only_ the newly compied shader :) yay
18:43 karolherbst: yes