00:00 Lyude: imirkin_: yes I do!
00:01 imirkin_: ok... so now ...
00:01 imirkin_: if (rt == 0xCF) rt = strtoull(getenv("TESTING"), NULL, 10);
00:02 imirkin_: and run TESTING from 0xc0 .. 0xff and 0x40..0x46
00:02 Lyude: gotcha, one sec
00:03 imirkin_: i.e. for i in `range 192 256`; do echo $i; TESTING=$i ./some-test -auto; done
00:03 Lyude: oh neat, never knew about the range command. I've always just used {192..256}
00:05 Lyude: imirkin_: btw, you need the asm debugging on for that or can I just turn it off
00:09 Lyude: well if I'm wrong I'll just run again, https://paste.fedoraproject.org/paste/9I-ScUj1tWC3IwcmN0Tay15M1UNdIGYhyRLivL9gydE=/raw
00:17 imirkin_: Lyude: no debugging
00:17 imirkin_: in fact, you might want to do >& /dev/null; echo $?
00:17 Lyude: imirkin_: i guessed right then https://paste.fedoraproject.org/paste/9I-ScUj1tWC3IwcmN0Tay15M1UNdIGYhyRLivL9gydE=/raw
00:17 Lyude: it tells you each time it runs with a different value so that should give the same effect
00:17 imirkin_: did you update the library code?
00:18 imirkin_: (and run with LD_LIBRARY_PATH)
00:18 Lyude: yep
00:18 imirkin_: (or whatever)
00:19 imirkin_: so ... fail.
00:19 imirkin_: -ENOTSUPPORTED =/
00:19 imirkin_: send a patch to kill it on fermi
00:20 imirkin_: in nvc0_screen.c::is_image_supported
00:20 imirkin_: er
00:20 imirkin_: is_format_supported
00:20 imirkin_: if class_3d < NVE4_3D_CLASS && binding & IMAGE && format == BGRA8: return false
00:20 Lyude: aw, was about to write that :P
00:21 imirkin_: sorry i sucked all the fun out of it
00:21 Lyude: hehe
00:21 Lyude: that's cool, I am just glad to have this bug fixed so I can work on actually fun stuff
00:39 airlied: skeggsb: btw I've got an nv9x machine I think here (that touchscreen), that after resume gnome-shell never renders again
00:39 airlied: skeggsb: until i restart it, I think it's running 4.10 from f25
00:45 Lyude: imirkin_: patch sent
00:53 Lyude: would be nice if github didn't initiate a merge just for a single patch
00:53 mwk: Lyude: there are options for that
00:54 mwk: but yeah
00:54 mwk: the defaults are not exactly perfect
00:54 Lyude: oh neat, I didn't know you could actually turn that off somewhere
00:54 mwk: IIRC it's right next to that big green merge button
00:55 mwk: not that I'd know, I usually push manually from the command line anyway :p
00:55 Lyude: same
00:57 nyef: ... Once again, I have outsmarted myself: I don't use windows, so I don't need NTFS, except that I *do* use windows occasionally, typically to verify that hardware works and to obtain things like USB traces, so I do need NTFS, but it's not in my kernel because I don't use windows... /-:
00:59 Lyude: btw, I still need one review on that arb_post_depth_coverage patch for nouveau if anyone wants to review/possibly push. pretty sure the only parts left that need a review is the patch that adds stuff to nvc0
01:06 imirkin_: Lyude: hakzsam reviewed that one, no?
01:08 Lyude: ah, wasn't him but someone else did actually review that
01:08 Lyude: guess they just need a push then
01:08 Lyude: unless that's his irc name]
01:08 imirkin_: :)
03:06 imirkin: Lyude: do you need stuff pushed for you?
03:23 imirkin: Lyude: well, either way, all pushed.
03:24 imirkin: Lyude: i made a tiny correction to your FS_POST_DEPTH_COVERAGE doc, hopefully that was ok
09:55 sodapop: how do I extract the blob ctxsw fw? Is it the same method with the one described here: https://nouveau.freedesktop.org/wiki/VideoAcceleration/
10:42 pmoreau: skeggsb: Anything to run to test the new disp code?
10:48 pmoreau: skeggsb: And are you planning to send https://github.com/skeggsb/nouveau/compare/cb25cb52da...8f3fa60de9 to fixes?
11:15 karolherbst: skeggsb: I'll be back tomorrow, would be awesome if you could take a look at my PMU counter branch over the next few days
12:14 mr_sm1th: Does lack of dynamic reclocking mean that the speed is set at boot time, or can I adjust the clocking speed manually for example when I want to do something graphically intensive?
12:16 karolherbst: mr_sm1th: there is a file in debugfs
12:16 karolherbst: usually it is /sys/kernel/debug/dri/0/pstate
12:16 mr_sm1th: Alright, I see, thanks!
13:37 imirkin: jamm: good job :)
13:37 jamm: imirkin: thanks :)
13:37 jamm: wouldn't have been possible without you, hakzsam and pmoreau
13:37 jamm: hakzsam: thanks for your reviews!
13:38 jamm: i still have another shader to work on (xfrm2nv110.vp) will send another patch for that
13:38 imirkin: hakzsam should review it obviously... some of those st's seem questionable, but what do i know
13:40 jamm: imirkin: you mean the st 0x1's ?
13:40 hakzsam: jamm: yeah, nice job. I wil review it for sure
13:41 jamm: hakzsam: thanks!
13:54 imirkin: i mean the st 0x0's :)
13:54 imirkin: but i don't really know all the rules
14:00 jamm: imirkin: you're right, i missed to put st 0xf there instead of st 0x0 (for sched (st 0x0 wr 0x1) (st 0x0 wr 0x0 wt 0x3) (st 0x0 wr 0x1) )
14:10 jamm: imirkin: should i resend the patch as a reply to the same thread?
14:16 jamm: sent it as PATCH v2
14:49 imirkin: jamm: ehh... your call. you can just provide some inline updates
22:57 thican: Hello everyone, could you please give me an answer about HDMI port on Nvidia GPU: can this HDMI port (for example on a GTX 1050Ti) be used as an HDMI _in_ put? I just want to use it for my Raspberry Pi, and I will be happy to have its signal on my current screen since I don't have an HDMI one.
22:58 thican: About the informations I've found on the Web, HDMI content has been designed to be only output on most devices.
22:59 thican: or you have to download an expensive external device.
22:59 thican: s/download/buy/ (silly me, if only I could steal an helmet and steal it again)
23:13 imirkin: thican: afaik the last nvidia-made VIVO-capable boards were in the GeForce 7xxx series, a very long time ago
23:13 imirkin: (pre-hdmi, of course)
23:14 thican: I am sad, thanks for the answer, imirkin. :/
23:16 Mortiarty: imirkin, did you have a chance to look into the h264 artifacts?
23:19 qlutoo: is anyone here deep in the official driver?
23:19 qlutoo: looking for a way to get nvdec/nvjpg outputting something in as few steps as possible
23:22 imirkin: Mortiarty: nope
23:22 imirkin: qlutoo: which gpu?
23:22 imirkin: Mortiarty: i've had ~no time to work on nouveau in the past 6 months or so =/
23:24 qlutoo: imirkin: tegra x1, gm20b
23:24 imirkin: qlutoo: dunno about the X1, but the K1 did not include the video decoding units inside of the GPU, they were part of a different device (although with an extremely similar API)
23:25 imirkin: gnurou would know best, but he left NVIDIA some time back. but he was working on upstreaming X1 stuff.
23:27 qlutoo: okay yeah, there's a device "/dev/nvhost-nvdec" that exposes same interface as gpu essentially
23:27 qlutoo: cdma cmd fifo
23:27 qlutoo: i just have no clue what to feed it
23:29 qlutoo: i'll wait and see if he shows up :)
23:37 imirkin: qlutoo: afaik the interface is pretty similar to the earlier MSDEC stuff
23:37 imirkin: if you can trace what the blob is sending, it should be similar to this:
23:38 imirkin: https://cgit.freedesktop.org/mesa/mesa/tree/src/gallium/drivers/nouveau/nouveau_vp3_video_bsp.c
23:38 imirkin: and other related files
23:38 imirkin: maybe tagr can point at some released stuff?
23:39 imirkin: iirc i did see some attempt at upstreaming NVDEC support for at least one of the tegra boards