00:09Tom^: imirkin: ok it didnt quite go as high as expected i was running a few other things aswell that were fluctuating it a bit, but it goes up to around 15% atleast :p https://www.youtube.com/watch?v=1LYIuEEgTPQ
00:10Tom^: anyhow need to eat something and get to work.
04:44fireion: Hi! I tried running nouveau on 840M (NV 110)(Optimus laptop with linux kernel 4.6.4). However on bootup noveau showed unsupported device. Am I missing something ?
04:52imirkin: is it a GM108?
04:53imirkin: if so, you'll need kernel 4.7-rcN for that
04:53imirkin: [and note that it still doesn't work for some people]
04:54fireion: Oh. Ok.
04:55imirkin: however if you're looking for higher performance, i suggest you give this branch a shot: https://github.com/karolherbst/nouveau/commits/stable_reclocking_kepler_v5
04:56imirkin: this should allow you to reclock the memory and core speeds. maybe.
04:56fireion: I would like to contribute. I have 2 devices with GM108, one is 930M and other 840M. 840M has nvidia closed source driver. Can someone help in getting started ?
04:57imirkin: grab the branch i pointed to and test it out, see if it lets you reclock
04:57imirkin: if it does, should be able to outperform the onboard intel chip
05:02fireion: imirkin, would it work with 4.6.4, or should I get 4.7-rcN ?
05:03imirkin: 4.7-rc, probably
05:03imirkin: it's an out-of-tree module... the idea is that you have a regular linux tree all built
05:03imirkin: and then you check out that tree, do "cd drm; make" which will build the nouveau module against your linux tree
05:09fireion: imirkin, I'm using archlinux kernel (4.6.4-1-ARCH), it's giving error, n rule to make target
05:10fireion: *no rule to make target 'modules'
05:11imirkin: probably means you don't have a properly configured source tree pointed at by /lib/modules/<version>/build ?
05:11imirkin: look at how the makefile works -- pretty easy to tell it to point to a diff tree
05:12imirkin: also, i very much doubt that tree will build against a 4.6 source
05:12imirkin: you'll need to revert some commits probably
05:12imirkin: anyways, i'm off to sleep
06:21orbea: that stable_reclocking_kepler_v5 branch works on 4.6 unless the recent commits changed taht
06:56unnew: karolherbst_work: so i tried to boot with blacklisted nouveau, run dmesg -w, and modprobe nouveau, here's what happens when i modprobe: https://paste.debian.net/hidden/d81395b5/
06:57unnew: after this, my screen is black (like in standby)
06:57unnew: i'm using a 32bits distro, if that's relevant
07:15karolherbst_work: ohh gp104
07:15karolherbst_work: skeggsb: somebody for you :p
07:15karolherbst_work: unnew: but I guess you should remove your mtrr settings (if you have any)
07:15karolherbst_work: I am getting old :/
07:27unnew: what's mtrr and how should i remove it?
08:14unnew: karolherbst_work: ^
08:21karolherbst_work: unnew: uhh, well if you didn't add it
08:21karolherbst_work: unnew: check your kernel boot parameters
08:34unnew: karolherbst_work: there's root=... and ro
08:35karolherbst_work: no idea then
08:35karolherbst_work: you could create a bug
08:35unnew: where? what's the bug, what should i describe?
08:36unnew: maybe it's my distro which compiled the nouveau module with some option by default but setting a param on commandline would disable this mtrr?
09:47karolherbst_work: unnew: no idea, really.
09:48karolherbst_work: unnew: you should open a bug according to this: https://nouveau.freedesktop.org/wiki/Bugs/
15:22imirkin_: karolherbst_work: i'm probably being picky, but i really really hate the title "bios/volt: Handle voltage table version 0x50 with 0ed header" - can you change that to "zeroed"?
15:24karolherbst_work: I could :p
15:25karolherbst_work: I am sure ben also commented on that :/
15:26karolherbst_work: and I thought I changed it alread
15:39dcomp: should nouveau be uploading a SEQ script similar to the nvidia one?
15:39karolherbst_work: dcomp: yes
15:39imirkin_: yes and no
15:39imirkin_: we don't use the same SEQ language as nvidia does
15:39karolherbst_work: well, some differences aren't that important :p
15:39imirkin_: but the scripts should be functionally very similar
15:41dcomp: I can see an SEQ script in my GM108 with values I can't seem to find in my nouveau trace
15:41imirkin_: are you using karol's branch?
15:42imirkin_: if not, i don't think reclocking was enabled for maxwell in mainline
15:42karolherbst_work: dcomp: well, does it work, or doesn't it work?
15:42dcomp: my card just spams FIFO SCHED ERROR 20
15:43imirkin_: dcomp: iirc your VRAM doesn't come up properly, right?
15:44imirkin_: you could try grabbing karol's branch, and then booting with something like nouveau.config=NvClkMode=7
15:45imirkin_: which should force a reclock right on start
15:45imirkin_: perhaps that'll kick the vram into high gear
15:45imirkin_: or at least ... into gear :)
15:45dcomp: stable_reclocking branch?
15:46karolherbst_work: imirkin: uhh, good idea actually
15:46imirkin_: karolherbst_work: i get them on occasion :p
15:46karolherbst_work: and obviously it will work with that :D
15:47karolherbst_work: somebody else had the same situation once I think
15:47karolherbst_work: not quite sure anymore though
15:48dcomp: is there another branch that works with 4.7
15:48karolherbst_work: dcomp: well, a rebase on master should work, allthough master is quite outdated :/
15:49dcomp: for future reference is master skeggsb/master?
15:49karolherbst_work: I meant my
15:49karolherbst_work: but skeggsb/master would work
15:50imirkin_: skeggsb/master doesn't have maxwell reclocking
15:50karolherbst_work: sometimes wrong commits get rebased too :/
15:50karolherbst_work: imirkin_: for rebase
15:50karolherbst_work: I usually keep my branches quite compatible with skeggsb/master
15:51dcomp: still getting an error with vga_switcheroo
15:52karolherbst_work: dcomp: git rebase -i
15:52karolherbst_work: and give me the list :p
15:52karolherbst_work: ohh wait
15:52karolherbst_work: it rebased, but compile error?
15:52karolherbst_work: ohhhhhhh right
15:52karolherbst_work: skeggsb is on 4.8
15:52imirkin_: rebase on skeggsb/kernel-4.7?
15:53karolherbst_work: that is tricky
15:53karolherbst_work: yeah, 100 commits missing or so
15:53imirkin_: then revert some of the drm-next commits
15:53karolherbst_work: dcomp: git rebase -i eb213eba55c0582c84d28f763fd978e153472e50
15:54karolherbst_work: dcomp: and remove the top drm-next commit
15:54karolherbst_work: and let rebase run through
15:55karolherbst_work: dcomp: but I want to go home now, so you have to wait 1.3h now :p
15:56dcomp: ok ... I've got some more test to run now at least
15:57imirkin_: ouch... 1h20 to get home? that sucks.
16:03dcomp: I couldn't get that rebase to work. But rebase skeggsb/linux-4.7 worked
16:04dcomp: by work I mean compile
16:11dcomp: imirkin_: well NvClkMode 7crashed gnome shell ... which I guess is an improvement
16:11imirkin_: pastebin dmesg
16:16dcomp: imirkin_: currently the dmesg is not a trace
16:18dcomp: imirkin: http://pastebin.com/sbAPy4sc
16:19imirkin_: try adding runpm=0
16:24dcomp: also .
16:24dcomp: [ 488.297405] CPU0: Package temperature above threshold, cpu clock throttled (total events = 1)
16:24dcomp: [ 488.297964] mce: [Hardware Error]: Machine check events logged
16:26imirkin_: dcomp: that appears happy
16:26imirkin_: are there any visible issues?
16:27dcomp: imirkin_: http://pastebin.com/XvUpRxwf
16:27dcomp: I've tried a DRI_PRIME=1 glxinfo
16:27dcomp: and it falls back to the intel
16:27imirkin_: are you running with DRI3?
16:27imirkin_: skeggsb: --^ lockdep appears to complain. not sure if it's something karol did or not.
16:28dcomp: I'm running what ever fedora rawhide is running
16:28imirkin_: skeggsb: struct mutex vs dmabuf lock ab/ba (maybe)
16:38imirkin_: dcomp: pastebin xorg log
16:38imirkin_: my guess is you only have dri2 and loading nouveau after Xorg starts
16:56dcomp: imirkin_: http://pastebin.com/rHMP5AyJ
16:56dcomp: I've had to extract it from journald
16:57imirkin_: i was wrong - you have DRI3 enabled
16:57dcomp: the fact that timestamps match between journald and dmesg is awesome
16:58imirkin_: can you do LIBGL_DEBUG=verbose DRI_PRIME=1 glxinfo
16:58imirkin_: (i'm mostly interested in the lines before the various "real" glxinfo output)
16:59dcomp: I'm currently insmod'ing from within gnome wayland session
16:59imirkin_: that's fine
16:59dcomp: should i do it from a vt?
16:59imirkin_: although you won't be able to unload
16:59imirkin_: the way that you have it set up
16:59imirkin_: coz X latches onto it
16:59imirkin_: you could load X with AutoAddGPU=false
17:00imirkin_: which will prevent it from doing that
17:00imirkin_: and then you'll be able to load/unload at will
17:07dcomp: this time I loaded the driver from a vt before loading gnome
17:08dcomp: and it seems to be using nouveau
17:09imirkin_: you can echo 0f > /sys/kernel/debug/dri/0/pstate in order to get higher perf
17:09dcomp: oh what does NvClkMode 7 do?
17:09imirkin_: note that at present maxwell is missing tessellation and images support (as compared to fermi/kepler)
17:09imirkin_: well, i made the educated guess that 7 was the lowest perf level
17:10imirkin_: if you cat that file, you should see all the available perf levels
17:10imirkin_: from perusing your dmesg, i think you have 7, a, and f
17:10imirkin_: note that NvClkMode takes it in decimal though, so if you want it to boot to 'f', you'll need to say '15' there.
17:11dcomp: good guess
17:11dcomp: Im still confused about why the card doesnt set ram up itself
17:12imirkin_: it normally does.
17:12imirkin_: the vbios script is supposed to handle it.
17:13imirkin_: iirc you had already tried to use NvForcePost=1
17:25imirkin_: dcomp: hm weird. it's as if your vbios doesn't have any init scripts
17:25imirkin_: dcomp: or at least we don't detect any... i see the "running init tables" message
17:26imirkin_: but i don't see anything about init scripts being actually executed
17:26dcomp: Ive extracted my vbios before
17:26imirkin_: i can't look in more detail now
17:27imirkin_: but you can run it through nvbios and see what that says
17:33imirkin_: danvet: you may be interested in the lockdep report in http://pastebin.com/XvUpRxwf - looks like some fun interaction between i915 and nouveau dmabuf...
17:35dcomp: btw that dmesg is18:33 < imirkin_> danvet: you may be interested in the lockdep report in
17:35dcomp: ^^ ignore butter fingers
17:36danvet: imirkin, known since forever
17:37danvet: assuming it's the struct_mutex one
17:37imirkin_: danvet: ah ok
17:37danvet:can't even find it in there
17:37imirkin_: search for "circular" :)
17:38danvet: yeah, struct_mutex considered harmful
17:38danvet: run on 4.7
17:38danvet: that should be fixed there
17:38imirkin_: it's a 4.7-rc7 kernel
17:38imirkin_: although with a somewhat funky nouveau
17:39danvet: hm right, if you try hard enough you can still hit it
17:39imirkin_: dcomp was trying hard. he ran "glxinfo" :)
17:39imirkin_: a pretty intense use-case
17:39danvet: yeah, that's just tempting it
17:41dcomp: btw its 100% reproducable
17:42dcomp: from a total of 3 attempts
17:42dcomp: it'll occur on the first attempt after each boot
17:42imirkin_: after the first print, lockdep turns itself off
17:51karolherbst: imirkin_: yes, it does
17:52karolherbst: have to move, but currently it wasn't possible otherwise
17:52karolherbst: dcomp: so you didn't get reclocking to work
17:53dcomp: its been a resounding success
17:53imirkin_: i think the main issue is that for some reason either your bios doesn't have init tables, or we don't find them
17:54imirkin_: and that in turn leaves a lot of stuff uninitialized
17:54imirkin_: also i had him turn off runpm, coz it seems like we don't reclock memory after it resumes from runpm
17:56karolherbst: mhh, so what is the best thing you actually were able to do?
17:56dcomp: DRI_PRIME=1 glxgears
17:57karolherbst: with stock module?
17:57karolherbst: I think I see the issues
17:57karolherbst: "0x30: 0xf1a => BOOST TABLE"
17:58dcomp: with your branch rebased on linux 4.7 and insmod with config=NvClkMode=7 runpm=0
17:58karolherbst: "0x00: 0xdd1 => PERFORMANCE TABLE"
17:58karolherbst: 32bit table pointer strike again
17:58karolherbst: skeggsb: okay, so maxwell1 too :p
17:58karolherbst: allthough it still looks somehow valid?
17:59karolherbst: let me check something
17:59imirkin_: karolherbst: it's not finding the init scripts...
17:59karolherbst: yeah, maybe due to the wrong table pointers
17:59imirkin_: pastebin nvbios output if you might?
17:59karolherbst: I have a branch with some 32bit nvbios stuff
18:00karolherbst: mhh odd
18:00karolherbst: that vbios feels "wrong", but mhh
18:02dcomp: nvbios: http://pastebin.com/9usc3evR
18:02karolherbst: imirkin_: https://gist.github.com/karolherbst/60d312568a85e4a74dad07b553d51145
18:02karolherbst: ha, I ran it with -v
18:02imirkin_: well *this* thing has no issues finding the init scripts...
18:03imirkin_: perhaps i misdiagnosed?
18:03imirkin_: or the vbios we get is somehow corrupt? dunno.
18:03dcomp: I think that dmesg is with debug=debug not trace
18:03imirkin_: dcomp: oh really? i thought it was trace
18:04karolherbst: dcomp: could you build nouveau from stable_reclocking_kepler_v5_4.7 ?
18:05imirkin_: wow, that's a serious condition table
18:05imirkin_: i think the max conditions i had seen before was like 16
18:05imirkin_: this thing has 59
18:05karolherbst: which table?
18:05imirkin_: Condition table at 0x3ef: 59 conditions:
18:06imirkin_: nothing wrong with it, just ... a lot of conditions :)
18:06karolherbst: well I hope nouveau doesn't have a static array for that
18:07imirkin_: it uses like a grand total of 5 of them
18:07imirkin_: nouveau reads the conditions on-demand, no caching
18:08karolherbst: ahh okay
18:11dcomp: insmod nouveau.ko debug=trace config=NvClkMode=7 runpm=0
18:11dcomp: on stable_reclocking_kepler_v5_4.7
18:12karolherbst: maybe wihtout debug=trace now :D
18:12dcomp: I've taken that one from journalctl not dmesg
18:12dcomp: so at least its complete
18:13karolherbst: sutpid X
18:15karolherbst: looks okay somehow
18:15dcomp: would the fact that 10f808 is in pmu but not init tables have anything to do with it
18:16inglor: v5_4.7 is stable proposed for 4.7? :D
18:16karolherbst: wtf is X doing...
18:19karolherbst: imirkin_: it also looks fine to you right? Except that X messup
18:19inglor: is there a way to reload the nouveau module (after compiling) without restart ?
18:20imirkin_: karolherbst: yeah i think so
18:20karolherbst: inglor: yes
18:20karolherbst: dcomp: please use this X config: https://gist.github.com/karolherbst/1f1bdd1a3822df74097f
18:21karolherbst: ohh wait
18:22karolherbst: dcomp: you have to replace the pci address for the nvidia card
18:22karolherbst: 07:00:0 for you
18:23imirkin_: karolherbst: yeah, seems fine.
18:25karolherbst: then let us remove X from the game...
18:25karolherbst: dcomp: any luck with that config?
18:26dcomp: karolherbst: sorry just having supper
18:30karolherbst: gallium opts for bioshock :O
18:30karolherbst: oh well
18:40dcomp: karolherbst: With that config there doesnt seem to be any differnece
18:41karolherbst: Xorg log then please
18:42dcomp: wait is it meant to allow me to load without crashing?
18:42dcomp: because Ive been loading nouveau in a vt before X then starting gnome shell
18:50karolherbst: dcomp: you can load after you start X though with that config
18:50dcomp: yeah jusy tried it and it works :)
18:51dcomp: Im off for now. If theres any tests you want to run on GM108 let me know
18:52karolherbst: what works
18:52karolherbst: glxinfo printi+ng usefull stuff or something else
18:53dcomp: karolherbst: X doesn't bring gnome-shell down and log me out, glxinfo shows nouveau.
18:55karolherbst: silly X
18:55karolherbst: does glxgear run
19:00dcomp: karolherbst: yes
19:00karolherbst: run it with vblank_mode=0
19:00karolherbst: and reclock to 0f :p
19:10dcomp: so it goes from 79 fps to 155 fps which although awesome
19:10dcomp: the intel card does 348 fps
19:10imirkin_: it's not a fair comparison
19:15karolherbst: dcomp: fullscreen?
19:16karolherbst: dcomp: you should try out real games with that now :p
19:17karolherbst: imirkin_: any idea why X sometimes messes up with DRI2 offloading? or loading the nouveau ddx?
19:17imirkin_: well, nouveau ddx won't load for him
19:17karolherbst: ahh right
19:17karolherbst: modesetting ddx
19:49eringeiogneign: I am using Debian testing right now (Kernel 4.6, MESA 11.2, nouveau 1.0.12) and an NVIDIA GTX 980 Ti. The same behaviour occurs for any other tested GNU/Linux OS as well. The system refuses to boot unless nomodeset is set. Otherwise it gets stuck with the message: switching to nouveaufb from simple
19:50eringeiogneign: I would really like to use basic GPU features via nouveau.
19:50eringeiogneign: If I understand correctly the card should be supported in the latest Nouveau?
19:50eringeiogneign: Would anyone experienced be able to assist in a solution?
19:51imirkin_: it should load with kernel 4.6 if you've installed the proper (non-free) firmware
19:52imirkin_: sounds like modesetting goes totally wrong for you though
19:52imirkin_: you should pastebin your dmesg when the system "refuses to boot" - in actuality it probably boots perfectly fine, you just don't see anything on your screen
19:53imirkin_: also what kind of monitor do you have, and how is it connected?
21:14eringeiogneign: imirkin_, hey, thank you for the reply, sorry for my late reply!
21:14eringeiogneign: I do not have any non-free firmware loaded (is this necessary, will this always be necessary)?
21:15imirkin_: eringeiogneign: it's necessary if you want any acceleration. nvidia GM20x+ chips require a crypto signature to load code onto the chip that is necessary for acceleration to occur.
21:15eringeiogneign: I see all text until that switching to nouveaufb thing comes up. It is a BenQ XL2730Z (144hz @ 2560x1440) connected via DisplayPort.
21:15eringeiogneign: imirkin_, does this firmware have a particular package name (perhaps I can locate it in Debian non-free).
21:16karolherbst: linux-firmware :p
21:16imirkin_: i don't know anything about debian
21:16imirkin_: you should find a debian-related support channel if you have debian-related questions
21:16eringeiogneign: N: Unable to locate package linux-firmware
21:16karolherbst: well, maybe it is named a bit differently
21:16karolherbst: but that's the official name
21:17eringeiogneign: Do all NVIDIA cards require this non-free firmware? Are there high-end modern cards that do not?
21:18imirkin_: all GM20x+ gpu's
21:18eringeiogneign: NV134 does not?
21:18imirkin_: the latest high-end gpu's that don't would be the ones in the kepler series
21:18imirkin_: GTX 780 Ti, TITAN, K40/K80, etc
21:18imirkin_: NV134 == GP104. which is >= GM200 :)
21:19eringeiogneign: Oh I see.
21:19eringeiogneign: Why do they do this? :(
21:19imirkin_: because you bought their gpu anyways
21:19eringeiogneign: What is the most feasable solution to get gaming graphics on free software driver?
21:20eringeiogneign: GTX 7xx and lower?
21:20Tom^: 780ti works quite ok here with karolherbst patches.
21:20eringeiogneign: And I actually replaced my 770 for a 980ti ... > . >
21:21imirkin_: eringeiogneign: AMD will definitely have the best support on linux
21:21imirkin_: they have an actual full-time team developing drivers
21:21imirkin_: and that team has access to documentation and pre-release hw
21:21eringeiogneign: Yeah yesterday I got informed about their new business and I learned what AMDGPU and AMDGPU-PRO was.
21:22eringeiogneign: Maybe I should pick up a AMD 4xx?
21:22imirkin_: i don't think that the RX 480 matches GTX 980 Ti in perf
21:23imirkin_: (but it's also like 1/3rd the price)
21:23imirkin_: i understand that there are AMD GPUs on the way that will be higher end too
21:23Tom^: didnt the 480 have some power issues, where its drawing to much from the PCIE
21:23imirkin_: vega or something
21:23imirkin_: Tom^: i think that was way overblown
21:23eringeiogneign: Why do they have to make this so complicated. ;c
21:23Tom^: imirkin_: *shrug* i just noticed some people complaining. on blogs :p
21:24eringeiogneign: Is there no chance at all NVIDIA will open up this cryptic stuff they enforce?
21:24imirkin_: right. aka "overblown".
21:24imirkin_: eringeiogneign: doubtful.
21:24imirkin_: eringeiogneign: they did end up providing some fraction of the signed firmware blobs that are necessary to operate those GPUs, but not all
21:24eringeiogneign: If I get 144 FPS+ on my Valve games (even on lowest settings, I need competitive performance), I would be happy.
21:25eringeiogneign: 240 Hz monitors are coming out (native) so 240 FPS+ might be a necessity in the future.
21:26imirkin_: i have no idea about any of that
21:26imirkin_: i always have the cheapest slowest GPUs
21:26imirkin_: so i'm happy when i get 5fps :)
21:26eringeiogneign: imirkin_, no. :<
21:26eringeiogneign: de_dust2 me now!
21:26eringeiogneign: Trying to find a nouveau Debian help article to see what they recommend le proprietary firmware.
21:26Tom^: eringeiogneign: http://i.imgur.com/Q738ehH.jpg
21:27Tom^: imirkin_: if i ever decide to buy a 1080ti im so sending you my 780ti
21:27eringeiogneign: Tom^, sexy!
21:27eringeiogneign: Tom^ by the way you can push the HUD all the way to the edges of the screen (in the video settings).
21:28eringeiogneign: I dislike HUDs moving towards the middle.
21:28karolherbst: Tom^: well with shitty graphics everybody goes above 200
21:28Tom^: eringeiogneign: i got to big monitor :p
21:28eringeiogneign: Tom^ what card and driver do you use? And OS?
21:28Tom^: eringeiogneign: 780ti nouveau, linux ofc.
21:29eringeiogneign: Last time I checked Linux was a kernel.
21:29Tom^: yea and that kernel runs on all the distros. :p
21:29Tom^: but fine, archlinux.
21:29eringeiogneign: Not on GNU/kFreeBSD Debian.
21:29karolherbst: Tom^: i seeth thee liketh to just 'round :p
21:29eringeiogneign: Tom^ arch linux scared me at the hard drive partitioning step. ;<
21:30Tom^: karolherbst: =)
21:30karolherbst: eringeiogneign: well, archs installer is a bit crappy
21:31Tom^: eringeiogneign: meh distro doesnt matter much, just get the one thats least in the way of your desires of setting your computer up. its still all just linux.
21:31eringeiogneign: It has an installer?
21:31eringeiogneign: I thought you had to manually set it up from terminal?
21:31karolherbst: but somehow arch user learned to deal with a crappy installer :p
21:31karolherbst: eringeiogneign: well, it is still crappy
21:43eringeiogneign: Tom^, you recon a GTX 780 Ti will work out of the box with recent software (MESA, Nouveau, Kernel 4.6 - 4.7)?
21:44imirkin_: eringeiogneign: no, you need karol's patches. unfortunately upstream has been slow at merging them.
21:44Tom^: work? yes but it wont upclock to 0f pstates without karolherbst patches
21:44eringeiogneign: Any estimate on when they will be merged?
21:44eringeiogneign: Or no ETA?
21:44imirkin_: almost certainly not 4.8
21:45eringeiogneign: My God it looks like AMDGPU is looking closer and closer as the only free option.
21:45eringeiogneign: X . X
21:45imirkin_: didn't i say that earlier?
21:45eringeiogneign: You did darling. <3
21:45eringeiogneign: And I appreciate you.
21:45eringeiogneign: But I do not really understand.
21:45Tom^: amdgpu is no fun tho.
21:46eringeiogneign: Apparently RX 480 is newer but R9 390 / Fury is better performing?
21:46Tom^: having a whole team of devs with specs and all, thats just blatant cheating!
21:46eringeiogneign: omg Tom^ do not scare me like that.
21:46imirkin_: eringeiogneign: and GK208 came out after GK110, but it's still about 1/10th the perf...
21:47imirkin_: RX480 is a "mainstream" gpu, not a high-end one.
21:47eringeiogneign: R9 Fury is the highest end then?
21:47eringeiogneign: Or R9 390?
21:47eringeiogneign: God so complicated.
21:47eringeiogneign: Fucking capitalism.
21:47imirkin_: you'll get the best info about all this from #radeon
21:48eringeiogneign: [22:47] == Cannot send to channel: #radeon
21:48eringeiogneign: They do not want me to talk apparently.
21:48eringeiogneign: Maybe they sense I have a proprietary GPU.
21:48imirkin_: all GPUs are proprietary
21:48eringeiogneign: Ssshhh let me have my mix of beautiful words. ;c
21:48eringeiogneign: All pizza's are proprietary.
21:48eringeiogneign: Until you eat them.
21:48eringeiogneign: Then they are nothing.
21:48Tom^: eringeiogneign: you probably just need to register with nickserv
21:49eringeiogneign: Wow Tom^, rebel.
21:49eringeiogneign: Are you trying to provoke them?
21:49eringeiogneign: With your beautifully free software driver rendered dot.
21:51eringeiogneign: I will throw $10 at anyone kickstarting a free / open software / hardware GPU model on Kickstarter.
21:51eringeiogneign:pokes Tom^ to say something smart and contributing to put me in a positive light
21:51imirkin_: add a few M's to it, and that should be enough
21:51eringeiogneign: You are very good at crushing my overly optimistic dreams. ;<
21:52imirkin_: it's what i do.
21:52eringeiogneign: At least Tom^ cares about my dreams!
21:52imirkin_: he's still young
21:52eringeiogneign: Young beautiful untainted minds.
21:52eringeiogneign: Then you turn 18.
21:53eringeiogneign: And you realise life has come to an end.
21:53eringeiogneign: With beautiful debts. Leeching children. Leeching government. Leeching wife. Leeching leeches.
21:53eringeiogneign: And before you know it you are 65 and do not have enough money for a new back.
21:53eringeiogneign: Life is beautiful.
21:55eringeiogneign: karolherbst why do the Linux people hate your heavenly patches
21:55eringeiogneign: Perhaps we should throw a coup.
21:55eringeiogneign: I heard that is the thing now.
21:55karolherbst: well I always say there are three phases to life: 1. everything is awesome and nothing has to be changed 2. everything is shit and has to be changed 3. it is okay the way it is and you can only change yourself// and depending on your viewpoints your are either in the first, second or third phase :p
21:56karolherbst: eringeiogneign: huh? who hates it
21:56eringeiogneign: I do not know why does it take them years to accept your lovely 0's and 1's? ;~;
21:56eringeiogneign: Is Linux shy?
21:56eringeiogneign: Are you sure you two are both committed to this relationship? Fully?
21:58karolherbst: skeggsb: see what happens if you need too long :p
21:58karolherbst: users are going crazy
21:58karolherbst: and I don't want anything to do with that, so it's your fault now
21:58eringeiogneign: I am not crazy. I am ~ special ~.
21:59eringeiogneign: I am on your side with this one karolherbst. It is all skeggsb's fault. Make a coup happen!!
22:00karolherbst: nah, I am long beyond phase 2 :p
22:00imirkin_: eringeiogneign: i guess you're attempting at being funny? it's not really working out that way...
22:01eringeiogneign: You guys are being way too serious.
22:01eringeiogneign: It makes me sad.
22:01eringeiogneign: I was already sad.
22:02karolherbst: I sam as serious as sam serious is
22:02eringeiogneign: karolherbst why does imirkin_ have to be like this?
22:02karolherbst: he is just being honest
22:03eringeiogneign: Well I rather have it he lie to me. :c
22:03eringeiogneign: And make me feel loved.
22:03eringeiogneign: I am going.
22:03eringeiogneign: I feel unwanted.
22:03eringeiogneign: Good bye my people.
22:45Lekensteyn: wtf was that conversation about... nevermind