MostAwesomeDude: Aw man, git's down again.
MostAwesomeDude: z3ro: Ping?
eboettcher: :/
eboettcher: I find myself installing kde4 stuffs
eboettcher: spstarr: will I like what I see?
spstarr: well, it depends what are you installing
eboettcher: well starting with the base :P
spstarr: pick a nice theme the default is crap
spstarr: if your composite works you will get a very nice panel
spstarr: transparent if you pick the 'glassy' theme
spstarr: but im using KDE trunk
eboettcher: qt 4.4.0
eboettcher: kdelibs from scm
spstarr: so post 4.1beta
spstarr: alpha
spstarr: whatever its called
eboettcher: * kde-base/automoc::kde4-experimental [N scm]
spstarr: oh is this SuSe?
eboettcher: no
spstarr: then svn, ok you'll get a better experience
spstarr: but KDM doesnt work for me right though
spstarr: otherwise, other things do
spstarr: PulseAudio does not
eboettcher: I can't install pulseaudio
eboettcher: it breaks wine
spstarr: hehe
spstarr: then no problem ;0
eboettcher: past deps.. cmake using stuff being built now
eboettcher: so right now I'm just installing the base system.. what packages do you think are worth checking out?
z3ro: MostAwesomeDude: pong
z3ro: MostAwesomeDude: I'll test the new code
MostAwesomeDude: z3ro: Have you tried the stuff from earlier that I pushed?
z3ro: just compiling now. :)
z3ro: MostAwesomeDude: awesome, it looks like it's working correctly. :)
z3ro: I'll check projective lighting too, as that uses a different arbfp/vp.
MostAwesomeDude: z3ro: Yay!
z3ro: ok projective lighting still has some bugs, but I'll look into that further later.
z3ro: great work though. :)
z3ro: "(Now available at your local grocer's. Multiply and add responsibly.)" hehe :P
MostAwesomeDude: z3ro: Cold medicine does weird things to me.
MostAwesomeDude: Working on LIT right now. It's tough.
z3ro: yeah. I've seen the r300 lit code is pretty complex.
z3ro: and probably not very robust, actually.
MostAwesomeDude: z3ro: Nah, it's pretty damn impressive.
z3ro: hmm
z3ro: brb, dinner.
glisse: MostAwesomeDude: my feeling is that doing branching on gpu is a things to avoid, so you should do the pow whatever :)
MrCooper: hmm, Xv overlay seems broken now with totem
MrCooper: all green
bridgman: MrCooper: which GPU ?
Zajec: fglrx driver is able to read EDID data from monitor from VGA-0 but radeon is not
Zajec: can I somehow debug this?
airlied: Zajec: rs4xx card?
Zajec: Mobility X700
Zajec: is this r400?
Zajec: I can see X700 (not mobility) on http://en.wikipedia.org/wiki/Radeon_R420
airlied: hmm we did have some probs on those I think.
airlied: the bios probably lies about which DDC lines are routed where.
Zajec: you may want to check my bug report: http://bugs.freedesktop.org/show_bug.cgi?id=15648
Zajec: can something be done about?
airlied: can you attach an fglrx loog?
StoneNewt: Zajec: it's an R410 fyi
airlied: log even.
Zajec: airlied: do you mean simple Xorg.0.log?
airlied: Zajec: up from fglrx with the monitor
airlied: yup.
Zajec: sure
Zajec: airlied: http://bugs.freedesktop.org/attachment.cgi?id=16736
Zajec: airlied: is this possible that my LCD uses too new version of EDID that is not supported by radeon? I'm just guessing but I can see:
Zajec: EDID Version: 1.3
Zajec: and in case ot LVDS (which is detected fine) it is 1.2
airlied: nope its just some routing gone wrong.. I gotta run, but we should be able to fix it, maybe Alex can spot something :)
Zajec: oh men, would be great :)
Zajec: agd5f: ping ;)
Zajec: agd5f: ping?
MostAwesomeDude: agd5f: Is there a way to only run a vop or sop, and do a nop on the other alu?
MostAwesomeDude: agd5f: Also the r3xx code has some in-place ops; e.g. EX2 temp1.x, temp1.x; Is that guaranteed to work?
glisse: MostAwesomeDude: what you really want to do is to always feed alu and scalar
glisse: nop won't make things go faster while keeping busy both unit by giving them things to do will :)
spstarr_work: is git radeon further than the last Fedora git build of radeon?
spstarr_work: looks at git log
spstarr_work: nice
MostAwesomeDude: glisse: Yeah.
timofonic: Hi
MostAwesomeDude: Hi.
eboettcher: Hello.
MostAwesomeDude: Howdy.
MostAwesomeDude: Stupid LIT. I'm still doin' it wrong.
airlied: spstarr_work: -ati in F9 updates-testing should only be day or two old.
spstarr_work: newer than in rawhide?
MostAwesomeDude: airlied: LIT is annoying me.
MostAwesomeDude: Think I'm just going to copy the stuff in r3xx.
airlied: MostAwesomeDude: hehe :)
spstarr_work: i will grab it but i dont think that fixes the issues i've seen ;/
airlied: spstarr_work: oh rawhide is old.
airlied: spstarr_work: I'm not working on rawhide at the moment.
spstarr_work: airlied: even if it's now f10-devel ok
spstarr_work: i'll grab it from testing
spstarr_work: airlied: in newer KDE its getting worse the composite issue
spstarr_work: airlied: now, if i leave it alone, sometimes it will lock up screen and if i touch the keyboard or the touchpad it will start writing to disk
spstarr_work: so now i have to leave it off completely
Zajec: does agd5f chat sometimes here? :)
MostAwesomeDude: Zajec: Theoretically. :3
Zajec: ;)
airlied: Zajec: we could try valgrind tracing it if you have two machines..
Zajec: i do... but zero knowledge about val.. tracing
airlied: what OS is that btw?
Zajec: openSUSE 11.0 beta 3
Zajec: 2.6.25 ; xorg-x11-server-7.3-101
airlied: it might be worth trying to work with -ati git master for now :)
Zajec: does ati git work with 7.3?
Zajec: Mesa-7.0.3-31
MostAwesomeDude: Zajec: Should.
Zajec: ok... i will try to get git
MostAwesomeDude: airlied: Can't wait for new memory management so Composite+3D stops glitching out.
airlied: MostAwesomeDude: yeah really need to just write it all already :)
MostAwesomeDude: airlied: Well, maybe I'll look into that after I finish the r5xx stuff.
MostAwesomeDude: Need something to occupy me until new AMD docs. :3
Zajec: i thought glisse is working on it...
Zajec: ACTION uses git and ati git first time
Zajec: should I use "autogen.sh" or "make -f Makefile.am" ?
airlied: Zajec: autogen.sh --prefix=/usr
Zajec: oki, compilation in progress
Zajec: that was easy :)
Zajec: uhm, or not?
MostAwesomeDude: Zajec: Doesn't work?
airlied: Zajec: well I didn't think it would work just I now know exactly what codebase you have :)
Zajec: ah, got it :) it was compiled into ./src.libs :)
Zajec: do i rembmer right i need to uninstall fglrx to have "radeon" driver working?
Zajec: i installed 8.5 by generating RPM
airlied: Zajec: nope.. it should work at least for this with fglrx installed.
airlied: Zajec: and we might need fglrx later :)
Zajec: oki, then i'll just try radeon
tormod: hi, I mentioned some ppracer artifacts while testing r500-support. actually I see the same on an rv410 card.
tormod: so it's a regression. but I don't know what it would be called in bugzilla. can I upload a screenshot somewhere?
Zajec: let's try
MostAwesomeDude: tormod: Screenie would be nice.
tormod: is there a pastebin for image files somewhere?
airlied: tormod: imageshack.us
tormod: airlied: thanks. hmm needs registration. what happened to the old cypherpunk/cypherpunk accounts? :)
MostAwesomeDude: tormod: Doesn't require reg. Just browse, hit upload, then copy your direct image link.
Zajec: have ati git driver partialy working... no direct rendering
tormod: MostAwesomeDude: thanks. they almost got me there :) http://img65.imageshack.us/my.php?image=ppracerrv515hg4.png
Zajec: i really think it's about fglrx
Zajec: however xrandr for LVDS works fine
StoneNewt: heh I use mouse + mod
MostAwesomeDude: Hmm. The billboards, right?
airlied: Zajec: fglrx might break direct rendering, but we might need to use it to find a fix for the DDC.
Zajec: ah, ok
airlied: so I'm assuming DDC didn't fix itself :)
Zajec: 1700FPS on glxgears without direct rendering? weird :)
bobbens: it's a horrible benchmark
airlied: Zajec: you have indireect rendering.
bobbens: especially with the new CPUs :)
Zajec: nope :| just checked...
Zajec: (i mean DDC)
Zajec: emm, airlied... i have problem i think
Zajec: i have notebook (with LCD built in), one computer and one LCD... is that enought?
Zajec: or do i need one LCD for notebook and one LCD for second computer (machine)?
airlied: Zajec: it mgiht help to have something on the VGA for the laptop so we can test it:)
airlied: Zajec: but swapping cables might work fine.
Zajec: will have to do so
airlied: Zajec: so can you attach a log from radeon running with the VGA plugged in
airlied: Zajec: using the git codebase
Zajec: airlied: done
Zajec: http://bugs.freedesktop.org/attachment.cgi?id=16747
airlied: Zajec: okay I suppose we should try tracing fglrx, this is a bit complicated :)
airlied: Zajec: first clone git://people.freedesktop.org/~z3ro/valgrind-mmt
Zajec: listen carefully :)
Zajec: got it... autogen.sh ; configre ; make - right?
airlied: Zajec: yeah might --prefix=/opt/valgrind-mmt
airlied: or something to keep it out of the way.
Zajec: oki
Zajec: weird tool... it wants me to use old glibc
airlied: oh damn.. it needs rebasing I think.
airlied: pastebin what it says?
Zajec: checking the libc version... unsupported version
Zajec: configure: error: Valgrind requires glibc version 2.2 - 2.6
airlied: z3ro_: you done a valgrind-mmt rebase recently?
Zajec: i do have:
Zajec: linux:/home/zajec # rpm -qa | grep glibc
Zajec: glibc-locale-2.8-11 ; glibc-2.8-11 ; glibc-devel-2.8-11
Zajec: uhm, and there is not RPM with 2.6 for me ;/
airlied: Zajec: I'll see if I can rebase it to a later valgrind, hopefully agd5f might think of a better way to figure it oput
tormod: MostAwesomeDude: billboards yeah, the trees look like the cut out and fold paper art you find in children's magazines.
tormod: notice also the text/numbers are scrambled.
MostAwesomeDude: Hmm, yeah.
MostAwesomeDude: No idea.
tormod: (the scrambled text is r500 only, works fine on r400
MostAwesomeDude: Hmm.
MostAwesomeDude: The scrambled text could be FP or RS problems.
MostAwesomeDude: As for the billboards, double-check your libpng.
tormod: seems like I have libpng 1.2.15
Zajec: about 150 of my packages depend on glibc 2.8 :/ and i am not sure if this is good idea to have glibc 2.6 and 2.8 installed at once
MostAwesomeDude: Huh, should be fine.
airlied: Zajec: don't bother trying..
MostAwesomeDude: tormod: It's just that I don't think it's the video drivers messing up the billboards, that's all.
Zajec: ok
airlied: Zajec: I'll see if I can fix valgrind.
Zajec: would be nice :)
Zajec: wish to help radeon/ati :)
tormod: MostAwesomeDude: I am pretty sure it was fine before I upgraded mesa
Phantomas: hello
Phantomas: im tired of waiting when ati will bring us some just good drivers
Phantomas: can radeonhd drivers give me the solution to all the problems i have with my 1950 pro
Phantomas: ??
Zajec: Phantomas: hm, i guess ppl here are noteven waiting for drivers from ati
Zajec: this # is about open ati driver
Phantomas: nice
Zajec: probably mostly depends on what are your problems :)
Phantomas: i have some questions for you
Phantomas: will your driver work properly with my 1950 pro
Phantomas: probs:
Phantomas: 1. i cant play ANY 3D game on suse
Phantomas: 2. matroska videos sucks when i open them
Zajec: Radeon 1950 is R520 so some day this will be supported by both: radeon and radeonhd... as far as I know currently you can not expect any nice acceleration from any one of theses projects
Zajec: so for games you will have to use fglrx i'm afraid
Phantomas: its really getting on my nervs
Phantomas: not so much for games
Phantomas: but matroska videos
Phantomas: ?
Zajec: matroska is just a container... it may contain MPEG, DivX, H.264 and others... but i can not really say anything about that
airlied: Phantomas: the -ati driver should have better video playing support
airlied: than radeonhd
Phantomas: no it has not for my card airlied
Phantomas: it doesnt work properly
Zajec: Phantomas: some main question: which distro are yo using and do you know which version of ati driver does it conain?
airlied: the version in SuSE likely doesn't though.
Zajec: ah, susre? i missed that
Zajec: *suse
Phantomas: Zajec yes suse 10.3 and latest version of ati fglrx drivers
airlied: Phantomas: you could try installing the xf86-video-ati driver from git maybe.
Phantomas: airlied is this the opensource drivers?
Zajec: yeah
Phantomas: so will this work for games and video?
airlied: Phantomas: video .. 3D support is udner development at the moment.
Zajec: Phantomas: i do not have openSUSE 10.3... but you can try something... check if there is "radeonrandr12.so" file in /usr/lib/xorg/modules/drivers
Phantomas: /usr/lib/xorg/modules/drivers/radeonrandr12_drv.so
Phantomas: only that
Zajec: ok, it's fine
Phantomas: well..
Phantomas: its not
Phantomas: :p
Zajec: then can try to use the newest stable version of -ati (this open one) driver
Zajec: it's fine about openSUSE 10.3 ;)
Phantomas: yes just kidding
Zajec: sax2 -r -m 0=radeonrandr12
Zajec: will make your system use "ati" driver in 6.8.0 version
Zajec: after that you of course have to reboot machine or simply use "init 3" and "init 5"
Phantomas: SaX2: command not found
Zajec: use this as root and use "sax2" not "SaX2"
Zajec: to switch to fglrx again you can always use
Zajec: sax2 -r -m 0=fglrx
Phantomas: sax2:: command not found
Zajec: ah, and keep in mind that 0 is ZERO (digit), not letter :)
Phantomas: :S
Zajec: yyy
Zajec: confused
Zajec: what does this command shows for you:
Zajec: rpm -qa | grep sax2
Zajec: it's weird as sax2 is by default in openSUSE...
Phantomas: sax2-libsax-8.1-258
Phantomas: sax2-tools-8.1-258
Phantomas: sax2-libsax-perl-8.1-258
Phantomas: sax2-ident-8.1-258
Phantomas: sax2-gui-8.1-258
Phantomas: sax2-8.1-258
Zajec: that's ok
Zajec: are you sure you use this as root?
Zajec: you know... "su" command
Phantomas: of course
Phantomas: yes i know
tormod: Zajec: (reading backlog) your Mob X700 is an rv410, like mine :)
Zajec: Phantomas: what about
Zajec: /usr/sbin/sax2 -r -m 0=radeonrandr12
Zajec: tormod: did you ever try to connect something to VGA-0?
tormod: Zajec, yes but not recently.
Zajec: hm? :)
tormod: I have a VGA output and a DVI, and I think both worked.
Zajec: you seem to be lucky ;/
tormod: I might remember wrong though :)
Zajec: hyh ;p
Phantomas: hey
Phantomas: its fucked up
Phantomas: :S
tormod: MostAwesomeDude: did you sort out the compiz issue where the screen is messed up after rotating cube, but you see it while rotating?
Zajec: Phantomas: ?
Phantomas: when i did what you said it restarted X
MostAwesomeDude: tormod: Ish. We still don't know quite why it happens.
Phantomas: and when i got in
tormod: MostAwesomeDude: what do you suggest for the ppracer issue?
Phantomas: it started doing things when i moved mouse
Phantomas: i pressed ctrl alt backspase
Phantomas: to save it
Phantomas: but now kmenu has gone right
Phantomas: :S
airlied: tormod: if you can find where the r4xx support regressed it would be nice :)
Phantomas: its wierd
airlied: tormod: not sure what you can do without git bitsecting it.
MostAwesomeDude: tormod: To be honest, right now I just don't know.
MostAwesomeDude: I'm buried neck-deep in FP assembly.
tormod: so you don't have this issue in ppracer yourself?
MostAwesomeDude: tormod: Not at the moment, no.
MostAwesomeDude: But a git bisect would be useful.
airlied: I suspect some of the r300 cleanups caused some regression
airlied: I'm seeing the text corrupt in etracer alright
tormod: hmm I don't have the hardware to do mesa bisecting...
MostAwesomeDude: tormod: ? How so?
Phantomas: its shit
airlied: Phantomas: you need to install the development tree really.
Phantomas: tell me how can i restore the system
tormod: I have this slow laptop where I prepare packages, then I build them on Ubuntu PPA.
MostAwesomeDude: Ah.
Phantomas: its messed up
Phantomas: :(
Phantomas: please give me a solution
Phantomas: hey you....
MostAwesomeDude: Phantomas: "sax2 -r -m 0=fglrx" is what was posted earlier.
Zajec: restore? do you mean using fglrx back?
Zajec: just use /usr/sbin/sax2 -r -m 0=fglrx
Phantomas: i dont know i just want to bring the kmenu in the correct position
Phantomas: :S
tormod: well, I can say the regression was before f86baae1 on r500-support.
Phantomas: http://www.imageshack.gr/files/cbxc5jyjwe4itjs4p8ux.jpg
Phantomas: look what i mean
Phantomas: :(
Phantomas: its shit
Zajec: Phantomas: it's just little funny :)
Zajec: did you try to restart KDE?
MostAwesomeDude: Phantomas: I'm not sure what I'm supposed to be seeing here.
Zajec: it doesn't look like driver bug
Zajec: MostAwesomeDude: probably KDE menu should be little more on left :P
Phantomas: Zajec not funny for me :S
MostAwesomeDude: If you're talking about how the K Menu is not on the far left, you could always just drag it back over there.
Phantomas: no i cant it will not let me
MostAwesomeDude: Right-click on kicker, choose "Unlock Panels"?
MostAwesomeDude: Then grab one of the little handles and drag?
Zajec: MostAwesomeDude: that's weird but KDE menu doesn't have handle
Zajec: Phantomas: really, try to restart KDE
tormod: bye
Phantomas: i did it
Zajec: ah
Phantomas: nothing
Phantomas: :(
Phantomas: help meee
MostAwesomeDude: One second.
MostAwesomeDude: -------------------K|%%|~~~~|##|
MostAwesomeDude: Grab one of the little bars to the right of the section
MostAwesomeDude: And drag leftwards.
MostAwesomeDude: --------K|----------%%|~~~~|##|
MostAwesomeDude: It's not a driver bug.
MostAwesomeDude: And I really don't know how to put it any more clearly.
Phantomas: ok thank god i fixed it i pressed left click on kmenu and selected move
Phantomas: its clear enough
Phantomas: thanks
Phantomas: but i fixed it
MostAwesomeDude: Sure...
Phantomas: thank you
Phantomas: i fixed it all
Phantomas: thank you
Phantomas: all
Phantomas: good night
Zajec: airlied: going sleep now, will try to catch you later and ask about valgrind :)
Zajec: good night also :)
MostAwesomeDude: airlied, agd5f, glisse: Pushed. Lots of small fixes; LIT is about half-working, tex swizzling works now, etc.
airlied: MostAwesomeDude: nice.. just trying to find why texenv crashes
MostAwesomeDude: airlied: texenv crashes? Did not know that. One sec.
airlied: I should say crashes my machine.
airlied: I've narrowed it down to the GL_ADD case I think
airlied: make
simpson: airlied: Well, I'd say it's more of a hardlock than a crash.
airlied: 01:27 < airlied> I should say crashes my machine.
airlied: :-)
airlied: oops typing on wrong keyobard earlier
MostAwesomeDude: XFD!
MostAwesomeDude: Okay, so looks like it definitely doesn't use arb_f_p.
MostAwesomeDude: But I can still get it to dump the prog and then exit; one sec.
airlied: MostAwesomeDude: its only the GL_ADD code that tanks.
airlied: if I don't do the GL_ADD bit it renders fine.
MostAwesomeDude: Well, it appears to lock the card; I'm going to just go ahead and kill it right after FP compile.
airlied: hmm found a texture mapping bug..
airlied: the r300 driver remaps texture units dynamically.
MostAwesomeDude: Hmm, could be a problem.
airlied: we don't handle that.. so if you run texrect and turn off texture 0
airlied: you lose all of them.
airlied: I'll see if I can fix that now.
MostAwesomeDude: The FP is clean; compiles fine and looks sane.
MostAwesomeDude: Definitely something further up the line.
spstarr: :)
spstarr: should build git ati driver
spstarr: but thats not my problem ;/
airlied: wierd my quick hack that should fix it doesn't :-(
spstarr: ;/
airlied: okay pushed a sorta working version.
airlied: the first time it breaks.
eboettcher: agd5f: ping
MostAwesomeDude: airlied:
MostAwesomeDude: airlied: You pushed a fix?
airlied: MostAwesomeDude: yup but its not complete.
airlied: MostAwesomeDude: it works with texrect if I disable 0 and then right click again.
airlied: which smells like some state emission.
MostAwesomeDude: Huh.
MostAwesomeDude: Appears to work fine out-of-the-box here, excepting normal AIGLX and Composite weirdness.
MostAwesomeDude: Oh, wait, nevermind. I see what you're saying.
eboettcher: damn... yet again I regret not using screen :P
airlied: doh..
airlied: translating, writing to atom, then updating instruction.
airlied: fail.
airlied: okay proper fix pushed
eboettcher: hm... I'm not too terribly impressed with kde right now
eboettcher: given kwin just totally failed :P
eboettcher: I'm happy with metacity + small panel + my favorite term program
eboettcher: it's hard to tell if the kwin deaths were xorg related
MostAwesomeDude: airlied: Ah, yes, order of operations. This should fix texenv too?
airlied: MostAwesomeDude: not sure :)
eboettcher: compiz crashed, but not every 20s like kwin
airlied: nope texenv still dies :)
MostAwesomeDude: Killing the comp in the process?
airlied: yup :)
airlied: well it kills the GPU
airlied: the machine works but gets really slow :)
MostAwesomeDude: Ah. Well, much as I would love to enjoy that, I have too many open windows right now. :3
airlied: hehe.. I've got 3 machines in front of me to play with :)
MostAwesomeDude: So jealous.
MostAwesomeDude: I need to buy a workstation for r6xx work.
eboettcher: well I put together a machine for my gsoc work
eboettcher: so I have laptop and that
eboettcher: which is nice because I have gdb running right now over here while I cause X to crash ^.^
eboettcher: if I pick up another psu I get get an AGP box up that won't be so stable, but will let me play with AGP r3/4xx cards
airlied: needs to get AGP box.
eboettcher: err...
eboettcher: I need a video smaller than 512x384 to get this to crash
eboettcher: I was trying out fosdem videos.. too big :P
Magnade: eboettcher: tell mplayer to resize it before sending it to vid card?
eboettcher: Magnade: I could... but I want to save my cpu for more kde build stuff going on :P I have some smaller videos.
airlied: lunch
agd5f: eboettcher: pong
agd5f: Zajec: there's some issue with some rv410 laptops and the gpio lines inthe wrong state
eboettcher: agd5f: well after setting my system up again I could never reproduce that xv crash.. may have been an old proto somewhere, or an old libXv more likely
agd5f: running ddcprobe first fixes it
eboettcher: agd5f: right now I'm looking into writing a program to help me keep track of my coverage of the register space
eboettcher: a quick program to show me what I've covered and what I have not -- and to see what's used and what is not
agd5f: eboettcher: ok. you should limit your regsiter space tothe space that should/could be used for a particular op rather than the whole space
eboettcher: agd5f: do you know if we'll be seeing a re-release of that r200 2d doc?
agd5f: eboettcher: at some point. new stuff is the main focus at the moment. once we finally get r6xx out
eboettcher: O.O suddenly everything started segfaulting
eboettcher: even sshd is segfaulting now
eboettcher: hard reboot it is ><
MostAwesomeDude: agd5f: Ping? When do you think the r6xx docs will be out?
agd5f: MostAwesomeDude: soon, but I don't know exact dates yet. final IP review so...
MostAwesomeDude: Ah. Sweet.
MostAwesomeDude: is furiously saving up for a new compy
airlied: hmm if I disable add in texenv we have broken blend.
MostAwesomeDude: Hmm.
airlied: hmm LRP :)
MostAwesomeDude: airlied: LRP isn't working?
MostAwesomeDude: Aw man!
airlied: MostAwesomeDude: not sure.. just the blend test uses LRP
airlied: MostAwesomeDude: so I'm blaming it :)
MostAwesomeDude: Well, one second, lemme find some tests and make sure it works.
simpson: airlied: It's a bit of a moot point, since I still get a hardlock with texenv, no matter what.
simpson: Oh, I see. You disabled the GL_ADD stuff.
airlied: simpson: you can turn off ADD :)
simpson: rolls his eyes.
airlied: just chagne the 5 to a 4 :)
simpson: airlied: Alrighty.
MostAwesomeDude: Whoo.
airlied: MostAwesomeDude: another crash?
MostAwesomeDude: airlied: texenv still hardlocks me, even with GL_ADD removed.
airlied: MostAwesomeDude: awesome..
MostAwesomeDude: ...I pulled out the old P3. It's sitting in the corner, parsing a
MostAwesomeDude: Fedora CD. It'll be a few hours, but it'll be worth it.
MostAwesomeDude: Too bad it doesn't have AGP or PCI-E.
airlied: MostAwesomeDude: PCI only
airlied: ?
MostAwesomeDude: airlied: Bingo.
airlied: wow crappy
MostAwesomeDude: Integrated Intel.
airlied: I have a box like that here though.
MostAwesomeDude: But, it can be a communications box.
MostAwesomeDude: Looks like about $500 USD for a new PCI-E box for an R600 card.
MostAwesomeDude: Man, I wish I were employed.
MostAwesomeDude: And not a student.
MostAwesomeDude: ...As long as I'm wishing, I want a girlfriend, a pantry full of ramen, and a 12-pack of soda.
MostAwesomeDude: :3
airlied: MostAwesomeDude: I think I see a bug in LRP.
airlied: but gimme a sec to think it out :)
MostAwesomeDude: airlied: Go ahead. I thought LRP was decently solid, but...
airlied: MostAwesomeDude: I think the temp reg is messed up
airlied: I think you might overwrite src2 with the first MAD
eboettcher: woot, MPX merged
eboettcher: how long 'till a common wm can use them?
airlied: MostAwesomeDude: http://pastebin.ca/1030744
airlied: the first inst ouutputs to dest 1, but the second reads from dest 1 twice
MostAwesomeDude: airlied: That's definitely a bug farther up the food chain. get_temp should never return slots in-between other slots.
MostAwesomeDude: Wait a sec, something's VERY wrong.
MostAwesomeDude: airlied: This came from texenv?
airlied: MostAwesomeDude: yup.. that just the LRP bit
airlied: its a TEX, LRP and MOV.
MostAwesomeDude: Do you have the Mesa FP source that produced this monstrosity?
airlied: http://pastebin.ca/1030745