Vash63: If I'm on radeon git master and get a 2.6.31 rc, is there anything else I need to get KMS working on an R700?
yangman: Vash63: no KMS for r6xx/r7xx yet
Vash63: Oh, I thought that was just no 3d.
fpoibaf: r300 performance dropped a lot after today update
fpoibaf: openarena went from 43 to 13 fps
fpoibaf: glxgears from 1900 to 900
fpoibaf: is it a know issue?
fpoibaf: no KMS, RV530
hifi: whats "today update"?
hifi: git or distribution?
fpoibaf: mesa master from git
fpoibaf: today vs 24h ago
mikkoc: there was some work on r300 i see
fpoibaf: yes, the shaders_cleanup merge
hifi: bisect it!
fpoibaf: Mmm... the problem went away after a recompile
airlied: wonders when the stencil regression popped up
airlied: not sure if it was always there for DRI2
EruditeHermit: airlied: did you ever manage to recreate the black screen on 3D in DRI1
airlied: I didn't get a chance to play with my r300
fpoibaf: EruditeHermit: I also have that problem when running demos like glxgears: all screen become black until I quit the demo
fpoibaf: games work fine though
EruditeHermit: every 3d app causes that issue for me
EruditeHermit: every being google earth, glxgears
EruditeHermit: that I have tried
EruditeHermit: blah sorry
EruditeHermit: I tried it again and I had to quit
EruditeHermit: fpoibaf: what card do you have?
fpoibaf: GL_RENDERER = Mesa DRI R300 (RV530 71C5) 20090101 x86/MMX/SSE2 TCL
EruditeHermit: airlied: one other question: should I file a bug report about radeon not turning my screen on after resume from suspend in KMS?
MrCooper: airlied: stencil regression?
MrCooper: airlied, agd5f, glisse: btw, any idea how many table entries the PCI GART can cache on the card? I imagine fetching them from system RAM is rather expensive...
EruditeHermit: airlied: btw nice speed boost again today
EruditeHermit: in 3d on kms
airlied: MrCooper: I think 1
airlied: MrCooper: its not so good :)
airlied: MrCooper: the maskedClear test is failing here
airlied: when it specs a stencil write mask
MrCooper: airlied: one friggin' page?!
airlied: MrCooper: afaik but I'm not 100%
airlied: I think it only has one TLB
MrCooper: if so then no wonder it's slow...
airlied: MrCooper: I don't think we can put them in VRAM either
hifi: I heard SPEED BOOST
ferret_: fpoibaf2: Hey, I had a look at the sauerbraten thing, but it turns out that game is way, way past the abilities of my little laptop
fpoibaf2: ferret_: which sauerbratn thing?
ferret_: I thought it was you...
ferret_: bug 22741
ferret_: I put it at the lowest settings of everything, got to the point mentioned, tried to turn everything on, but then it started going at exactly 0 fps :D
fpoibaf2: what card?
ferret_: stupid piece of crap xpress 200m (rv370 iirc but with no tcl)
fpoibaf2: ah, see http://wiki.x.org/wiki/RadeonProgram
ferret_: It's significantly different from what you had but I thought to try it anyway
fpoibaf2: osiris tested non TCL cards and found it very slow when enabling shaders on sauerbraten
ferret_: Yeah, it's about then it stopped rendering frames at all
fpoibaf2: maybe your cards takes a different code path...
ferret_: I think it's just bad ;p
ferret_: With *everything* turned off and 800x600 resolution it was getting 20 fps
ferret_: And it looked like quake in software mode
fpoibaf2: stopping rendereing frames... do you mean it lockups or only got very slowly
ferret_: Probably just really slowly... like one frame every 10 seconds or something. I lost patience with it
fpoibaf2: probably that problem only affects TCL cards or maybe only R5xx cards
fpoibaf2: lunch time now :)
ferret_: Hmm, can I submit data to that chart from my system?
fpoibaf2: sure it's a wiki!
ferret_: I'm not sure it would make sense, because most of the stuff on there I can run but I can't say if it's running at a reasonable speed
airlied: stencil clear failure is to do with user clear code, passing the stencil ref in
airlied: going to be now, but its something silly with the clear value clamping and signed/unsigned int
taiu1: airlied: prey-demo fails with assert for me, I'v put some info at http://pastebin.ca/1494595
taiu1: Im I correct in assuming that it needs smth similar to http://cgit.freedesktop.org/mesa/mesa/commit/?id=b484c71036e0d0b30ac7685ba50a9008d09f5047
taiu1: yeah, comment out .MapTexture in radeonSpanRender* and play :)
lipatden: Hi all
lipatden: Is this the right place for Radeon Xpress 200M cards?
lipatden: great, I can't seem to locate the radeon driver in Fedora 11
ajax: you're joking.
ajax: it's xorg-x11-drv-ati, it's installed by default.
lipatden: ah, I was looking for xorg-x11-drv-radeon
ajax: yeah, we should probably rename it
lipatden: Well not necessarily, since FireGL, Rage etc are all covered too right? so ati is good, just confusing for me :P
ajax: firegl, yes. rage, no.
lipatden: At a long distant point (circa F9) I could load ati drivers into my Linux with 3d accel, what's my best option now (I've NEVER had it working on Fedora 10)?
adamk_: That should also be working by default.
adamk_: Is it not?
ajax: atropine:~% glxinfo | grep renderer
ajax: OpenGL renderer string: Mesa DRI R300 20090101 x86/MMX/SSE2 TCL DRI2
adamk_: So it is working for you.
lipatden: define default? With which driver?
ajax: on an r500, but, also f11. should be working out of the box, assuming there's nothing something special about xpress200 i'm forgetting.
ajax: radeon. the one you're running.
lipatden: well right now, I'm staring at a black background with little white characters - nice to have a 1280x800 text console, sucks that it's my only option (X won't start)
adamk_: Ohhh... Heh... I just realized I read ajax's response as lipatden.
adamk_: Well X not starting is definitely a problem.
lipatden: yeah :S
adamk_: Can you grab the /var/log/Xorg.0.log file from a failed attempt to start X and pastebin it?
mjg59: ajax: xpress200 is rs480
mjg59: ajax: I'm pretty sure there's been some number of people having trouble with it
lipatden: firstly, how do I ensure I'm actually using the ati driver? delete xorg.conf?
ajax: lipatden: that, or just make sure it says Driver "radeon"
adamk_: lipatden: Yes, it should work with xorg.conf, in fact.
lipatden: Well, I deleted my xorg.conf, that's what caused my currently monochrome problem
lipatden: don't have a handy mechanism to transfer the log file, pastebin at a console?
adamk_: Well try to start X without an xorg.conf and pastebin the resulting /var/log/Xorg.0.log file.
lipatden: Will try
lipatden: the other problem is system-config-display faults out
lipatden: why oh why do I bother with these upgrades?
adamk: Because it makes life more interesting.
lipatden: Yikes, X in VESA mode is fugly
ajax: lipatden: also, 'yum install nopaste' will let you pastebin from the command line
lipatden: ah, thanx for the tip
adamk_: Ahhh... I was wondering about that. I searched for pastebinit, which didn't exist, and then tried just pastebin, which gave me my very own pastebin server to run on my machine :-)
lipatden: grrr, ignore ^^
lipatden: adamk_: Now that sounds like fun
lipatden: note: the pastebinned log is WITHOUT xorg.conf
ajax: huh, that's special.
adamk_: Anyone have thoughts as to why it's trying to use the vesa driver?
ajax: (==) Using config file: "/etc/X11/xorg.conf"
ajax: pretty sure that log is _with_ xorg.conf
adamk_: Ohhh... lipatden, didn't you say you deleted the xorg.conf file?
lipatden: cool, I'll try again, maybe I got the one I actually booted into
lipatden: 2 secs
lipatden: dammit, init 3 gives me nothing at all, gotta reboot
ajax: huh, wonder why it doesn't try to give you 10x7.
lipatden: I just want Compiz Fusion :'(
ajax: well, because vbe says the 10x7 modes are disabled. yay vbe.
lipatden: Man, that just grates! running as root, startx works, going to try runlevel 5 and gdm...
lipatden: I KNOW I deleted my xorg.conf and it broke. Now it works. Argh, I love this OS :)
ajax: yay, my existence is validated.
lipatden: I want it to break again so I know I'm not nuts
lipatden: Right, next challenge, 3d accel...
ferret_: lipatden: make sure you use a version of mesa greater than or equal to 7.0.1
adamk_: He's using F11... So it better be newer that that.
ferret_: I've heard bad things about fedora ;p
ajax: trust me, i've heard worse.
lipatden: so you're saying I can get 3d with the radeon driver?
lipatden: hey, look at the effects!!! compiz works!
lipatden: Seriosuly, this hasn't been working for two and a bit years!!
ferret_: I have the same card
lipatden: Woohoo! :D
lipatden: Thanks all
ferret_: (Well, not exactly the same, but it is an xpress 200m)
ajax: lipatden: np
lipatden: Yep, Xpress 200M on a Dell Inspiron 1501, 4GB RAM, dual-core 64-bit Sempron
ferret_: Hmm, I didn't realise they still put that card in newer stuff :)
ferret_: Maybe it's a different chipset underneath
lipatden: I've been running this thing since 2006
lipatden: So it's original
lipatden: 1501 isn't a current model, 1520 is replacement
ferret_: You upgraded the RAM?
ferret_: Good choice :)
lipatden: http://pastebin.org/1518 for full dmi data
lipatden: Honestly, ever since the kernel developers tool out whatever the call is ATI (AMD) claimed they NEEDED fro the proprietary driver, I haven't had joy with this card, now it's sweet as silk
lipatden: wow, typos... the developers took out what ATI claimed they needed for their proprietary driver (around 2.6.27 or .26 I believe)
mjg59: lipatden: Got a reference?
mjg59: The only stuff I've seen was the ACPI headers, which were because they were fucking around with internal ACPI structures rather than use the correct API
ferret_: Yeah, that sounds like what he is referring to
lipatden: mjg59: sorry, stepped away. Reference?
mjg59: lipatden: To stuff being removed from the kernel causing problems for fglrx
lipatden: now is when I prove I am nuts and can't come up with the goods...
lipatden: I'm searching for a reference now, but it came down to this: We, the lords of kernel, decide what function we expose to the outside world. All of the in-tree drivers have moved away from (insert call here), only you people at ATI, who guard your code, use it, so we see no need to keep it in here. Break!
lipatden: I seem to recall it happened around Xorg 7.3
agd5f: glisse: bw patch looks good. we should probably switch that radeon native_mode struct to just use drm_mode struct
glisse: agd5f: yup i got tired working on that patch as i had to restart it after loosing it one time, will try to get back to that, i am on pool allocator
lipatden: ho hum, radeon driver is freezing with compiz
lipatden: oh look, and without
Magnade: i just upgraded yesterday to -git for ubuntu and now i have 1024x768 instead of 1280x960 or whatever this laptop does curious if anyone else has reported issues yet
adamk: I haven't see any reports of that on this channel. Does the old resolution show up in 'xrandr'?
Magnade: i didnt look in xrandr did look at gnomes display thing only showed 1024 and 800
Magnade: Screen 0: minimum 800 x 600, current 1024 x 768, maximum 1024 x 768
Magnade: default connected 1024x768+0+0 0mm x 0mm
Magnade: 1024x768 61.0*
Magnade: 800x600 61.0
Magnade: looking at log now i see im on vesa driver
adamk: That was going to be my next guess.
koolfy: lipatden: what you are talking about the kernel happened at exactly 2.6.29
koolfy: fglrx wasn't fixed and probavly won't
koolfy: they releaded a 9.6 version that doesn,'t support older chipsets
koolfy: only HD afaik
adamk: Everything after 9.3 has only supported HD card. It has nothing to do with kernel changes.
adamk: He seemed to be referring to an earlier situation, but one that no one else seems to know about :-)
koolfy: Never said it had
koolfy: but I don't think they'll bother fix the "old" fglrx kernel issues now...
adamk: And I never said you said it had.
koolfy: ok :)
koolfy: xpress20M is total crap.
koolfy: many problems with Xpress200M never happening with x1650 pro
koolfy: IF only I could tear it off my laptop...
Magnade: adamk: got it working
Magnade: i didnt have radeon named as driver so it for whatever reason now started picking up vesa instead
adamk: Well that's annoying, but easy enough to fix.
MrCooper: agd5f: btw, it looks like we're re-uploading the bicubic BO data for each frame... surely once will do with KMS
agd5f: MrCooper: yeah once would do with non-kms as well probably
agd5f: MrCooper: we can just reserve some space for it in xv init like we do for the r6xx shaders
MostAwesomeDude: r600 shaders live in VRAM?
agd5f: MostAwesomeDude: the ddx ones for exa and xv
MostAwesomeDude: Ooh, nifty.
agd5f: could put them in gart too
agd5f: MostAwesomeDude: consts can live in memory too :)
MostAwesomeDude: agd5f: The possibilities, they are endless. :3
agd5f: MrCooper: done :)
Rabenkla1e: hi, it seems, that i discovered a bug within the free radeon driver regarding font rendering.
Rabenkla1e: here is a screenshot: http://img21.imageshack.us/i/200907141948291024x768s.png/
agd5f: Rabenkla1e: 3d game?
Rabenkla1e: That is a screenshot of the new Monkey Island episode within wine
agd5f: Rabenkla1e: probably a bug in wine or the 3d driver
Rabenkla1e: my installed packages: media-libs/mesa-7.4.4, x11-base/xorg-server-1.6.2-r1, x11-drivers/xf86-video-ati-6.12.2
Rabenkla1e: on gentoo
agd5f: Rabenkla1e: what chip?
Rabenkla1e: Also with x11-drivers/xf86-video-ati-9999 it does not work
Rabenkla1e: ATI Xpress 200M
agd5f: Rabenkla1e: xf86-video-ati doesn't do 3d
agd5f: 3d is handled by mesa 3d driver
Rabenkla1e: on #wine someone told me, that it seems to be a bug of my graphics driver
agd5f: r300 3d driver in your case
MostAwesomeDude: Rabenkla1e: Does it work with the software rasterizer?
Rabenkla1e: MostAwesomeDude: Software rasterizer? You mean the vertex/pixel/shader options in winecfg?
agd5f: Rabenkla1e: LIBGL_ALWAYS_SOFTWARE=1
Rabenkla1e: agd5f: Then I'll have a try using the -9999 version of mesa tonight
MostAwesomeDude: It'll run really, really slow, but basically, if it still renders wrong in software, then it's a Wine bug.
agd5f: Rabenkla1e: if the sw rasterizer doesn't work, then it's a bug in wine
MostAwesomeDude: If the software rasterizer renders it correctly, then it's our bug.
Rabenkla1e: the window stays black and those messages are repeated ~10/sec:
Rabenkla1e: fixme:d3d_shader:shader_glsl_select >>>>>>>>>>>>>>>>> GL_INVALID_OPERATION (0x502) from glUseProgramObjectARB @ glsl_shader.c / 4261
Rabenkla1e: fixme:d3d_shader:shader_glsl_load_constants >>>>>>>>>>>>>>>>> GL_INVALID_OPERATION (0x502) from glUniform4fvARB @ glsl_shader.c / 646
MostAwesomeDude: Rabenkla1e: Hah, those are Wine bugs, probably unrelated.
tball: What is the status of r600 3d acceleration?
tball: I know its not quite ready yet, but is compiz running?
agd5f: tball: in progress. currently stuck on a bug in the buffer aging code.
Rabenkla1e: agd5f: I don't think it's a bug in wine, as at least 4 people managed to play this game with the wine version I'm using
agd5f: Rabenkla1e: did it render ok with swrast?