airlied: a few ppl pinged me, it mostly scrolled off, back from personal holiday, still on work holidays for a few more days ;-)
glisse: airlied: you should cut the internet ;)
airlied: glisse: well then I'd be bored at home watching childrens tv ;-)
glisse: i still watch old cartoons that i used to watch what i was younger but i admit i am having hard time with new kid stuff
airlied: well I think I know more kids songs now than radeon code
Tommeh: And that is a lot. :)
gormux_: hi all
gormux_: is the control of evergreen cards's fan possible ?
gormux_: cause I think mine is always at full power, making a lot of noise :/
Tommeh: http://www.x.org/wiki/RadeonFeature
Tommeh: There's a lot of power management improvements coming in 2.6.35, supposedly. But I dunno if any will be for Evergreen :(
Tommeh: I doubt it, looking at the Matrix.
gormux_: ah, it is the "power management" thing ?
gormux_: ok :)
Tommeh: Powerplay - near the bottom
gormux_: that's the first thing I'd want :)
Tommeh: I guess it's one of the last thing's AMD release.
Tommeh: I'm not even sure if all the Evergreen documentation is out yet :(
Tommeh: Don't quote me on that though.
gormux_: everything is fine, except that I cannot play World of Goo, but this won't be anytime soon as it seems
Tommeh: It's unfortunate. I've been waiting some time for reliable 3D on R600.
gormux_: is there some usable 3D on R600 ?
gormux_: for the base things though, the drivers are really ok on evergreen
chainsawbike: 3d on my rv770 works fairly well for my needs
gormux_: I'm running openbox + awn + xcompmgr on a 3600*1080 display (dual head) with no problems
Tommeh: Yeah, I run KMS w/compiz and it's nice :)
chainsawbike: damn... ive only got 3360x1050 (dual head , compiz)
Tommeh: Some screen flickering on KMS that I don't get with UMS, but it's only when changing modes/booting/starting wine apps.
Tommeh: And it's limited to the 2nd screen.
gormux_: 22' 1680*1050, 37' 1920*1080 :p
gormux_: I have strange things with some apps
gormux_: for example if I launch dosbox in fullscreen, the second monitor goes off, the display is weird (not well placed in the screen), and when I quit my main monitor becomes the 2nd, and the 2nd stays off :)
gormux_: but it's probably related to xrandr
gormux_: must be dosbox that does not handle the screens correctly and messes everything
Tommeh: gormux: yeah, I think given that my flickering is affecting only the one screen, that it's RandR playing up.
gormux_: my gf likes to play Theme Hospital while watching vids on the second screen... so since I use radeon, I set dosbox in windowed mode
gormux_: before, I used a zaphod setup
Tommeh: Theme Hospital, what a game :)
Tommeh: And the original Theme Park.. :D
gormux_: she only plays theme hospital from this serie :)
Tommeh: Serie?
gormux_: ah, maybe not the right word
gormux_: hum... the Theme *
gormux_: :p
gormux_: the theme something series ?
gormux_: ah, it seems zaphod modes are supported in radeon ?
gormux_: I may try this
Tommeh: Ahh, I see what you mean - she only plays Theme Hospital from the 'Theme' series of games :)
Tommeh: doesn't know what zaphod modes are... Sorry :9
Tommeh: *:(
KotH: modes with two heads and three arms
gormux_: Tommeh: 2 separate X screens
gormux_: permits a lot of things, but I use it basically
gormux_: to have 2 X displays, :0.0 and :0.1
gormux_: so fullscreen apps does not interfere with the other display
Tommeh: Ahhhh! :)
Tommeh: Sounds very useful.
evil_core: but radeon cards even with proper power managmenet accont suck as low power as intel ones?
airlied: very unlikel
evil_core: because intel has better architecture?
evil_core: and r500 is very bad at pm comparing to newer cards?
evil_core: friends shitty sonys with intel works asas twice long as mine T60p with much nmore bigger battery
evil_core: I also dont understand why windows generally works nearly as long as linux one, and profiling with powertop doesnt help too much, and with windows vista/7 they got plenty of animated things on desktop
evil_core: so it must be thousands of wakeups per second
airlied: intel graphics are in the chipset and a lot less powerful
airlied: they also share main memory
airlied: so don't have to power VRAM separate
evil_core: not so less powerfull with free drivers ;)
twnqx: hm
twnqx: which reminds me
twnqx: does gfx card switching work on t500?
evil_core: I dunno if its true, but many users stated they were able to play nfsmw undr wine with X3100
airlied: twnqx: it mostly works, though its not dynamic
airlied: evil_core: no a lot less powerful with any drivers
twnqx: if it's manual it would be ok
airlied: twnqx: I wrote support for it a few months ago
twnqx: i only need the radeon because the crap intel isn't connected to the display port
twnqx: wait, the display port doesn't detect my projector anyway...
evil_core: wtf?
evil_core: its impossible
twnqx: dunno, it detects a tft perfectly alrtight
evil_core: it doesnt make any sense
twnqx: it's a computer
twnqx: they tend to not make sense :]
evil_core: because additional card can be powered off when not needed
Janhouse: Nice, Can't even play 0 A.D. with opensource radeon drivers :(
evil_core: it can be for sure, friend disabled radeon because of battery
Janhouse: http://www.ubuntu-pics.de/bild/65110/0_a_d__036_gLIick.png
evil_core: Janhouse: be happy, because you can play KQlives, xbill, supertux, dosbox, zsnes, icebreaker, Head Over Heels, and many other nice games!
Janhouse: I can't actually... Most of them have terrible FPS
evil_core: frame corruptions will not be fixed for r500?
evil_core: you can, they are 2D games
Janhouse: I am happy because I can watch videos on Ubuntu :D
evil_core: 3D games sucks
Janhouse: but I can't play games...
evil_core: I am happy too, because gallium started workkign
Janhouse: gallium?
evil_core: and I can play 1080p on KMS
evil_core: yeah,q3a works perfectly with that in 1600x1200
Janhouse: I have laptop. Sadly CPU and GPU is connected in some way so I have to set min_cpu_frequency to 1.6 GHZ or even compiz lags
evil_core: but true combat elite not
evil_core: and warsow also
Janhouse: so no games for me
evil_core: I dont use ocmpositing manager
evil_core: it only causes problems, and many VRAM/RAM
evil_core: compositing sucks
Tommeh: But it is nice to use. :)
evil_core: I like ancient window managers
evil_core: and ancent themes
evil_core: but I am a bit strange ;)
evil_core: people in the past usedlinux, because they iked unix ideas, wonderful CLI and windows managers(unix like)
evil_core: now they want free windows copy like modern GNOMe/KDE
Tommeh: Well, there's still choice .. Which is nice :)
evil_core: in the past there werent double click, middle button was for movin objects, and keyboard shortcuts were way different and better than windows one
Tommeh: But I do find it offensive when people seem to think that I just want a 'free Windows' because I like using compiz
Tommeh: I use compiz because it's more enjoyable, not because it's 'like Windows'.
evil_core: buntu and especially compiz get around linux many windows users only because aero wasnt running on their machines
evil_core: and often uses vista themes in gNOME, or even that shitty one from windows XP
Tommeh: Well, try not to tar everyone with the same brush ;)
evil_core: airlied: yare you using pageflip+pm2 and gallium on your T60p?
airlied: evil_core: using d-r-t at the moment I think
airlied: not really doing much more than irc/firefox on it at the moment
twnqx: i wonder how to get rid of the screen flashes when clocks change
Janhouse: wow
Janhouse: http://www.ubuntu-pics.de/bild/65121/selection_037_Z9czKE.png
Janhouse: blur window now works with opensource radeon drivers...
Janhouse: only problem is performance on my laptop :D
Janhouse: it is terrible ^^
evil_core: airlied: btw, do you prefer your newer laptops over FlexView T60p?
Janhouse: what is that?
evil_core: Janhouse: FlexView is legendary screen used in IBM Thinkpads
evil_core: its IPS
adamk_: Last I tried it, blur wouldn't work if you had to screens going.
volter: Hi there!
volter: I'm looking forward to buying a new graphics card.
volter: I'm using two screens with 1920x1080 and OpenGL-Compositing.
volter: Is there any reason to look for a 4670 as compared to a 4350?
adamk_: I use two screens on an 4350, though neither are as large as 1920x1080. It's not a speed demon, but it's certainly usable.
adamk_: I'm not familiar with how the 4670 performs in comparison, though.
volter: adamk_: I currently use a 2600 and it's kind of working as well, but could be better.
Thunderbird: or perhaps a radeon 5450 but I guess this driver doesn't support it that well yet
volter: I also see a black-screen-crash every couple of hours.
volter: Thunderbird: That's what I fear as well!
adamk_: The only issue I've heard with the 5xxx cards concerns the unsupported watermark for mobility GPUs, but I have no personal experience with those either.
volter: And I thought those crashes might be related with the specific graphics processor.
adamk_: volter: I'd be more inclined to think it's a driver bug than something specific to the GPU.
volter: adamk_: I thought there was no hardware acceleration support!
adamk_: Oh, sorry.
adamk_: I got my signals crossed and was thinking #ati for a second.
adamk_: You are correct, there is no acceleration for HD5xxx cards.
volter: adamk_: I thought about a driver's bug concerning the specific GPU.
volter: adamk_: Does that sound reasonable at all?
evil_core: volter: try using gallium only
adamk_: It's possible, but I'd still be surprised if it were specific to the HD2600.
adamk_: Well that's horribly bad advice.
volter: adamk_: Is it?
adamk_: The r500 gallium driver is nowhere near ready.
adamk_: Last I heard, it just managed to do glxgears.
volter: So I'll get a new card anyways.
GNU\colossus: waves
volter: waves back and blushes
volter: The main question is now, whether to look for something "bigger" than the 4350.
Tommeh: adamk_: the r300g driver does R500 right?
Tommeh: If so, Phoronix had it out-performing the classic mesa driver in OpenGL rendering.
adamk_: Tommeh: Yes.
adamk_: And, in fact, I mistyped above.
adamk_: When I said that the r500 gallium driver is nowhere near ready, I mean r600 gallium driver.
Tommeh: Ah. You meant R600?
Tommeh: Heh, makes sense :)
adamk_: I shouldn't let myself near a computer first thing in the morning.
Tommeh: Try being British, I've been up 7 hours already :P
volter: What is the conclusion about the drivers now? :)
Tommeh: Stick to the regular R600 driver, don't go gallium3d (yet)
Tommeh: 3D should work though, if you update everything.
volter: Tommeh: OK!
volter: Argh, there are almost no cards with two DVI-connectors!
volter: Adapter-crap for nothing!
adamk_: I use an HD4850 with two DVI connectors.
GNU\colossus: ding-dong, the DVItch is dead!
volter: GNU\colossus: Tell that to my screens!
GNU\colossus: just buy new ones
volter: GNU\colossus: !!!
GNU\colossus: I'll take those pesky old ones off of you, so you needn't worry about recycling them properly
volter: GNU\colossus: Sounds fair!
gormux_: argh
GNU\colossus: volter: you'll notice that my fees are fairly moderate.
gormux_: Ronnie James Dio is dead :/
gormux_: cries
volter: hands over the two Eizo 2431 S-PVA screens.
GNU\colossus: volter: that'll be €50
GNU\colossus: per screen, of course
volter: feels like doing a bargain.
volter: So, back to topic now!
Tommeh: Heh
volter: OK, I'll just jump into the 4650 and will replace a too short DVI-cable with a DVI-HDMI-cable in the same run.
volter: Thank you for your help!
algol: do scalers work only with LVDS connectors?
algol: it seems that DRM_MODE_SCALE_* constants are used only in radeon_lvds_set_property()
mjt: um. Seeing lots of 'invalid edid' messages when loading 2.6.34 radeon.ko, for a port (dvi) that's disabled.
mjt: on rv690
mjt: err rS690
mjt: ..and an instant system crash when loading radeon.ko without loading fbcon first, apparently (it's not just monitor, system stops responding to pings too)
agd5f: algol: the scalers work with any encoder, but we only use them with LVDS
agd5f: evil_core: pageflipping isn't implemented yet for dri2 on radeon
PeterNL: Hi. Does anybody know about tv-out not working?
PeterNL: Hi. Does anybody know about tv-out not working? I have an Hd2600XT (RV630) and my 1st monitor works just fine with the open drivers.
agd5f: PeterNL: try 2.6.34
agd5f: if you are using kms
PeterNL: Does that matter? What changed?
mjt: lots :)
mjt: (but it depend on what you're comparing it with)
PeterNL: I'm using 2.6.33 now. Really lots?
agd5f: PeterNL: there were several tv-out regression fixes
Tommeh: There are tonnes of fixes in 2.6.34, and a lot of speed improvements.
mjt: "regression" sounds wrong in this context
mjt: when it didn't work before anyway... :)
PeterNL: I'm using Ubuntu 9.04 (jaunty). What should I use? http://kernel.ubuntu.com/~kernel-ppa/mainline/
mjt: 2.6.34 has been released yesterday
mjt: but rc7 is most close
PeterNL: Are there any debs for jaunty? Or only for karmic/lucid?
PeterNL: Or should I just install the rc7-lucid?
mjt: no idea
mjt: either learn how to build kernel or wait till it will be built for you by ubuntu folks ;)
PeterNL: I'll ask in the overcrowded #ubuntu...
shadowmaster: there's no crowd like overcrowd
PeterNL: Sure. When was the rc7 source released?
shadowmaster: 2.6.34 released? :/
shadowmaster: I guess I'll need to check that out. With KMS.
Tommeh: You're still running Jaunty :o
PeterNL: I know...
Tommeh: PeterNL: where do you get versions of the DDX/mesa/X etc.?
PeterNL: xorg-edgers
Tommeh: They still support Jaunty? wow!
Tommeh: Good ol' Tormod
PeterNL: I just rebooted, because FlightGear crashed, and now my monitor shows "Out of Range" and the picture is noisy... What did I break today?
algol: agd5f: is that due to lack of manpower or simply lack of interest?
agd5f: algol: lack of need for the most part. the monitor itself usually has a scaler built in, so there's no need to use the on GPU one
agd5f: lvds panels don't usually have a scaler so they need to use the on-gpu scalers
algol: agd5f: true
agd5f: algol: it's tricker with external monitors as well since the "native" mode is less well defined.
algol: agd5f: doesn't EDID return info on native resolution?
agd5f: algol: it specifies a preferred mode. which is usually the native one on digital monitors like dvi/hdmi, but analog is another story
agd5f: algol: you could probably check a few edid bits to get it fairly reliably, but then they usually have a built in scaler anyway
djkrikke: Hi
adamk: Hello.
djkrikke: well, I was wondering
djkrikke: I have a HD4850 sapphire 512mb version
djkrikke: and neither the proprietary drivers or the opensource ones seemed to do "everything" a few months ago
djkrikke: I mean, either compiz worked, but I couldn't see any movies at the same time, or compiz was very slow etc
djkrikke: were there many modifications already since january/february?
adamk: Yes.
adamk: But even back in January compiz and video playback should have worked quite well with the open source drivers.
djkrikke: I remember that everything seemed to work on fedora core, but on ubuntu I had those weird errors again
adamk: Well the open source driver in Karmic didn't have 3D acceleration for your GPU.
mjt: yeah it definitely worked in january
djkrikke: allright, I'll try a reinstallation in a few days then
djkrikke: but I also remember
djkrikke: I installed a kubuntu version a few weeks ago, and everything seemed to be working smoothly
mjt: ghrrm. I see *tons* of messages like this, about not-connected dvi-d-1 output:
mjt: [ 267.649397] [drm:drm_edid_is_valid] *ERROR* Raw EDID:
mjt: [ 267.649400] <3>00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
mjt: (note it's formatted wrongly too)
djkrikke: but at only up to 1920x1200, whenever I pick 2560x1600, everything lags like hell
adamk: I regularly run with two monitors combined at 2720x1280 without any issues with compiz or video playback.
djkrikke: allright, I'll boot the partition and go for an update
djkrikke: maybe it solves the problems
PeterNL: so I decided to compile the .34 kernel myself, on Jaunty... But where do I download the right sources?
PeterNL: A, B, C or something else? https://help.ubuntu.com/community/Kernel/Compile#Get the kernel source
PeterNL: A, B, C or something else?
PeterNL: Dman, just copy-past the link if it's not completely underlined by your irc client...
PeterNL: Damn,* just copy-paste* the link if it's not completely underlined by your irc client...
djkrikke: adamk, I have the problem that I removed the fglrx* packages, but on reboot, xorg can't find the fglrx module
djkrikke: what's the best I can do, dpkg-reconfigure?
adamk: djkrikke: You should have just disabled the driver in the Driver Manager, and it would have created a new xorg.conf file you.
adamk: I'm guessing that you didn't change your xorg.conf to use the 'radeon' driver instead of 'fglrx'?
djkrikke: I installed it using the ubuntu howto, that means with .deb files
CartoonCat: hellos
adamk: djkrikke: Hmmmm.. Very Ubuntu wiki I've seen on the matter says to use the driver manager :-)
adamk: djkrikke: In any case, you can probably just get away with removing /etc/X11/xorg.conf and letting Xorg autodetect what driver to use.
djkrikke: right
CartoonCat: Im having issue with google Earth. The wireframe is leakng at the edges of the textures. GE says its using "DRI R300" but I thought that was merged into mesa, is Lucid using the old drivers? (9.10 worked fine with no issues)
djkrikke: or maybe reinstall quickly and use the built-in uninstall functions
adamk: djkrikke: As long as fglrx is completely uninstalled, removing /etc/X11/xorg.conf should be fine. If you want an xorg.conf file you can always create one with 'Xorg -configure'
adamk: CartoonCat: DRI R300 is the correct renderer string for r300 through r500 GPUs.
adamk: CartoonCat: That has not changed in recent versions of Mesa. At least not that I've noticed.
CartoonCat: ok,wanted to make sure
CartoonCat: Now how do i figure out what is causing this leak or shear, or what ever it is called ?
djkrikke: adamk, how can I test if my drivers are correctly installed?
adamk: djkrikke: Check the output of 'glxinfo | grep -i render'
djkrikke: ok
djkrikke: Software rasterizer, this is not good :)
djkrikke: I removed my xorg.conf before I rebooted
adamk: Pastebin your /var/log/Xorg.0.log file.
chithead: and dmesg too
djkrikke: ok
djkrikke: my mouse is gone
djkrikke: great
djkrikke: sudo dpkg-reconfigure xserver-xorg just doesn't do anything it seems
djkrikke: is this normal?
jcristau: yes
djkrikke: (it doesnt even make a xorg.conf file)
glisse: mjg59: if we get lid event (close->open, open->close) we can assume that the lid status acpi report is not broken ?
mjg59: glisse: Yeah, I think if state changes then you're probably good
mjg59: I haven't seen any that have it reversed
mjg59: But you can't guarantee that the initial state is correct
glisse: ok, thx for all this insight in the painfull world of acpi
glisse: if there is any acpi hw engi i can stab in the back let me know
djkrikke: libgl1-mesa-glx can't be found in the apt-repositories, is this normal?
adamk: No, that's definitely not right.
djkrikke: ok, will post the xorg.log
djkrikke: http://pastebin.com/zsFbDYsw
djkrikke: AIGLX error: dlopen of /usr/lib/dri/r600_dri.so failed (/usr/lib/dri/r600_dri.so: cannot open shared object file: No such file or directory)
glisse: error is selfexplicit
djkrikke: but an hd4850 is rv770 chip
djkrikke: or doesn't that have anything to do with it?
adamk: djkrikke: You're using Karmic...
djkrikke: yes
adamk: djkrikke: Didn't I tell you that there was no 3D acceleration in the open source drivers in Karmic?
agd5f: djkrikke: r600 3D driver supports both r6xx and r7xx chips
djkrikke: hmm, can't remember it
djkrikke: is it supported in lucid?
adamk: Yes.
adamk: Yes, in fact, I said "Well the open source driver in Karmic didn't have 3D acceleration for your GPU" :-)
djkrikke: oh right
djkrikke: sorry, didn't see that sentence
djkrikke: let's see how good ubuntu is at self-upgrading :)
cafaro: Getting a black screen with radeonhd driver.. (using a Mobility HD3650), here are the errors I receive: http://pastebin.com/dp7XbJGT
adamk: You should be asking about radeonhd in #radeonhd :-)
cafaro: ok
adamk: But, most likely, you just want to use the radeon driver.
cafaro: someone told me here I needed a newer kernel for that (.33 i believe), so I figured I would try radeonhd before getting a newer kernel
Daekdroom: Newer kernel for what?
cafaro: the radeon driver
Daekdroom: The driver itself doesn't require the .33 kernel
Daekdroom: Whoever told you so was probably refering to a specific feature
cafaro: ok
adamk: 3D acceleration will work with both radeon and radeonhd with 2.6.32.
adamk: Or, at least, it should.
cafaro: here's my log of radeon then: http://pastebin.com/ytiAqBiP
adamk: And what's the problem when using radeon?
cafaro: adamk: black screen
cafaro: I get a flicker for 1second, and then it goes black
NTU: 2.6.32 had so many LVDS quirks in it.... i couldn't even use it
adamk: Why would the kernel matter for LVDS quirks if KMS isn't being used?
NTU: i was using KMS ;-)
adamk: cafaro: If I had to guess, I'd say that this is the issue:
adamk: (II) RADEON(0): Query for AtomBIOS Get Panel EDID: failed
adamk: I guess all I could really suggest is to try a newer version of the radeon driver.
cafaro: adamk: I followed these steps: http://www.x.org/wiki/radeonBuildHowTo#xf86-video-ati.28ddx.29
cafaro: and I get no screens found ( http://pastebin.com/KxXNSTxA )
cafaro: here's xorg.conf: http://pastebin.com/s3nZDidy
chithead: no modules found, your xorg install seems broken
chithead: /usr/lib/xorg/modules contains what?
cafaro: chithead: http://pastebin.com/VAQ7e1mJ
chithead: I guess your modulepath in xorg.conf causes this
cafaro: ModulePath "/opt/xorg/lib/xorg/modules,/usr/lib/xorg/modules"
cafaro: it's a comma, couldn't see that
cafaro: retries...
cafaro: "no screens found" again
adamk: cafaro: Why not just use the default modulepath and copy the radeon_drv.so file over since you're having this problem?
cafaro: what's the location of the default modulepath, if I may ask..?
adamk: In this case /usr/lib/xorg/modules, probably. You'd actually put it in /usr/lib/xorg/modules/drivers/
cafaro: I was actually trying the lastest stable version, which is the same as the driver i was using before
adamk: 6.13.0 is considered stable, as far as I know.
adamk: And it's definitely newer than what you were using.
cafaro: compiled for 1.7.5, module version = 6.12.6
adamk: There was even a 6.12.7 release, as I recall.
adamk: So, no, you were not using the latest stable version before.
cafaro: I think the documentation is a little outdated
cafaro: I was using the radeon build guide from x.org
cafaro: adamk: radeon 6.13.0 log: http://pastebin.com/xMqHykWq
cafaro: adamk: still getting a black screen, radeon error seem to be gone though
cafaro: adamk: weird thing is, that X isn't being started automatically, and "/etc/init.d/gdm start" isn't doing anything either...
soreau: cafaro: Have you tried enabling kms at all?
soreau: The log looks fine, other than that afaict
cafaro: soreau: no I haven't
soreau: I would see if kms helps the situation
cafaro: soreau: it seems it's enabled by default? http://wiki.debian.org/KernelModesetting#ATI
cafaro: soreau: how can I check if it's enabled?
soreau: dmesg|grep modeset
soreau: Also, your X log tells you
arrenlex: When I autoscroll with firefox (middle-click a page and move the mouse below the icon to have it scroll automatically), I can't cancel it. It keeps scrolling for 5 or so seconds after I click to cancel. Previously I was on nouveau and this problem did not exist.
arrenlex: Is this an issue which the video driver might possibly cause? Or is it just not possible that radeon could be at fault for this behaviour?
cafaro: soreau: I think it's enabled
soreau: cafaro: What does dmesg say?
cafaro: soreau: [drm] radeon defaulting to userspace modesetting.
soreau: That is UMS, not KMS
soreau: k is for kernel and u is for userspace
cafaro: alright
cafaro: soreau: btw, would I need 2.6.33 kernel, or is 32 fine?
soreau: I would recommend going with a later kernel if it is feasible
soreau: IIRC .32 was the first kernel to have 3D support for HDxxxx series cards
cafaro: ok
adamk: For KMS, 2.6.33 is needed, isn't it?
soreau: hm.
adamk: I could be wrong.
soreau: That is a good question
soreau: only one way to find out ;)
soreau: or, just upgrade the kernel and don't bother with knowing :P
evil_core: cafaro: use multiple lines
evil_core: I am using it like that
evil_core: and commenting when it causes problems
cafaro: evil_core: are you talking about ModulePath?
evil_core: right
cafaro: ok
evil_core: cafaro: http://carme.pld-linux.org/~evil/radeon/dirhack/
evil_core: cafaro: there are building-scripts, config files, and other useful scripts that generally works for me
cafaro: okay, thanks!
evil_core: but remember to tune them for your chip, they are r300 specific
evil_core: r300 means r400-r500
NTU: r300 means r300 too :)
djkrikke: http://pastebin.com/Wes2xUmC
djkrikke: What can be wrong this time?
djkrikke: I upgraded to lucid, but still software rasterizer
NTU: DDX might not be installed
NTU: (EE) Failed to load module "ati" (module does not exist, 0)
djkrikke: yes indeed
djkrikke: what is DDX?
soreau: xf86-video-ati
soreau: You need to install the xorg-server-driver-radeon package or whatever deb calls it
djkrikke: hmm
djkrikke: Ive got that package
soreau: You have the -ati package which is the wrapper
soreau: and if you have -radeon, it's not installed correctly
djkrikke: oh, so which one i need?
djkrikke: the -ati
djkrikke: right
Studio271: Should I hold a question about texture compression capabilities of the Radeon driver until someone is willing, or sohuld I just blurt it out?
djkrikke: soreau: so I have to remove the -radeon package and install the -ati package?
soreau: no. you want both
NTU: this is exactly why I don't use distros with stupid package names like fedora, opensuse, debian, ubuntu etc... they change the name on you to not only confuse you, as a person, but ./configure scripts sometimes as well since they move the libraries around. I'm not saying you should change your distro, but to those who say I'm a "masochist" ..this is _why_
djkrikke: hehe, no problem
djkrikke: what distro do you recommend?
soreau: Studio271: don't ask to ask a question. just ask it with as much reasonable detail as possible and if someone knows the answer, they will likely reply
djkrikke: right, I needed them both indeed
soreau: I recommend the distro that best fits the users experience level and purposes
djkrikke: it works great!
djkrikke: soreau: what distro are you using then?
soreau: djkrikke: I use arch and ubuntu. I also have used gentoo
NTU: uses his own (I'm on windows 7 x64 so no /versions please)
Studio271: Well, just for possible moral boosting, though; Thank you all for your excellent work on the driver(s)! I have Penumbra:gRequiem working nearly perfectly (and at playable framerates!) on my system: Archlinux x86_64, HD4770, following the instructions here: http://bbs.archlinux.org/viewtopic.php?id=79509&p=1
Obscene_CNN: is so confused by how to write r600 assembly code
soreau: slaps the 7 out of NTU
Studio271: That aside, I've been trying to get Left4Dead going with Wine; as you may know, arch64 uses lib32-* packages for all of that handling; I'm fairly certain that Penumbra even uses the 32-bit libs for that, as well, which is why my problem is confusing to myself.
Studio271: I can use driconf to "enable s3tc texture compression even if software support is unavailable" and get left4dead working, but with no textures at all (just shaded poygons and weirds trails form models that I assume are shadow-related).
adamk: s3tc is not supported, at all, on r600l
adamk: So you're aadvertising the extension, but the driver can't handle it.
Studio271: but it gives and error and refuses to load if I don't do the driconf thing; they look like this: http://i42.tinypic.com/14y7h9j.png
soreau: adamk: r600l new driver!?
soreau: adamk: phoronix time! xD
adamk: soreau: Yeah, something like that.
adamk: soreau: The new laggium interface.
Studio271: ok, adamk: you have concisely answered my quesiton. :D
soreau: lulz
Studio271: I'll save you the headache of asking the obvious question no developer wants to hear after saying a feature is not available. :P
adamk: When it's done.
adamk: And I'm not a developer, but thanks :-)
Studio271: heh, well, thanks anyway
NTU: adamk: no?
Daekdroom: A new mesa interface? Already?
Studio271: so, despite my, say, compiling and placing libxtc_dxtn.so into /opt/lib32/usr/lib/, the radeon driver itself doesn't know what to do with it yet?
adamk: Correct.
Studio271: gotcha
adamk: I'm not quite sure what happens to the textures, but I imagine they go through some sort of super duper phaser deconstruction before being reassembled on the other side of the galaxy.
Studio271: haha
Studio271: wonders if he can tell l4d not to use TC...
Studio271: anywho, I'll just have to wait it out; hopefully development of TC beats the porting of the source engine to linux. ;P
adamk_: Hey ops, can we at least put in the topic that you have to use a registered nick to speak in here?
Studio271: yeah, "/msg nickserv help" explains it concisely
grey-water: hello
grey-water: hi. i'd like to override the kms modedetection for my crt monitor. is that possible?
grey-water: apperantly kms is the only way to get R600 openGL working properly
Obscene_CNN: has ums working properly on r600 as far as he can tell
grey-water: do you have any problems with OGRE?
grey-water: liek crashing^^
Obscene_CNN: whats OGRE?
dileX: I know
Studio271: OpenGL RE
CtrlAltCa: hi; i'm getting some strange video memory corruption on some operations (mostly window focusing and scrolling); here's an example: http://i41.tinypic.com/70aryf.jpg ; xorg-server-1.7.7, xf86-video-ati-6.13.0, mesa-7.8.1, libdrm-2.4.20 (slackware64-current); video card: ATI Technologies Inc M76 [Radeon Mobility HD 2600 Series]. Any hint on where to start searching for a solution?
grey-water: object-oriented graphics rendering engine i think
grey-water: its that goog old "IRQ's not enabled, falling back to busy waits"
grey-water: (i checked for firmware as stated above)
Obscene_CNN: I don't think I have ever tried it. I get warnings in piglet due to busy waits though
grey-water: a #radeon-log from february suggested that one has to use kms these days
grey-water: but since 60hz is deadly on a crt that's not really an option for me right now
grey-water: so either there is some UMS trick i don't know about or I need to convince KMS that my monitor is a bit more powerful than 800x600@60
grey-water: I understand busy waits are somekind of cheap replacement when X.org can't get to the IRQs?
Obscene_CNN: busy waits just poll the device until its done or it meets some condition.
grey-water: so that shouldnt prevent opengl from working
Obscene_CNN: correct
grey-water: thank you.
person132: I'm trying to get DRI working with radeon on a Mobility Radeon 7000 under Slackware 13.0.
person132: The Xorg.log shows everything being enabled, but glxinfo doesn't tell me I have direct rendering.
person132: The Xorg.0.log is here: http://pastebin.com/yFBFzSmS and my xorg.conf is here: http://pastebin.com/1qUCrjXQ
soreau: person132: Not sure why you're using xaa rendering method (or an xorg.conf for that matter) but nothing in your X log indicates that DRI isn't working
soreau: Can you pastebin the complete output of LIBGL_DEBUG=verbose glxinfo?
person132: http://pastebin.com/QnafXbgn
soreau: That looks more like the output of 'glinfo' to me
soreau: Does ls -l `which glxinfo` show anything of interest?
soreau: but judging by the output, it looks like DRI is working
soreau: person132: What makes you think DRI is not working?
person132: I guess I made a false assumption based on the fact that glxinfo didn't indicate direct rendering
person132: And that glx_texture_from_pixmap seems to be missing, even though AIGLX seems to enable.
adamk: person132, soreau: FYI, glxinfo on Slackware 13.0 is not glxinfo.
adamk: It's actually a symlink to glinfo.
adamk: Due to a packaging error with Mesa.
adamk: It was fixed in an update.
soreau: Ah, I was right ;)
soreau: person132: I assume you're getting this message when trying to start compiz?
person132: Yes.
soreau: If so, try LIBGL_ALWAYS_INDIRECT=1 compiz --replace ccp
soreau: or compiz --replace --indirect-rendering ccp
soreau: or both together ;)
person132: I used that command line; it didn't help.
soreau: ok, pastebin the complete output
person132: Compiz still reports no manageable displays
person132: http://pastebin.com/Tg0fwwnq
adamk: Whcih desktop environment are you using?
person132: Blackbox.
adamk_: yeah, that's not gonna work.
soreau: heh
adamk_: Blackbox is a window manager, not a desktop environment.
adamk_: And you can't replace a running blackbox session with compiz.
person132: That would explain it.
adamk_: Either create a new session to start compiz stand-alone, or launch it from an actual desktop environment that let's you replace the window manager.
person132: What would I need in a new session?
adamk_: Whatever you want to run with compiz.
adamk_: Maybe a panel or dock.
adamk_: (For example).
person132: I should have run a console IRC client - I'm going to test it out.
person132: My god... It's beautiful
Daekdroom: I don't like console.
Daekdroom: I'm horrible at memorizing keyboard shortcuts D:
person132: Thank you very much for helping me with my rather stupid problem. My desktop is now responsive and beautiful.
soreau: No problem ;)
evil_core: soreau: why indirect-rendering, wouldnt it slowdwon compiz?
soreau: evil_core: He's using AIGLX/DRI1
soreau: With that setup, you must use indirect rendering
Daekdroom: See, I always thought DRI standed for Direct Rendering Infrastructure
person132: Will EXA give me better performance than XAA?
soreau: Probably, yes.
soreau: Daekdroom: yes but AIGLX stands for Accelerated Indirect GLX
MostAwesomeDude: EXA will perform far better than XAA at this point.
Daekdroom: soreau, which is sorta of an irony, but heh.
soreau: person132: With X versions > 1.5 you should use exa
soreau: MostAwesomeDude: Even on a 7000?
evil_core: soreau: for DRI2 too?
soreau: evil_core: With DRI2, you do not need to use AIGLX
soreau: but you can
evil_core: should I disable AIGLX in xorg.conf?
soreau: (if you r machine doesn't lock up like it does here with --indirect-rendering ;)
soreau: evil_core: If you use compiz with DRI1, no
MostAwesomeDude: soreau: EXA has Render accel, which is pretty important.
soreau: but it doesn't hurt to have it enabled with DRI2
soreau: MostAwesomeDude: Right.
Daekdroom: Now, I'm using xorg-edgers radeon driver and when I have compiz activated and try to play fullscreen videos through VLC or gstreamer, the videos are played with a low frame-rate. Any ideas on how to fix it? It does the same thing on both gallium or classic mesa.
soreau: MostAwesomeDude: I had a question mark there though ;)
evil_core: I am not using compoisiting currentlym only gallium with pageflip(its broken for classic) and dynpm
Daekdroom: evil_core, did you compile the new DRM or did you grab a kernel somewhere?
evil_core: and it sometimes lockups after few hours(usuallu after 1-3 days), but its probably due to dynpm some bug
evil_core: compile, with Obscene patches
soreau: Daekdroom: Does it make a difference if you try Unredirect Fullscreen Windows in ccsm>general options?
evil_core: he made many of them especially for me :D
Daekdroom: Will check now
MostAwesomeDude: soreau: I don't have a 7000 sitting around, but EXA should be faster than XAA on there.
soreau: MostAwesomeDude: Ok, thanks.
Daekdroom: Oh right. Obscene is the guy that keeps posting performance patches on phoronix, isn't he?
evil_core: he is
evil_core: but he uses r600 or r700
evil_core: but he was so kind to make also for r300(up to r500) too :)
evil_core: and gallium with pageflip rulez
Daekdroom: soreau, flawless and bright. Thank you :D
soreau: :-)
zhasha: MostAwesomeDude: do you want mine? I'll give it to you for free
evil_core: not as quick as UMS and many games dont run, but q3a is good enough,m no tearing and aggresive power management
MostAwesomeDude: zhasha: PCI?
zhasha: AGP x4
MostAwesomeDude: :C
MostAwesomeDude: No thanks.
evil_core: lol
zhasha: it's an All-in-Wonder card :P
evil_core: he should give M/B and CPU for it also for free ;)
evil_core: hmmm.7000?
evil_core: I was trhinking that bridged r700
evil_core: 7000 must suck at performance
zhasha: I think it's a 7200 though, I'm not sure
Daekdroom: lspci
MostAwesomeDude: No, I have an AGP board, but digging it out always makes me sad.
Daekdroom: Ah.
zhasha: Daekdroom: the computer it came from is somewhat disassembled at the moment
MostAwesomeDude: I need to, though. I need to review the XGI patches on the bugzilla and ack them. :T
jcristau: didn't the xgi guy end up getting git access and pushing his code drop directly?
evil_core: Daekdroom: if you are interested at pm, I would recommend http://carme.pld-linux.org/~evil/radeon/dirhack/useful_scripts/radpm - simple frontend ;)
Droste: MostAwesomeDude: what did AGP do to you that you hate it so much? :-)
MostAwesomeDude: jcristau: I have zero idea. I don't think anybody actually ever tested his code either.
MostAwesomeDude: Droste: Same thing it did to everybody. Gave us hardware with horrible incompatibility bugs.
Daekdroom: evil_core, when it comes to having a IGP, PM isn't really interesting.
evil_core: what his code does?
evil_core: Daekdroom: IGP can be also in laptop, right?
Daekdroom: evil_core, which is not my case, so I'm like.. whatever.
evil_core: ok, no problem
evil_core: I lik dynpm, I was waiting 2 months for working solution after initial Zajec patches
Daekdroom: What is your videoboard model?
evil_core: r500, FireGL V5200
evil_core: its UXGA T60p
soreau: What are the XGI patches supposed to do?
soreau: MostAwesomeDude: Why does getting your agp card out make you depressed?
MostAwesomeDude: soreau: Old board, AGP cards are buggy, etc.
evil_core: http://www.mail-archive.com/mm-commits@vger.kernel.org/msg19321.html
soreau: MostAwesomeDude: They are?
evil_core: XGI are SiS chips?
soreau: MostAwesomeDude: My agp card has been working fine for years now
ajax: some of them are. some of them are trident chips.
evil_core: SiS chips are deepshit, I saw a few and they sucked even under windows
MostAwesomeDude: I've got the XGI Volari. Yay?
ajax: MostAwesomeDude: which one?
evil_core: ouch, I missed XGI with SGI ;)
MostAwesomeDude: ajax: Dunno. AGP 4x, has a stamp for Dx 9.0. Never actually plugged it in, but Donnie says it works.
evil_core: and was wondering how SGI can use that big shit for their workstations ;)
MostAwesomeDude: "Works" could just be VESA, of course. :3
ajax: MostAwesomeDude: probably V5 or V8 then.
ajax: V3 and i think 8300 are trident cores, the rest are all sis cores.
ajax: two great tastes etc.
MostAwesomeDude: Ugh, disgusting.
evil_core: I like Trident name
MostAwesomeDude: just realized that he'll have to undelete the Mesa driver if he can get it to run :C
ajax: the Z7/Z9/Z11 are sis-based server chips, and are actually my second favorite server chip
ajax: which isn't saying much, since there's only three major server chips, and they're all awful in one way or another
Daekdroom: Not having money to buy brand new hardware can be saddening, I suppose.
ajax: MostAwesomeDude: i don't think the trident dri driver ever worked very well
ajax: i'd be surprised if it textured
MostAwesomeDude: ajax: It's not like I work very well, either. :3
MostAwesomeDude: But I guess it's one more thing on the plate. I already have things on the plate.
ajax: stick to things with shaders
MostAwesomeDude: Yessir.
ajax: although, i think we did get the volari docs at one point, and if they're really dx9 parts they should have some shader support
chithead: the current amd g34 boards have matrox g200
ajax: if only they did
ajax: really it's "g200se"
MostAwesomeDude: Heh, kind of like ES1000.
ajax: which is sort of like the difference between radeon 9200 and 9200se, only, much much worse
Daekdroom: Lol
MostAwesomeDude: ajax: If the Volaris really do have shaders, they will make the most crippled, hilariously bad Gallium driver s ever.
ajax: speaking as someone who has to care about server chips for their day job, rn50 >> volari z* >>>>> g200se
Daekdroom: I remember when this store tried to sell me a 9200se back when we had X1950XTX as top notch..
Daekdroom: evil_core, what the hell does pageflipping do?
ajax: pageflipping is an alternate method of doing glXSwapBuffers where, instead of blitting from back buffer to front buffer, you update the scanout engine to point to a different region
ajax: so you essentially "flip" the back and front buffers on alternate frames
evil_core: Daekdroom: insted of copying back to frontbuffer, you change pointers, quicker and needed for vblank
Daekdroom: So it prevents generating useless data?
evil_core: no
ajax: not "needed", but much easier.
evil_core: useless operation if yo ucan call it useless
evil_core: but you dont anyerate any more data w/o it
evil_core: Daekdroom: you know whats double buffering?
Daekdroom: evil_core, yeah
evil_core: os its strictly related to that
evil_core: copying takes more time than switching between buffers
Daekdroom: Hm.. I see.
evil_core: I got UXGA T60p, its nearly as 1080p in number of pixels, so I got significant improvement with it
evil_core: but still nbot as quick as UMS which doesnt support it ;)
evil_core: ok I am going sleep
evil_core: good night
amarsh04: is fine with his 9200se (-:
amarsh04: passive cooling, appropriate for a PII-266
Daekdroom: I have a Core 2 Duo E4300, and a not much more powerful videocard xD
agd5f: evil_core: there's no pageflip support for dri2 yet
soreau: agd5f: You will start sounding like a broken record here soon if you have to say it again tomorrow ;)
agd5f: :)
Obscene_CNN: agd5f, you speak r600 don't you?
agd5f: Obscene_CNN: more or less
Obscene_CNN: does the proposed solution I made here make sense to you https://bugs.freedesktop.org/show_bug.cgi?id=27901
NTU: oh right. Obscene_CNN if you want to ask about r700_assembly.. ;)
agd5f: Obscene_CNN: yeah, probably. Richard would know better
agd5f: Obscene_CNN: see the notes section of this page: http://dri.freedesktop.org/wiki/R600ToDo
agd5f: Obscene_CNN: also the r700 ISA document has details as well
Obscene_CNN: okay thanks, I'm going to try and take a stab at figuring out how to implement it.
Obscene_CNN: I probably will end up asking Richard to do it though
Obscene_CNN: according to http://www.x.org/docs/AMD/r600isa.pdf it looks like the sin function must be in radians
Obscene_CNN: sin and cos that is
agd5f: R600 trig functions input must be normalized from radians by dividing by 2*PI. Valid input domain is [-256, 256] which corresponds to an unnormalized input domain of [-512*PI, 512*PI].
Obscene_CNN: yes, but the reported workaround implies differently
Obscene_CNN: oh, well its quitting time here
Studio271: I don't know where to go to mention this, and I'm not exactly an experienced programmer, but...
Studio271: for piglit, I had to throw a "#include
Studio271: in order to get it to compile
Studio271: probably a recent gcc change, or my system is crazy, who knows. I'm sure if it was important, someone would have noticed it by now. ;/
mentor: Studio271: What error?
Studio271: mentor: /home/drew/DOWNLOADS/piglit-v1/tests/glean/environ.cpp:68:55: error: ‘snprintf’ was not declared in this scope
mentor: Yeah, that's fixed in git, for 4 days :)
mentor: Thanks anyway :)
Studio271: hm... ohhhh... I got v.1. >.<
mentor: Such is development
Studio271: and here I was, thinking I was off to a good start in trying to get into maybe doing a little work on the driver for fun.
mentor: I wouldn't say that you haven't
Studio271: the tests are fun to watch
mentor: :)