Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

Radeon IRC Logs For 2009-12-28

Search This Log:


cxo: you know we should have a kconfig snippet on the wiki to help folks like Demo1
Demo1: where is general setup? :(
Demo1: How to enter?
tstellar: Demo1: Are you running menuconfig ?
Demo1: nop
tstellar: Demo1: Did you run genkernel --menuconfig all
Demo1: ou
Demo1: w8
Demo1: :D
gregorian: Hi, I came here yesterday to resolve my 3.5 minutes gnome startup. I've written a script to read X's log every one second. I found the regions where startx takes a lot of time.
amarks: what is this fb patch backported by airlied to xorg-server 1.7.3.902 about? what issue(s) was it for?
gregorian: http://www.pastebin.org/69286
gregorian: That is Xorg.0.log being 'tail'ed every one second
Demo1: i dont have there enable deprecated sysfs...
gregorian: The regions following the lines that repeat for several seconds must be the ones that are the bottlenecks
tstellar: Demo1: What about enabling the radeon driver, can you do that?
Demo1: i need to put there asterisk?
bridgman: gregorian; do you have a matching xorg log ?
gregorian: I'm sorry?
tstellar: Demo1: Yes if there is an asterisk, then that means it is built-in if there is an 'M' then it is built as a module.
bridgman: what you pastebinned is a once-per-second sampling from the xorg log
bridgman: do you have the actual log as well ?
Demo1: thewe was nothing
gregorian: This is my complete xorg: http://pastebin.com/69073
Demo1: at direct rendering manager
tstellar: Demo1: Go ahead and enable the Direct Rendering Manager.
tstellar: Demo1: Then you can select the radeon driver.
Demo1: i put there asterisk
tstellar: Demo1: For the radeon driver?
Demo1: ATI radeon
tstellar: Demo1: Yes enable that feature.
Demo1: done
tstellar: Demo1: Ok, exit menuconfig
bridgman: gregorian; second link doesn't work - one is www.pastebin.org, other is pastebin.com, was that deliberate ?
Demo1: can i fix udev?
Demo1: or i dont need that?
tstellar: Demo1: You can do that later. Someonen in #gentoo can help you with that question.
gregorian: It was, but I uploaded it again today morning. Could you give me sometime? I'll see if I can upload a new one
Demo1: ok
bridgman: www.pastebin.org/69073 seems to work
Demo1: ok, building kernel
tstellar: Demo1: When the kernel is done building, reboot your system and it should work.
Demo1: I hope so
tstellar: Demo1: Although you might want to check /boot to see if genkernel copied the new kernel image there. I can't remember if it does.
Demo1: how can i check if everything is ok with drvier?
tstellar: Demo1: When you run glxinfo you should see something like this:
gregorian: bridgman: Yes, that is the log
tstellar: Demo1: OpenGL renderer string: Mesa DRI ...
Demo1: damn, something went wrong
Demo1: can i do it again?
tstellar: Demo1: If it says Software Rasterizer, you aren't using hardware acceleration.
tstellar: Demo1: Do what again?
Demo1: makeconf
tstellar: Demo1: Did the kernel build fail?
Demo1: yep
tstellar: Demo1: Do you know why?
Demo1: nop
gregorian: Demo1: Does it say permission denied anywhere?
tstellar: Demo1: Can you post the ouput from genkernel.
Demo1: can i just do it gain? :)
tstellar: Demo1: Well, if the kernel failed to build, it will probably fail again.
Demo1: ill try
Demo1: if it fails again
Demo1: then
tstellar: Demo1: You should be able to determine what went wrong by looking at the ouput.
gregorian: bridgman: I would like you to know that my radeonfb loads quickly at bootup
Demo1: no wi need just w8
Demo1: *** End of Linux kernel configuration.
Demo1: *** Execute 'make' to build the kernel or try 'make help'.
Demo1: * >> Compiling 2.6.31-gentoo-r6 bzImage...
Demo1: I need to w8?
tstellar: Demo1: Yes, you should wait until it is done.
Demo1: before previous time i had to run make
Demo1: i cant see any progress
tstellar: Demo1: Genkernel supresses the build output, so you won't see any progress until it is done.
Demo1: ouh
Demo1: ok
gregorian: bridgman?
Demo1: do you know how to fix this http://dpaste.com/138456/ ?
tstellar: Demo1: Yes, like I said earlier, you need to disable that feature in General Setup->enable deprecated sysfs..
Demo1: couldnt find
Demo1: :(
Demo1: damn
tstellar: Demo1: It should be the 17th item in the General Setup section.
gregorian: It looks like bridgman isn't here. Could anyone else look at my Xorg log sampled every second and tell me what's causing the bottleneck during startx?
bridgman: sorry, just got back - a tree fell down outside, had to check to see if it landed on anything important
gregorian: oh sorry!
bridgman: looks like maybe a long timeout during display sensing
gregorian: radeonfb doesn't suffer from the same problem because bootup is quick
gregorian: bridgman: What do you suggest?
bridgman: I haven't seen this reported with other hardware so might be specific to xpress 200
bridgman: maybe file a bug ticket attaching the two logs and mentioning that the delay seems to happen with radeon but not with radeonfb, vesa etc...
gregorian: I did not face this problem when I used Slackware 12.1 and Fedora 10. They both used the radeon driver. Is it really a bug?
gregorian: Also, one of the people who was helping me yesterday claimed he had he same configuration
bridgman: maybe, maybe not - might be specific to a different version of the driver I guess
bridgman: I haven't seen it reported with other Xpress 200's either
gregorian: Is there a way to "tell" the final configuration to radeon/X without having it scan?
gregorian: I have only one screen and do not connect external displays
bridgman: I don't think so - afaik the issue is the desktop manager making a call to X which has the side effect of doing a display scan
gregorian: Does xterm count as a desktop manager?
gregorian: It takes 1.5 minutes
bridgman: I don't remember the exact call; it's basically a "tell me about the screens" kind of call IIRC
gregorian: vesa does not need to make that call?
gregorian: radeonfb can correctly determine my resolution in an instant
bridgman: it's an randr thing IIRC, don't know if vesa or the other older drivers even implement it
bridgman: I don't really know the details; more questions -> lower quality answers ;)
bridgman: going from the vaguest of distant memories here
bridgman: so not even sure if this is the same issue
gregorian: Alright.
gregorian: How long does it usually take to resolve a bug after reporting it?
bridgman: between a day and several years
gregorian: ...
gregorian: Anyone can think of a work around?
bridgman: depends largely on the correlation between area of the bug and the area a developer happens to be working on, ie what part of the code is fresh in their head at the moment
bridgman: and on whether the problem seems to be one which is likely to affect a large number of users
bridgman: you said it didn't happen with slackware and fedora; which distro are you running now ?
bridgman: I know it's in the logs but they take a couple of minutes to open on dialup
gregorian: gentoo, but with the kernel.org kernel
Demo1: gentoo is da best
bridgman: radeonfb is disabled when you're running radeon ?
gregorian: I'm using the framebuffer as we speak, and radeon is also running.
gregorian: If it were disabled, I wouldn't be getting the correct resolution on framebuffer right?
bridgman: I don't think they are supposed to coexist; wondering if one is interfering with the other
bridgman: probably
gregorian: No, same issue even when framebuffer wasn't running
gregorian: I had to enable the framebuffer
gregorian: Besides, my slackware always had framebuffer on
gregorian: Will the channel still be active in two hours?
bridgman: it's 24/7 but probably different people / time zones
Demo1: Do i need to delete my xorg conf to work open source driver perfectly?
gregorian: I came about 2 hours ago, but no one was probably there. I'll return in 2 hours
gregorian: Thanks for everything bridgman!
bridgman: it's total luck of the draw; the devs are more or less scattered across the time zones
bridgman: no prob, wish I could help more
Demo1: ?
gregorian: I expected so. Thanks though!
gregorian: Demo1: Try to run it without xorg.conf.
gregorian: If it doesn't work, use xorg.conf
Demo1: ok
tstellar: Demo1: You might want to run X -configure to generate a new xorg.conf, if you were trying to use fglrx before.
Demo1: but first i need without xorg.conf
Demo1: ?
tstellar: Demo1: You always need an xorg.conf
Demo1: nop
Demo1: not always
gregorian: tstellar: Gentoo handbook says not to use xorg.conf unless it's necessary
Demo1: yeah, when i was using ubuntu i didnt have xorg.conf
gregorian: Demo1: You won't have anything but a little time to lose if you try it without xorg.conf right?
Demo1: so i need to reboot, then boot in, kill gdm,and generate xorg.conf
Demo1: ?
gregorian: Press Ctrl + Alt + F2 instead
gregorian: See if you get a terminal
Demo1: but i will have to kill x
tstellar: Demo1: I am pretty sure X won't start without xorg.conf.
Demo1: to be able to run Xorg -configure
Demo1: tstellar: will start
Demo1: Configuring xorg.conf should be seen as a "last resort" option. It really desirable to run without one if possible, and to do all your configuration via HAL policy files.
gregorian: tstellar: You might want to check out the configuring X section on the Gentoo site.
gregorian: I'd like to know what options I can enable for graphics chipet in Xorg.conf
gregorian: Is there a site for that?
tstellar: Demo1: I stand corrected. I just started X without xorg.conf
Demo1: :)
tstellar: gregorian: man radeon
tstellar: Demo1: Have you rebooted your system yet?
Demo1: nop, i fixed udev
Demo1: now im still compiling
Demo1: How long it would take approx to compile openoffice, I have pentium 4 3GHz ?
gregorian: tstellar: It describes the options and tells me the series of my card, but it doesn't mention what it supports
gregorian: demo1: 20 hours
tstellar: Demo1: A really long time.
Demo1: gregorian: ok...
gregorian: demo1: Do the binary install
tstellar: Demo1: Try openoffice-bin
Demo1: whats difference?
gregorian: Second one is already compiled
gregorian: tstellar?
tstellar: Yes, oppenoffice-bin is pre-built.
gregorian: I mean, I know the chipset series code and have the descriptions of the options but how do I know what my chipset supports?
bridgman: man radeon will give you a list of the xorg.conf options implemented in the driver
bridgman: there are a few more implemented in the x server but man radeon is a good start
gregorian: I referred to man radeon
gregorian: It doesn't tell which options are supported by which cards
taiu: gregorian: what's your cpu usage, top processes when you run radeon drv
Nightwulf|work: hi all
Demo1: and how long it would take to compile compiz?
gregorian: taiu: X takes 5.1% of CPU
bridgman: are you looking for which options your chip can support in theory, or how well each is working today ?
bridgman: afaik all chips support all options, at least I can't think of any that can't in theory
gregorian: bridgman: I want to know what options I can enable in xorg for my chipset.
bridgman: I don't remember any of them being chipset-specific
bridgman: a few are bus-specific
bridgman: just taking another look to be sure...
gregorian: I saw some option (commented option) related to LVDS in the xorg.conf. If that's related to my problem, can I do something to it?
bridgman: all the mobile parts have lvds IIRC
gregorian: Does that mean I can't do something about this problem by editing xorg.conf
Demo1: gregorian: What you want to do?
bridgman: we don't know what the problem is, so your question is hard to answer
bridgman: Demo1; delays during X startup, seemingly related to display sensing
Demo1: ou w8 i have something... with me
bridgman: gregorian, is it just startup or app launch as well ? you mentioned xterm IIRC
Demo1: http://dpaste.com/138463/
gregorian: I don't know what 'IIRC' means but I have no problems (as far as I can see) with applications. Just the startup.
bridgman: If I Remember Correctly
bridgman: was it xterm you asked about ?
Demo1: I once added something to xorg.conf and everithing went a bit faster
gregorian: Xterm took 1.5 minutes through xinit xterm
gregorian: Demo1: My problem lies in the time taken to startup
tstellar: Demo1: That is normal.
Demo1: tstellar: so i can ignore?
bridgman: ok, so that involved an X start
bridgman: apps launch fast ?
gregorian: Decently fast
gregorian: They are not slow
tstellar: Demo1: Can you post your /boot/grub/menu.lst and ls -l /boot
Demo1: http://dpaste.com/138464/
Demo1: and http://dpaste.com/138465/
Demo1: i need to add make.conf real_root=/dev/sda3 ?
tstellar: Demo1: No that line should go in your /boot/grub/menu.lst, but it is already there.
Demo1: so i can reboot?
tstellar: Demo1: Is 7:26 close to your local time?
Demo1: nop
Demo1: local time=9:34
tstellar: Demo1: Ok, it must be utc, You can reboot now.
Demo1: can i?
tstellar: Demo1: Yes.
Demo1: ok
gregorian: bridgman: DefaultConnectorTable looks interesting
bridgman: I think that's for chip-to-connector wiring, not display sensing (?)
gregorian: It says something about using driver defaults for each chip
Demo1: driver is working
Demo1: i can see it now
Demo1: no problems with udev
Demo1: :)
tstellar: Cool, can you post the ouput of glxinfo
gregorian: bridgman: LVDSProbePLL also looks interesting.
Demo1: http://pastebin.com/d2388f45f
gregorian: bridgman: The problem is with sensing something right? If it doesn't sense anything then will it solve my problem
bridgman: I'm *guessing* that the delay has something to do with *display* sensing
bridgman: something where the code has timeouts and is taking those timeouts for some reason
bridgman: but I don't know
gregorian: If I turn them off and try I don't lose anything right?
tstellar: Demo1: It looks like you are still using the software driver.
Demo1: :/
Demo1: what should i do?
tstellar: Demo1: Can you post xorg.log.
bridgman: gregorian; right
Demo1: http://pastebin.com/d70be62c7
bridgman: this is a laptop, right ?
gregorian: yes
bridgman: if you see the LCD display "bloom" turn the power off ASAP
bridgman: there are rumors that's bad for the LCD, let's be safe
Demo1: do you think i need to generate xotg.conf?
gregorian: bloom? I will take 5 seconds of holding down the power button to turn off
gregorian: What do you mean by bloom?
bridgman: go white, look funny, like a CRT with the brightness turned up too high, anything like that
tstellar: Demo1: I think you need to enable AGPGART in your kernel.
Demo1: omg
Demo1: :D
Demo1: i need to make menuconfig?
tstellar: Device Drivers->Graphics Support->AGP Support
gregorian: I will try DefaultConnectorTable first. Here goes..
tstellar: Demo1: yes
Demo1: cd /usr/src/2.6.31-gentoo-r6 && make menuconfig ?
gregorian: gnome loaded mucccccccccccccch faster
tstellar: Demo1: yes
bridgman: interesting
gregorian: less than 1 minute
gregorian: My log file is much smaller
Demo1: AGP Support->ATI chipset support?
tstellar: Demo1: Maybe, can you post the ouput of lspci.
gregorian: My screen looks beautiful. It's much better than what it is when it took 3.5 minutes. Does this mean that the radeon default takes *longer* time and loads a *worse* configuration
Demo1: tstellar: http://pastebin.com/d608ddb7d
bridgman: gregorian; that might just mean it was sensing fewer connectors
bridgman: can you describe "better" ?
gregorian: Screen appears to have been at a lower resolution than what I assumed. Font looks crisp and sharp.
Demo1: tstellar: ?
tstellar: Demo1: I think you need to enable Intel chipset support, but I am not sure, so why don't you enable intel and ati.
Demo1: ok
gregorian: bridgman: What's a connector?
Demo1: all warnings i can ignore right?
Demo1: when compiling
gregorian: yeah
tstellar: Demo1: Yes
bridgman: either a jack on the back of the laptop where a display can be connected, or the internal connector to the display, or if you have a docking station one of the connectors on the dociking station
Demo1: i hope this time will be ok
bridgman: bios knows about the docking station connectors and checks them even if docking station not connected
Demo1: i can see right now changes
Demo1: how can i delete ATI catalyst control system from system->preferences?
bridgman: gregorian, 3am & time for zzz
bridgman: I would still think about filing a bug ticket so devs have something to look at
Demo1: compiling stopped
Demo1: confused
tstellar: Demo1: Was there an error?
Demo1: http://pastebin.com/d39334d24
tstellar: Demo1: Did you use genkernel to compile it?
Demo1: nop
Demo1: cd /usr/src/linux-2.6.31-gentoo-r6 && make menuconfig
Demo1: scripts/kconfig/mconf arch/x86/Kconfig
tstellar: Demo1: You should use genkernel: genekernel all
Demo1: why?
Demo1: so i need to recompile?
tstellar: Demo1: Becuase genkernel will compile and install everything for you correctly.
Demo1: or i can try to run make
Demo1: so what to do?
gregorian: tstellar: I have solved my problem. Bridgman advised me to report a bug. Where do I do it?
tstellar: Demo1: I don't know how to generate initrd by hand, so I can only help you if you use genkernel.
tstellar: gregorian: If your problem is solved, then maybe you don't need to file a bug.
Demo1: ok, ill try another time
gregorian: initrd? I never used any initrd for my kernel. Demo1, isn't the handbook working for you?
Demo1: tstellar: what was te commant for make all configure
tstellar: Demo1: If you run genkernel all and then reboot, I think this time your problem will be fixed.
gregorian: tstellar: I disabled BIOS probing of scanners and forced the default driver for the connectors. If you feel that's ok, I'm fine with it
Demo1: just genkernell all?
tstellar: Demo1: yes
Demo1: the previous settings saved?
Demo1: at menuconfig
Demo1: ?
tstellar: Demo1: They should be , but you can do genekernel --menuconfig all if you want to double check just to be safe.
Demo1: ok
gregorian: tstellar?
tstellar: gregorian: I am not really sure what you did, but if it works for you, then you don't need to file a bug.
gregorian: Alright, thank you all for your time.
Demo1: apgart wasnt
Demo1: hope everithing goes well again
Demo1: but how to delete ati catalyst control center?
tstellar: Demo1: Did you enable agpart->ati/intel chipsets.
Demo1: yep
tstellar: Demo1: Ok, cool
tstellar: Demo1: How did you install catalyst control center?
Demo1: maybe emerge -av ati-drivers ati-drivers-extra
tstellar: Demo1: emerge --unmerge ati-drivers-extra should remove it.
Demo1: seems ive removed
Demo1: :/
Demo1: ou i know how
Demo1: with sh
Demo1: sh ./radeon-display-driver-i-dont-remember.run
Demo1: how can i remove that?
tstellar: Demo1: I am not sure the best way to uninstall that.
Demo1: can i do it with this: run emerge --depclean
Demo1: ?
tstellar: Demo1: If you didn't install it with emerge, I don't think that will work.
Demo1: hmm
tstellar: Demo1: Has your kernel finished compiling?
Demo1: nop
tstellar: Demo1: I have to step away for a few minutes, when your kernel is done compiling reboot and post glxinfo.
Demo1: ok
Demo2: http://pastebin.com/d5048017f
tstellar: Demo2: It is working!
Demo2: wow
Demo2: its so cool
Demo2: Thant u so much
Demo2: :)
Demo2: ;)
tstellar: Demo2: No problem, if you want to try an xorg.conf file, there are many options you can use to improve performance.
tstellar: Demo2: man radeon has a description of the options.
tstellar: Demo2: And running X -configure will generate a working xorg.conf file for you.
Demo2: ok, i will chech
Demo2: check
tstellar: Demo2: But maybe for now just enjoy the open source driver as it is ;-)
Demo2: Xorg -configure
Demo2: For now till what?
tstellar: Demo2: I am just mean maybe you can use it for a little while before you try to tweak the settings, but it is up to you.
tstellar: Demo2: I have to go now, good luck.
Demo2: ok bye
_KAMI_: hi
moobie: hi
moobie: Anyone tried kernel 2.6.33-rc2 with r700?
mcgreg_: yes
mcgreg_: rv730xt aka 4670 here
moobie: mcgreg_, do you experience any problems with it? And what distro are you using?
mcgreg_: no problems at all. I use debian unstable amd64. the kernel I compiled myself is from www.kernel.org
mcgreg_: what problems do you have?
moobie: My screen is all black if I use 2.6.33-rc2
moobie: with 2.6.32-9 I don't got that problem
moobie: Though with 2.6.33 I don't see any errors whatsoever
mcgreg_: hmm do you use kms? did you install thta new firmware file?
moobie: I installed the irq firmware file
moobie: Do I need new of the other firmware files?
mcgreg_: it is called r700_rlc.bin
moobie: I got that one
moobie: /lib/firmware/kernel-2.6.33*/radeon
moobie: And I see it loads fine
mcgreg_: and when does your problem starts? after in itializing kms or when starting xorg?
moobie: Both
moobie: No sry
moobie: I actually see the splash screen
moobie: but with a lower resolution than with 2.6.32
moobie: But if I look in my kern.log file afterwards, it loads kms just fine with 2.6.33. Also the irq firmware
moobie: mcgreg_, though I see something different with 2.6.33. This line:
moobie: kernel: [ 20.811585] Console: switching to colour frame buffer device 240x67
moobie: The current 2.6.32-9.13 kernel from Ubuntu works fine with KMS enabled, it looks to have drm upto the .32 release plus a few backported patches. The latest Ubuntu kernels from the mainline along with a custom compiled DRM-Radeon-Testing applied to Linus' git and Ubuntus 32-9.13 git source have all failed the same way.
moobie: It might be a drm problem?
mcgreg_: well... I really dont know. but it might be a framebuffer problem perhaps
mcgreg_: I had something similiar
mcgreg_: once
mcgreg_: but kms reported there is some different framebufer loaded ... and later it had problems
mcgreg_: but I really cannot tell you since i'm not an expert to this at all
moobie: ok
moobie: thx for you try :)
moobie: I'll try ssh into it and maybe fill a bugreport
mcgreg_: you're welcome+
moobie: <- booting into 2.6.33 ;)
amarks_: is sauerbraten meant to render correctly?
amarks: i turned off shaders but still looks strange
moobie: hi
Demo1: hi
moobie: Anybody know when the radeon devs are here?
Demo1: nop
amarsh04: bridgman was here 2.5 hours ago
moobie: O :)
moobie: Ok*
Kurko: Does anybody know what is broblem: when I enable KMS then I get this error: (EE) RADEON(0): [dri] RADEONDRIGetVersion failed because of a version mismatch.
moobie: Kurko, actually thats the exact same problem I have
moobie: lets solve this together :)
Demo1: Version mismatch :D
moobie: I only have this problem with 2.6.33, but not 2.6.32
soreau: Kurko: moobie: You need to build mesa against latest libdrm
moobie: soreau, well actually I only got the problem with kernel 2.6.33
Demo1: the newest one which i gen was 2.6.31
soreau: moobie: Does fbcon module get loaded before radeon?
Demo1: Which distro u have?
moobie: Ubuntu
Demo1: oh
Demo1: ok
moobie: soreau, want to see my xorg.0.log files for both 2.6.32 and 2.6.33? There os definetly something wrong in the 2.6.33 kernel
moobie: Xorg.0.log with kernel 2.6.32: http://pastebin.com/m5ab92eee
moobie: Xorg.0.log with kernel 2.6.33: http://pastebin.com/m54edb7d0
Kurko: I build mesa and libdrm but I get still this error
moobie: hmm kernel 2.6.33 is reporting: (II) [KMS] drm report modesetting isn't supported. in my xorg.0.log file
soreau: moobie: Is the radeon module loaded per lsmod?
moobie: soreau, I can't say because my screen is all black
moobie: Well I could ssh into my computer though
Demo1: My screen went black on Xorg configure
Demo1: :/
Demo1: Now im not using xorg.conf
moobie: Demo1, actually I think its related to my errors in Xorg.0.log ;)
Demo1: moobie: :D
moobie: soreau, I will post my dmesg with the kernel 2.6.33
moobie: soreau, http://pastebin.com/m7b2992c0 <- kernel 2.6.33 log
Demo1: When I had ubuntu, i didnt have xorg.conf
moobie: Hmm,
moobie: (EE) RADEON(0): [dri] RADEONDRIGetVersion failed because of a version mismatch.
moobie: [dri] This chipset requires a kernel module version of 1.17.0,
moobie: [dri] but the kernel reports a version of 2.0.0.[dri] If using legacy modesetting, upgrade your kernel.
moobie: [dri] If using kernel modesetting, make sure your module is
moobie: [dri] loaded prior to starting X, and that this driver was built
moobie: [dri] with support for KMS.
moobie: [dri] Disabling DRI.
moobie: Is 2.0.0 higher than 1.17.0?
Demo1: it seems to be so
soreau: moobie: Looks like fbcon never got loaded afaict
moobie: soreau, could I ssh into my box, stop gdm and drm, and then load fbcon?
moobie: or would it already be too late
moobie: and stop radeon ofcourse
moobie: brb
Kurko: is it right to use --refix=/usr when build ddx and mesa and drm?
seb_: Kurko: following the wiki --prefix=/opt/xorg should do it fine
Kurko: now I got this error when I use KMS:
Kurko: (EE) AIGLX error: Calling driver entry point failed
Kurko: (EE) AIGLX: reverting to software rendering
mcgreg_: does suspend to ram works for anybody on kms (with 2.6.33)?
MichaelLong: not for me: AGP + rv250
soreau: works for me with drm-radeon-next: AGP rv350
soreau: but tv-out gets no signal after resume :P
mcgreg_: well, on my radeon 4670 pci-e it does not
soreau: Kurko: Did you build mesa against the libdrm you installed?
seb_: i have /opt/xorg/lib/libGL.so: file not recognized: File truncated every time i build mesa
seb_: anyway i can't build xf86-ati and mesa on a debian testing whatever i try ...
soreau: Did you try simply installing libdrm to /usr then try building the other two again?
Kurko: soreau: first I build xf86-video-ati and then drm and then mesa is this right order?
soreau: no, you want to build libdrm first
seb_: thx
dougztr: I've noticed that my displaylink usb monitor segfaults Xorg in conjunction with radeon, but it works with vesa. Anybody have a workaround or patch?
Kurko: I got same error even tough I built libdrm first
moobie: hi
moobie: does any of you get messages from a bot, which is trying to send you nasty things?
Wizzup: moobie: do you have any idea what channel it originates from?
moobie: Wizzup, I am only in #ubuntu and #radeon
moobie: It might be from #ubuntu
Wizzup: May be possible
moobie: Is my mesa or ddx too old if I get this message in Xorg.0.log?
moobie: (EE) RADEON(0): [dri] RADEONDRIGetVersion failed because of a version mismatch.
marvin24: probably drm to old
moobie: marvin24, drm in kernel or libdrm?
marvin24: moobie: libdrm
moobie: marvin24, ok. I get the problem with latest kernel 2.6.33-rc2
moobie: kernel 2.6.32 had no such problems
marvin24: what are you running?
moobie: marvin24, Ubuntu with tormod ppa. I have libdrm 2.4.17+git20091221
moobie: Should be early enough?
marvin24: yes
marvin24: maybe you can post the output of xorg.log somewhere
moobie: sure
moobie: Xorg.0.log with kernel 2.6.32: http://pastebin.com/m5ab92eee
moobie: Xorg.0.log with kernel 2.6.33: http://pastebin.com/m54edb7d0
moobie: marvin24, there you go. And thanks
marvin24: your ddx has changed
marvin24: (self build without kms)
moobie: marvin24, it has?
marvin24: (II) RADEON(0): Built from git commit 9d0f3af7278dc939fd4e6f3ea69d9f488a9fbed7
moobie: marvin24, just to be sure. The ddx is the xorg radeon module?
marvin24: yes
moobie: Well I haven't installed another ddx between the two run
moobie: I just rebooted and started the other kernel
moobie: marvin24, is the radeon the drm kernel radeon module?
marvin24: btw. I'm using http://ppa.launchpad.net/xorg-edgers/ppa/ubuntu for the packages
moobie: marvin24, me2
marvin24: no its the ddx
moobie: Well I use the lucid packages for karmic
moobie: To get mesa 7.8 with glsl support
marvin24: there also may be problems with firmware loading
marvin24: can you also post the output of dmesg?
marvin24: and kernel 2.6.33rc2...
moobie: http://pastebin.com/m7b2992c0 <- kernel 2.6.33 log
moobie: marvin24, do you need the one with 2.6.32?
marvin24: do you have installed the firmware files (r{6,7}00_rlc.bin)
marvin24: no
marvin24: only > 32 - I think
marvin24: seems to be ok, firmware gets loaded
moobie: marvin24, yes I have. It seems to load it fine
moobie: As of what I can see from the log
marvin24: so there must be a problem with drm/ddx somewhere - you should enable experimental api for drm
marvin24: if you build it yourself
moobie: I haven't build it myself
moobie: I use tormod and mainline kernels
moobie: marvin24, are you using lucid or karmic?
marvin24: karmic
moobie: with mainline kernel and karmic repo?
marvin24: currently 2.6.32.2 + some patches
marvin24: I'm not using tormod
marvin24: I normaly compile everything (drm, mesa and radeon) myself
marvin24: what is the difference between xorg-edgers and tormod?
moobie: Actually its xorg-edgers, but tormod made them :)
moobie: marvin24, tanks for you help
moobie: I will try reinstalling xorg-edgers from karmic repo instead of lucid later
moobie: bye
marvin24: good luck
idletask: Hello
soreau: If your mouse cursor turns into a square of garbage, is there any way to 'reload' it to make it show the cursor?
marvin24: switch resolution/console and back
soreau: marvin24: Thanks for the suggestion, but neither of those helped
marvin24: anyway, there is something wrong here
idletask: soreau: and restarting X altogether?
soreau: idletask: I'm pretty sure that would fix it, but I was wondering if there was anything else I could do
idletask: soreau: well, better check that, who knows :p
idletask: soreau: I've had my cursor disappear, but this was only a case of switching to console and back - but I've never had garbage, except when playing Dungeon Siege under wine
soreau: idletask: Well I started another X session on a different vt, played a couple games, then switched back to this one and it was the garbage square so not exactly just using 'normally' :)
idletask: Ah, yes
idletask: Which makes me think that I have a panic if I just launch a bare X server on another vt and start a wine game
idletask: But...
soreau: I get something similar, have to turn off second output on both x sessions before the game will not freeze the box
idletask: soreau: I only have one output
idletask: Well, one screen, I mean
soreau: idletask: I don't use wine
idletask: Even though my card has two outputs
idletask: I do :p
idletask: And I regret to admit that fglrx does a much better job with it right now
idletask: One thing is that I can launch games in an emulated desktop with fglrx - I can't with radeon
idletask: "emulated desktop" == "window"
spstarr: I've a question and I want an opinion if you think this is a bug in the simulator (both LL and opensim's):
spstarr: If someone in a special group rezzes an object that is too big for the sim to render, the simulator returns non-deeded items. Worse, if those non-deeded items are from the sim landowner they get returned to them. feel this is a design flaw. If there are objects rezzed in the sim, they should never be returned unless the sim owner sets any autoreturns. The sim should not be returning non-deeded items automatically it should instead notify the user
spstarr: that the object being rezzed is too big to fit and reject the rendering of it altogether. I would suggest changes to the simulator to not return non-deeded items because people dont always deed items (or forget to) and this causes a big mess.
idletask: I can't help at all, I don't know what opensim and LL are, for starters :/
_KAMI_: hi
spstarr: er
spstarr: that was wrong channel
spstarr: eep
kdekorte: Anyone having issues with the drm-radeon-testing branch?
spstarr: kdekorte: what sort? im using that as of 2 days ago
kdekorte: spstarr, http://bugs.freedesktop.org/show_bug.cgi?id=25779
spstarr: it hasn't been updated in 5 days
kdekorte: yeah it has been dead for me for 5 days
Ghworg: It's almost as if everyone is on holiday or something
kdekorte: They deserve it...
kdekorte: I was just wondering if this was a problem that only I was seeing or not
Wizzup: Is there a way to see the temperature of my GPU? something in /proc/ I suspect?
Ghworg: Not yet
Wizzup: Ok
maderat: bug kms mesa master is known? Segmentation fault at address 0x100000004
kdekorte: maderat, how do you create it?
maderat: some staff from koji fc13 libdrm and some mesa + kernel 2.6.33-rc2
kdekorte: did you build mesa yourself?
MostAwesomeDude: Don't mix'n'match; libdrm's API changed recently and you probably should build them all together.
maderat: kdekorte you need backtrace?
MostAwesomeDude: At the least, use packages from the same repo.
kdekorte: no... just some basic questions first
kdekorte: if you build mesa yourself, run make distclean and then rebuild
maderat: like MostAwesomeDude say
maderat: make clean && make && make install
maderat: 3 times
kdekorte: did your try make distclean
kdekorte: and then rerun configure?
maderat: no
kdekorte: try that
kdekorte: and make sure if you are using libdrm from koji you have libdrm-devel from koji as well
maderat: libdrm-devel-2.4.17-1.fc13.x86_64
seb_: hey guys i have the 'black symptom' on a precompiled deian 2.6.32-trunk kernel, at boot KMS is enabled as soon as i launch Xorg, it starts and sounds working behind a black screen...
seb_: xorg.0.log report http://pastebin.com/m3059dfe5
seb_: do you happen to know if behavior would change on a 2.6.33 vanilla one ? does it worth trying building a new one ?
seb_: would someone give me a hand ?
amarsh04: seb_ I see 2 (EE) lines in your Xorg.0.log
dougztr: hey all, I'm trying to get my displaylink mini usb monitor working in conjunction with radeon, but it causes a segfault. the displaylink device does however work with the vesa xorg driver. anybody here have one of these or know of a solution?
amarsh04: first: [dri] RADEONDRIGetVersion failed because of a version mismatch
amarsh04: second: Acceleration initialization failed
amarsh04: Is anything not stock Debian?
seb_: amarsh04: everything is git freshly pulled and built from nongit.freedesktop.org/
seb_: +a
MostAwesomeDude: seb_: You need to rebuild your libdrm with --enable-radeon-experimental-api then rebuild DDX.
Demo: what can i do to speed up my radeon?
Demo: i hate delayings when running compiz
hifi: buy a better card
seb_: MostAwesomeDude: bullshit, i did that plenty of times
MostAwesomeDude: seb_: The error in your log means that your DDX isn't built with support for KMS, which usually means that your libdrm doesn't support it either.
Demo: on ubuntu compiz went better
Demo: much
amarsh04: seb_, please be polite, I haven't rebuilt from git recently, and had a few false starts before being successful
amarsh04: you haven't provided any information about your setup Demo
Demo: amarsh04?
maderat: make distclean doesnt help
amarsh04: yes Demo, you were running ubuntu, what are you running now? What card? Have you checked /topic ?
Demo: Now Gentoo
amarsh04: were you on earlier with the other Gentoo folk Demo?
seb_: amarsh04: sorry it drives me crazy
idletask: Demo: I run gentoo and have the ddx working
idletask: Demo: do you use the x11 overlay?
Demo: y
monkey_mouse: hi, i have a radeon 9550 and i'm having an issue with the proprietary drivers, they are working fine for the moment but if i log out ubuntu goes into the safe graphics and i have to install them again, any ideas?
MostAwesomeDude: monkey_mouse: #ati
hifi: nobody bothered to read he has 9550 which has not been supported with the proprietary drivers for a long time?
MostAwesomeDude: Well, I'm not here to convert him. If, for some reason, he likes fglrx, who am I to tell him to change?
Demo: Is it possible to install fglrx driver on radeon 9550?
hifi: I just have a feeling he's not really using fglrx
MostAwesomeDude: After all, if he can get fglrx to behave, he probably also can wrestle bears and do other similarly hard-boiled things.
MostAwesomeDude: Demo: Nope, support was dropped a while ago. You need to use the open-source drivers.
Demo: yep
Demo: so how to speed up
Demo: ?
Demo: there should be something
Demo: some way
Demo: only way i see xorg conf
Demo: but i dont know really
hifi: MostAwesomeDude: you made my day :)
monkey_mouse: well i should mention i'm using hardy, i had them working for a while but i tried to change my screen resolution and it crashed and had problems ever since
hifi: monkey_mouse: *do* you want to use fglrx for real?
amarsh04: Demo, do you have Option "AccelMethod" "exa" in your "Device" section of xorg.conf?
MostAwesomeDude: monkey_mouse: Well, if you want to stick with Hardy, then you'll need to go to #ati.
Demo: my xorg.conf is empty!!!!
monkey_mouse: should the openGL work fine with the new version 9.1 of ubuntu
hifi: monkey_mouse: depends what you use
hifi: overall it should work fine
hifi: you can always test it with a livecd
hifi: though I must admire you are tunning hardy just to get fglrx working
hifi: s/tunning/running/
monkey_mouse: one thing that is weird is that my svideo output is set as the default display now, it was not before i reinstalled the newest drivers from ati's website
amarsh04: Demo, I just posted my entry for a radeon 9200se (AGP) on http://paste.debian.net/55071
hifi: monkey_mouse: you shouldn't install newest drivers from ati's website...
Demo: its only what u have on your xorg.conf?
monkey_mouse: where should i get the drivers?
hifi: maybe better if you go to #ati and ask there
amarsh04: that's the total non-comment lines on my xorg.conf Demo
Demo: amarsh04: ok
amarsh04: on Debian "man radeon" provides information about options and defaults
tsdgeos: hi, what's the level of support for Radeon HD 4650/RV730PRO ?
tsdgeos: reading http://www.x.org/wiki/RadeonFeature and http://www.x.org/wiki/radeonhd%3Afeature i understand 2d works but 3d does not
tsdgeos: is that right?
adamk_: tsdgeos, 3D works. Certainly well enough to run compiz, ut2004, nexuiz, openarena, etc.
tsdgeos: cool
tsdgeos: then either that pages are old or i am not able to understand them well enough :D
BioTube: OpenGL 2.0 is supported
adamk_: tsdgeos, However, I don't think any distribution currently ships with the code necessary for that functionality.
adamk_: F12 comes close, but does require an extra package to be installed.
adamk_: tsdgeos, It's quite possible that it's out of date. That page seems to be specific to the radeonhd driver (which, btw, has it's own IRC channel).
adamk_: tsdgeos, But accelerated 2D is available via both the Xorg radeon and radeonhd driver, and Mesa provides the 3D support.
tsdgeos: adamk_: first link is about radeon driver as i understand
adamk_: Ahhh, I missed that first link for some reason.
tsdgeos: adamk_: anyway good to know it works
tsdgeos: didn't want to buy a laptop with unsupoprted hardware :D
MostAwesomeDude: Just avoid HD 5xxx for now.
tsdgeos: adamk_: i understand next revisions of distros should include that support by default? F13, ubuntu 10.04?
adamk_: tsdgeos, Most likely, yes.
tsdgeos: nice :-)
seb_: whatever i do ./configure --enable-radeon-experimental-api when building libdrm i NEVER get KMS support for libdrm and ddx either, i really dunno where investigate ...
MostAwesomeDude: seb_: At the end of libdrm configure, does it say that the Radeon API was enabled?
seb_: no
_KAMI_: I hi
soreau: seb_: Where are you installing libdrm?
seb_: soreau: in /usr
_KAMI_: I am plan to use 2.6.33 rc kernel with radeonhd driver
_KAMI_: is it possible?
_KAMI_: with kms?
seb_: MostAwesomeDude: actually yes :
seb_: libdrm 2.4.17 will be compiled with:
seb_: Radeon API yes
adamk_: _KAMI_, No. radeonhd does not support KMS.
MostAwesomeDude: seb_: Okay, good. Now, when you configure your DDX, it should say "Kernel modesetting: yes" or something like that.
_KAMI_: oh
_KAMI_: adamk: I just wanted to use that because hdmi output is workable with trat
_KAMI_: that
_KAMI_: but I should try radeon with 2.6.33? or just ati driver?
adamk_: _KAMI_, OK... But radeonhd still doesn't support KMS :-)
_KAMI_: okay
adamk_: _KAMI_, radeon supports KMS. And, as far as I know, the HDMI code for audio support is available in some version of KMS.
_KAMI_: so I will try with radeon driver
adamk_: Not sure exactly where that code is at the moment, though.
_KAMI_: that is what I need:)
seb_: yes it does !
seb_: (suddenly popped from nowhere)
_KAMI_: adamk thanks!
_KAMI_: hi adamk
_KAMI_: I tried it but I have no succes with X
_KAMI_: I have some missing firmware (I used precompiled kernel)
_KAMI_: then i donwoaded it and I have no X with 2.6.33 rc2
_KAMI_: and kms with radeon driver
_KAMI_: http://pastebin.ca/1729645 few kernellogs and xorg log http://www.pastebin.ca/1729647 for kms enabled -> adamk
seb_: one man said fedora is (at the moment) a very well desktop implemented distribution, what do you guys thinking of this distribution ? (i mean regarding the future and the present of the linux desktop)
maligor: they tend to include very experimental sw in it
Hackus: Fedora 12 is quite excellent in my opinion.
Hackus: Have an entire office equipped with it.
Hackus: 12 workstations.
Hackus: 4 of them are virtual.
seb_: good on you Hackus
Hackus: I have 4 virtual Xp workstations. I must say, the performance increases in Virtualization in Fedora 12, compared to F11 is night and day.
maligor: I'm not sure I'd trust fedora for workstation usage :P
Hackus: I am a very skilled computing professional, however, so I am not sure if Fedora 12 would be good for your average I.T. grunt.
Hackus: Only problems I have with F12 is Radeon support is poor, lots of bugs. So some of our openGL engineering apps have tons of problems.
Hackus: For startes, no support for two video cards, DRI doesn't work.
MostAwesomeDude: Better be filing those bugs.
Hackus: yep.
MostAwesomeDude: I've got two video cards. Both work with DRI. Which cards?
Hackus: VGA Arb is broken.
MostAwesomeDude: vgaarb works fine.
MostAwesomeDude: Which cards?
Hackus: Our workstations have R670 cards in them.
Hackus: That would be ATI 3870x2
MostAwesomeDude: Hm. I've got an RS780.
Hackus: Well, I did try a R700 series card and it does work, it just doesn't work when you have two of them plugged in.
MostAwesomeDude: And I've successfully paired that with an RV630, an R700 of some sort (can't remember the number for HD 4850x2), an R580, an RV410 (which has a chipset-specific bug), and an RV280.
Hackus: Well, I should be more specific.
Hackus: DRI support stops working for the card.
MostAwesomeDude: I really really wanted to do three cards, and double up with either the RV280 or maybe an R128, but I don't have a third monitor.
Hackus: So, it doesn "work" you just end up with a huge frame rate slow down.
MostAwesomeDude: DRI works on both cards. How'd you set up your xorg.conf?
MostAwesomeDude: Did you do Xorg -configure?
Hackus: I tried all sorts of xorg configs for days. Couldn't get it too work.
Hackus: I rebuilt all of the software, drm, ddx, kernel 2.6.32-rc2
Hackus: Simply won't work with two cards.
Hackus: DRI, excuse me.
MostAwesomeDude: Hm.
idle_task: 2.6.3*2*-rc2?
MostAwesomeDude: Hackus: Wait, were you not using a Fedora kernel?
Hackus: I was, but that doesn't work.
idle_task: runs the drm-radeon-testing kernel
Hackus: I think my last attempt I tried using 2.6.33-rc2.
Ghworg: Doesn't dri get disabled when you use vgaarb, I'm sure I've read that
Hackus: Yep.
Ghworg: By design I mean, not a bug
Hackus: No biggy I guess, I just pulled the cards out of the boxes, and it works fine with a single card.
Hackus: Its not like the radeon can do crossfire yet.
MostAwesomeDude: vgaarb can be disabled on each card that opts out of it.
Hackus: "radeon modules:
MostAwesomeDude: vgaarb disables itself when one or fewer cards are using it.
MostAwesomeDude: All KMS chipsets opt out of vgaarb.
MostAwesomeDude: Ergo, vgaarb is not a problem for Radeons. (Except ES1000, but if you know what that is, then you know why.)
Hackus: I tried turning off KMS, loading the radeon module manually with different modeset settings. It crashes hard core.
MostAwesomeDude: Okay, what's the bug number?
Hackus: I filed the bug on redhats bugzilla.
Hackus: I would have to look it up, just a sec.
MostAwesomeDude: does not mind posting on RH bug reports
Hackus: Redhats Bugzilla is INCREDIBLY SLOW today...
spreeuw: this way only the real bugs can come through
spreeuw: bugfiling is harsh enough work as it is ;p
seb_: yeah! i got my RV710 finally working with KMS, dds and mesa 3D, glxgears report 1450FPS not too bad :-)
Hackus: I filed the bug under the xorg component, radeon.
Hackus: xorg-x11-drv-ati
Hackus: Just a question, why is there two radeon drivers in the tree?
Hackus: And what is the diff between radeonhd and radeon.
dmb: ugh
dmb: Hackus, i suggest looking on phoronix for the answer to that question
Hackus: The bug for DRI was filed on Bug 547006
Hackus: bugzilla red hat.
Hackus: God, did that take a long time for that search. :-)
spreeuw: oh glxgears at what res?
spreeuw: mine does 245 FPS
spreeuw: at 1920*1200
spreeuw: and it renders 2 cogs off screen half
spreeuw: because of the ratio I guess
Hackus: That seems slow.
spreeuw: no its ok really
spreeuw: yes with dri enabled
spreeuw: not that glx gears is no benchmark
Hackus: Oh, OK, then that is actually pretty good. :-)
spreeuw: note
Hackus: Must have a decent processor.
spreeuw: no its hardware accelerated
Hackus: it can
Hackus: t be hardware accerlrated if DRI is not active.
spreeuw: thta uses cpu too, but I had the feeling you thought its software rasterizer entirely
Hackus: What I mean to say is if DRI is turned off, you can'
Hackus: t have hardware acceleration.
spreeuw: OpenGL renderer string: Mesa DRI R600 (RV730 9490) 20090101 x86/MMX+/3DNow!+/SSE2 TCL DRI2
spreeuw: OpenGL version string: 2.0 Mesa 7.8-devel
Hackus: Is that from glxinfo?
mokoloko: 20090101 lolwut
spreeuw: yes I always run current git
spreeuw: oh that date doesnt mean much
spreeuw: probably a specification
mokoloko: yeah but it's funny :P
mokoloko: in a very lame way
lordheavy: somebody got the anholt.cfg file for openarena ?
MostAwesomeDude: Sure, one sec.
MostAwesomeDude: mokoloko: That date is the last time that one of the GL API files got regenerated.
MostAwesomeDude: Not anything special.
MostAwesomeDude: lordheavy: http://people.freedesktop.org/~airlied/scratch/oa_anholt_demo.tar.gz
lordheavy: ho thanks !
mokoloko: ok
MostAwesomeDude: Hackus: Before I comment on that bug...
MostAwesomeDude: Hackus: Did you build your own DDX?
MostAwesomeDude: Because it looks like your DDX was built without KMS, which means it was built against a libdrm without KMS.
MostAwesomeDude: Oh, wait. Also on a livecd? Hm.
MostAwesomeDude: Actually, there's like five things wrong with that bug report.
MostAwesomeDude: It'd be pretty nice if you could attach a dmesg, Xorg.0.log...
Hackus: I didn't think much about attaching the whole Xorg log as what is the point? So I just took snippets of the dmesg and Xorg, etc.
lordheavy: wow ! now 88 fps in openarena with a rs880 ! nice, last time i've checked (1 week) it was 60 fps :-)
Hackus: Anything over 60fps is sorta overkill. I don't think the human eye notices.
yangman: Hackus: it's hard to get an overall view of potential issues without the entire log. if the average layman can tell which *exact* lines are useful for any particular bug, they wouldn't be laymans now would they?
spreeuw: Hackus: no you need 100+ in a timedemo
spreeuw: 200 would be ok
Hackus: 100+ frames a second in a timedemo? Whats a timedemo?
lordheavy: i can notice a difference beetween 60 and 88 fps :-)
spreeuw: that more or less ensures it no where goes below 72
spreeuw: a timedemo is a recorded match
lordheavy: Hackus : http://dri.freedesktop.org/wiki/Benchmarking see openarena with http://people.freedesktop.org/~airlied/scratch/oa_anholt_demo.tar.gz
spreeuw: played as fast as possible
Hackus: Oh, well, for benchmarking I can see why. :-)
spreeuw: gives a more real world performance estimate
spreeuw: also mention your res
Hackus: But really, I can't tell between 60 fps in WoW vs 82 Frames a second.
MostAwesomeDude: Hackus: Well, at any rate, something's fishy, and I'm betting that it's a libdrm that's not built with the right stuff.
spreeuw: and settings
Hackus: Well, after the libdrm is built, is there a way I can get the build info?
spreeuw: I never wowed but that needs no speed skills
spreeuw: or time related skills
MostAwesomeDude: Well, at the end of configure, it'll say which APIs it has enabled.
Hackus: I use WoW as a chat client. :-)
Hackus: So what do I know. LOL
MostAwesomeDude: You need to use --enable-radeon-experimental-api.
Hackus: I did that.
Hackus: What I want to know is, after the libs are built, how do you confirm it has all the right options.
nuffilein: Hackus: for wow everything more then 30fps should be enough...but for a fast shooter you'll need at least 60 fps...
spreeuw: Hackus: follow a guide with the correct order and build flags
evil_core: http://www.mail-archive.com/bk-commits-head@vger.kernel.org/msg01043.html
MostAwesomeDude: Well, libdrm_radeon won't be built *at all* if you don't enable the experimental API.
Hackus: Guys come on, you know how many times I rebuilt my ddx, libdrm, and kernel over the past week?
evil_core: why pageflip, KMS and hyper-z patches never went into master or some other branch?
MostAwesomeDude: Don't rebuild your kernel.
MostAwesomeDude: It's fine the way it is.
Hackus: Well, I wanted fabulous IRQ support.
MostAwesomeDude: evil_core: One second.
Hackus: FABULOUS IRQ support is FABULOUS you know....
Hackus: LOL
evil_core: s/KMS/s3tc/
Hackus: Pageflip isn't supported yet on the R600/R700 GPU's yet is it?
MostAwesomeDude: evil_core: Pageflipping never made it into KMS, because nobody wrote it.
evil_core: it wsas on mailing lists, but doesnt apply on currecnt mesa
MostAwesomeDude: S3TC should work; a patch series went in last week that fixed it.
Hackus: yes and I didn't see it in the KMS sections of the source either so that confirms that.
evil_core: MostAwesomeDude: with multitexturing?
MostAwesomeDude: HiZ/HyperZ works on non-KMS. We haven't decided exactly how we want it implemented for KMS.
Hackus: OOoo....texturing.
MostAwesomeDude: evil_core: Kind of. It's known to be broken in the HW.
Hackus: I have a question?!
evil_core: so how quake3 worked in the past?
MostAwesomeDude: But we don't forbid it. Most app authors know better than to do compressed textures with multitexturing.
MostAwesomeDude: You'll notice that Q3/OA doesn't enable s3tc by default.
Hackus: Did they ever figure out how they are going to support GPU's with onboard memory in a nice standard way?
MostAwesomeDude: What do you mean? Discrete GPUs with their own memory have been supported since the dawn of time.
evil_core: its broken on r500 too?
spreeuw: MostAwesomeDude: s3tc only up to r500
Hackus: Well, I thought there was a problem that each card had to have its own memory management?!
spreeuw: with those recent patches
MostAwesomeDude: spreeuw: Well, that's because nobody unpaid (nha, osiris, myself) has done it. None of the paid guys can do it.
MostAwesomeDude: Hackus: For KMS land, yeah, but it's not really painful.
evil_core: spreeuw: uo to, means mine r500 should work with s3tc/multitexturing?
spreeuw: I only meant to clear it up for the person who asked for s3tc state in 6/700
spreeuw: r3/5 do yes
MostAwesomeDude: Right. To summarize the five or so discussions:
spreeuw: have for years
Hackus: I think I am talking about GEM guys.
Hackus: Know anything about it?
MostAwesomeDude: Arg, nevermind.
MostAwesomeDude: Hackus: Yeah, what about it?
evil_core: spreeuw: but mine HW is broken? (it got garbled textures in q3 when s3tc enabled in driconf)
Hackus: Was it ever finished?
MostAwesomeDude: A long time ago.
MostAwesomeDude: And put in the kernel.
MostAwesomeDude: Ditto TTM.
spreeuw: evil_core: no they recently broke it for r3/5
MostAwesomeDude: It's all there, it all works.
spreeuw: but another recent patch fixed it again
evil_core: bit his patch isnt in master?
MostAwesomeDude: Yet.
spreeuw: it was a drm patch in the kernel
spreeuw: iirc
spreeuw: or else libdrm
MostAwesomeDude: S3TC was broken on KMS, there's a patch, the patch will be in master soon.
Hackus: Anybody know where the GEM source files are?
evil_core: MostAwesomeDude: are this patches public?
spreeuw: evil_core: you can disable quake3's s3tc
spreeuw: it offers nothing
evil_core: more fps when its enabled, but maybe because id doesnt work
evil_core: I got 4* more fps with s3tc enabled
spreeuw: how much videomem do you have?
evil_core: 256MB onboard
spreeuw: 4fps?
spreeuw: lol
evil_core: RV5200
pillar: hello, isn't there anymore an option to change overlay in catalyst command center? I have a TV connected and it is not using the full screen area and I used to be able to correct that - now I cannot find the option anymore (command center has a new style)
evil_core: 4 * more whan w/o
MostAwesomeDude: Hackus: http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=tree;f=drivers/gpu/drm;h=1f73b304b50411d4263691b8e993b83fe865af59;hb=HEAD
MostAwesomeDude: drivers/gpu/drm/drm_gem.c
spreeuw: evil_core: yeah I figured that
spreeuw: that not alot
spreeuw: are you low that you need them then?
MostAwesomeDude: pillar: #ati
Hackus: Oh, its part of the drm directory.
pillar: MostAwesomeDude ok thanks
Hackus: I wonder why I didn't notice it before?
evil_core: yeah, tc-elite lags as hell
spreeuw: evil_core: btw disabling vsync helps 10% here
MostAwesomeDude: evil_core: http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=commit;h=2babbf3065b4274b6394efe2fa3a4a857b7ea412
spreeuw: tc_elite is quite heavy yeah
spreeuw: it has very inefficient maps
spreeuw: like the one with the trains
spreeuw: or maybe the driver is just slow for those particular tricks
evil_core: MostAwesomeDude: thx, so it means its in master?
spreeuw: seems to e theres a huge gap between gamemakers and 50% of the userbase
spreeuw: makers are all on windows with 3x fatser drivers and cards
MostAwesomeDude: evil_core: Not yet. Learn2gitweb plz.
spreeuw: evil_core: this is kernel drm
Hackus: Oooooo...look at the drm_gem.c file. Its purrrrrty. :-)
spreeuw: if you check out that branh you should have it
spreeuw: I think it should be in there if you build this kernel http://phoronix.com/forums/showthread.php?t=20484
Hackus: Ok, well....it is sorta pretty. There are non condition jumps in the code. Maybe I shuold rewrite that. :-)
evil_core: spreeuw: I know, I am using latest drm-radeonptesing
evil_core: I got obsession, making git pull many times a day and git logg
Hackus: evil_core that means your a git.
Hackus: :-)
eosie: MostAwesomeDude: I think S3TC is already in 2.6.33-rc2
evil_core: Reversed (or previously applied) patch detected! Assume -R? [n], so I got it probably yet
amarks: is sauerbraten meant to render correctly with git mesa on r700? it is rather a mess when i tried it, i turned off shaders and still broken.
mentor: Would it be a good idea to have the mesa DRI drivers log through X (it requires pulling in some more dependencies)?
evil_core: so if s3tc is in drm-radeon-testing, then it looks thats broken on r500
Wizzup: amarks: Last time I tried it X died after a few seconds in the menu
amarks: Wizzup: it hasn't crashed for me, i can configure stuff in the menu, go and start to play a game, but the rendering is very broken, it's unplayable
amarks: this is with the latest trooper edition
Wizzup: amarks: Ok, I was using an old one probably. There's only one in portage iirc
amarks: Wizzup: yeah, try the ebuild from http://bugs.gentoo.org/show_bug.cgi?id=111566
Wizzup: amarks: Will try soon,bbiab, thanks
amarks: Wizzup: you need the conf, master, init and enet.diff also
evil_core: amarks: try disabling s3tc
amarks: evil_core: any other names for s3tc?
evil_core: amarks: no, run driconf
er0x: anyone got running kms on r600 with .33?
MichaelLong: <-
er0x: mine kernel freezes at requesting radeon/R600_rlc.bin
er0x: i know i got it
amarks: evil_core: there are low/high textures, i don't see any reference to compression?
er0x: maybe i need to rename it to rv635_rlc.bin?
adamk_: er0x, No, you do not need to rename it.
er0x: adamk_: then maybe any ideas why kernel freezes at requesting it?
adamk_: Nope.
adamk_: Other than, perhaps, a bug.
er0x: got same with drm-radeon-testing
adamk_: Is radeon compiled into the kernel or as a module? If it's compiled as a module, are you loading it from an initrd?
er0x: radeon/drm is build-in
er0x: or maybe mine fw is broken
adamk_: Did you use the CONFIG_EXTRA_FIRMWARE option to compile the firmware into the kernel?
er0x: didint used
adamk_: Well that'd be a problem, then.
adamk_: You need to compile the firmware into the kernel using that opiotn.
adamk_: option, even.
er0x: adamk_: thx for help :)
adamk_: No problem.
idle_task: Hello again
er0x: now kms works :)
er0x: hmm 3d is faster than fglrx wtf?
spreeuw: with what?
er0x: kde effects
er0x: i can resize windows :d
spreeuw: waw, progress
spreeuw: desktop 2.0 is almost there now
maderat: i can't run kms with libdrm 2.4.17
spreeuw: when in doubt use git for everything
maderat: kernel 2.6.33-rc2 fc12 full update and drmrpm.src.fc13 rebuild for fc12
spreeuw: maderat: get the git libdrm
spreeuw: and build it in the correct order
spreeuw: I dont know about distro packages
idletask: This is why I love gentoo and its overlays
spreeuw: holy cow, theora slows my box to a crawl ;p
spreeuw: http://ccc.et-cip.rub.de:81/r1.html
spreeuw: but I love the browser integration otherwise ;p
spreeuw: might this be a radeon issue?
spreeuw: top shows 95% X cpu use
spreeuw: 5% seamonkey
knoppers: the website is hosted at my university, rub.de, cool :D
knoppers: here the stream is slow too
knoppers: high cpu
shadowmaster: (22
spreeuw: I can barely switch windows
shadowmaster: uhh :(
spreeuw: pon an AMD X2 5600
spreeuw: maybe an exa bug
Ghworg: I blame flash
spreeuw: its not flash
Ghworg: I blame flash anyway :-)
spreeuw: its a built in video decoder in mozilla engines html5 draft
edgecase: mozilla builtin sucks, get greasemonkey script to use totem plugin
edgecase: it tweaks youtube to use totem, uses hardware video scaling
spreeuw: np I have mplayer
edgecase: or that
spreeuw: http://streaming-26c3-ice.fem-net.de:3000/saal1 1 2 3
idletask: Hmmm
spreeuw: hehe mplayer works at 0.05 load
idletask: I run drm-radeon-testing kernel tree with up-to-date libdrm and mesa and KDE OpenGL screen savers are slloooow
spreeuw: oh 0.03
idletask: But then I know why :/
spreeuw: and still dropping
spreeuw: 0.01 hehe
Reilithion: My screen likes to jitter when the radeon module is loaded. It seems like a refresh rate might be off a bit.
spreeuw: Reilithion: on a flat panel?
Reilithion: yeah
spreeuw: without modelines?
spreeuw: let it autoprobe everything
Reilithion: spreeuw: No modelines.
Reilithion: My xorg.conf was mostly autogenerated.
spreeuw: maybe the devs can help then
spreeuw: or you make a modeline and tune it down a few decihz
Reilithion: Also, just this boot, there's a 1 1/2" margin of black at the bottom, and the top of the picture is not visible off the top of the screen.
idletask: Question... I have updated X, and since then, fonts are "not the same"... I know this is probably offtopic for this channel but... Why? (xorg 1.6.5-> 1.7.x with some x)
Reilithion: xrandr --output VGA-1 --auto fixes the margin problem. But it still jitters.
Ghworg: idletask: dpi set differently probably
idletask: Ghworg: that's strange, I didn't change anything from my xorg.conf (which, in fact, I don't even have - I let X detect everything)
idletask: Well, anyway
idletask: is happy to report that drm-radeon-testing, with KMS, yields to a more than playable nexuiz
idletask: I have trouble though if I try and configure video settings to more than "medium"
amarks: bloom seems to grind nexuiz to <1fps
amarks: and IIRC had to disable shaders for proper rendering for me
amarks: alienarena just crashed my box: failed to parse relocation!
amarks: :(
idletask: amarks: what is bloom?
amarks: idletask: no idea lol but it no workie
Ghworg: idletask: I'm just guessing, but the couple of times my fonts have changed on me that was the cause
idletask: Ghworg: well, if this means that the dpi setting has changed without even my having setting it in the first place... :/
Zajec: agd5f: could you define "engine is idle", please?
Zajec: agd5f: or maybe, i will just ask other question
Zajec: agd5f: if there are 10 commands sent to CP, and CP finishes processing one of them, does it emit GUI IDLE IRQ?
agd5f: Zajec: no
agd5f: the engien is still usy in that case
agd5f: engine means the entire GUI engine (CP, setup, shaders, etc.)
Zajec: agd5f: so when we want to upclock... we should lock CP mutex (to stop it receiving new commands), wait for GUI IDLE IRQ, reclock, unlock CP mutex
Zajec: is that right?
agd5f: it's basically the same as polling the GUI_IDLE bit in GRBM_STATUS reg, but the CP does it for you
Zajec: (same for downclock)
agd5f: Zajec: yeah, I think so
Zajec: agd5f: hm, ok, can try it... but not sure if that will work :| i'm afraid it may be something else causing my single frame corruptions
spreeuw: idletask: I can run it on normal in 720
spreeuw: but it needs an fps doubling for any serious competition I agree
spreeuw: especially slow is the rocket
spreeuw: and the dynamic light of that blue blob gun
agd5f: Zajec: yeah, power management sucks :(
spreeuw: but it all looks very impressive
amarks: Zajec: i took out the set_engine_clock calls as per your patch and when using dynpm=1, I can't get it to crash and is behaving like dynpm=0, so seems set_engine_clock is aggrevating the fence_wait crash
agd5f: amarks: makes sense. I don't think the engine likes being re-clocked while it's running
Zajec: amarks: thanks for testing!
Zajec: amarks: i want to believe agd5f :)
Zajec: i'm leaving for new year soon
Zajec: but will be back to PM in 2010 :)
agd5f: Zajec: have fun! it'll be here when you get back ;)
Zajec: :)
Zajec: don't do everything, please ;)
Zajec: leave sth for me :P
agd5f: Zajec: btw, I cleaned up my patches, no new functionality. http://www.botchco.com/alex/xorg/pm/
agd5f: Zajec: I won't have much more time for pm next year. I need to get evergreen working
Zajec: agd5f: ok, thanks for info
agd5f: at least initially
Zajec: oh, nice :)
Zajec: agd5f: could you eventually get back to blit on r600 Mesa driver? i know, it's kind of upsetting personal request... :)
Zajec: agd5f: but I'd love to see it to get some better Wine support :)
agd5f: Zajec: yeah. one of these days I'll get back to it
Zajec: agd5f: it's sounded quite easy last time as we have blit ready as for EXA
agd5f: shoudl help gnomeshell too I think
Zajec: agd5f: probably just hooking current code to Mesa's function
Zajec: great! :)
amarks: agd5f: can you clarify again please on tearing, i still see it with xv on mplayer in a window on r700 with drm-radeon-testing/kms/irq and kwin is set to use vblank, full screen is tear free, is kwin or mplayer broken or is it a video issue? i tried gl output with mplayer and this crashed the box :(
agd5f: amarks: depends on the vblanks support in the compositor
agd5f: amarks: lots of compositors unredirect fullscreen windows in which case it will use the Xv anti-tearing stuff
amarks: agd5f: it's only tearing non full screen so is that kwin's fault?
idletask: Hey
idletask: I use kwin also and have no problem with it
agd5f: amarks: if you have compositing enabled, then it's up to kwn
idletask: But I've had trouble, though (like too many people), with the "IRQ fimrware" :( (which I thank agd5f for providing)
amarks: agd5f: i turned on vblank in kwin, perhaps it's not working
idletask: agd5f: a question if you don't mind... Some 3D games, even native to Linux, turn off compositing from the window manager (I use kwin, and quake3 disables compositing), however some other 3D games (like nexuiz) don't - how is that?
idletask: Oh, and I use opengl rendering for desktop effects in kwin, if that's any help
agd5f: amarks: could be. I'm not too familiar with how kwin does things
agd5f: idletask: some compositers unredirect fullscreen windows
idletask: agd5f: isn't the fact that there are several ways to "do things" a problem somehow?
idletask: It makes for a lot of combinations
agd5f: idletask: several ways to do things?
idletask: Well, compositing vs 3d, since this seems, in my case, the matter at hand
idletask: Why does Quake3 turn it off and why doesn't Nexuiz do
agd5f: idletask: the 3D apps don't know about compositing
agd5f: the compositor does
idletask: agd5f: it just looks, however, that even though 3d apps don't know about compositing, they seem to have an effect on it
idletask: Most of them, at least
agd5f: idletask: DRI1 doesn't support redirected rendering
agd5f: you need dri2
agd5f: that's why all these compositer hacks exist to unredirect certain windows
idletask: Well, I have dri2 loaded, the Xorg log says so
chithead: pastebin Xorg.0.log
chithead: or glxinfo | grep -i render
agd5f: idletask: if you have kms and dri2 enabled, you should turn off the hacks in the compositer
evil_core: compositors sucks!
idletask: agd5f: describe "hacks"?
agd5f: idletask: unredirect fullscreen or 3d windows or whatever
evil_core: you gain only blink, battery drain and problems
idletask: Hmm
evil_core: and high memory usage
evil_core: so less RAM for system/games
evil_core: and you should use texture xv, not x iverlay
evil_core: over*
idletask: checks kwin settings
mentor: Yay
mentor: It works again
Shuren: hello, i need some feedback
idle_task: Shuren: on?
Shuren: today i tried out the new mesa 7.6.1, but with it, aiglx not work in radeon hd2400 (r610)... mesa 7.7 works ok
Shuren: the 7.6.1 changelog talk about r600 support... mistake is mine or other people see this behaviour?
idle_task: r600 support for aiglx? Explicitly mentioned?
Shuren: not for aiglx, for 3D, but with mesa from 11 december, it works well
Shuren: http://www.mesa3d.org/relnotes-7.6.1.html
Shuren: New features
Shuren: * Upgraded GL/glext.h to version 56, GL/glxext.h to version 25, GL/wglext.h to version 17
Shuren: * New 3D driver, r600, for Radeon R6xx, R7xx hardware
Shuren: and then, works well with development version 7.7
Shuren: seems to be a regression in 7.6.1, it's strange
idle_task: Well, I have only used mesa's git tree over the last few months
idle_task: But then, I don't know if, and don't think that, a "new 3d driver" means that this driver is complete
idle_task: All I know is that I have partial hardware 3D acceleration using libdrm/ddx/mesa git trees
adamk: Shuren: Define "not work"?
spreeuw: Shuren: the present git runs pretty stable
spreeuw: I havent had hardlocks yet
spreeuw: lately anyway
Shuren: Shuren: Define "not work"? <--- glxinfo says Yes to direct rendering, xorg.log says EE when loading AIGLX, and games are not playable, damned slow
adamk: Well games being slow has nothing to do with AIGLX.
adamk: Pastebin your /var/log/Xorg.0.log and the output of 'LIBGL_DEBUG=verbose glxinfo'
Shuren: now i have mesa 7.7 that works well
adamk: Most likely 7.6.1 would have worked, but you had some configuration or setup error.
adamk: But we'll never know now :-)
adamk: For future reference there's not much point in reporting problems if you aren't able/willing to help track them down. :-)
Shuren: well, upgradepkg and back :) wait for a while
adamk: No, I probably won't. I'm off in a few.
adamk: But someone else might be around to figure out what's going on.
Shuren: thank you for now, tomorrow i track down debug infos, good night
yao_ziyuan: i have a radeon 9200 card
yao_ziyuan: the radeon driver in opensuse 11.2/kubuntu 9.10/fedora 12 don't play OpenGL games like supertuxkart well
yao_ziyuan: should i use http://wiki.x.org/wiki/radeonBuildHowTo ?
bridgman: yao_ziyuan; the OpenSUSE radeon instructions recommend upgrading to mesa 7.7 for improved performance
bridgman: have you done that ?
bridgman: http://en.opensuse.org/Radeon
idletask: can confirm, since he runs Mesa's git tree, than Mesa >= 7.7 _does_ improve performance
bridgman: one click install down at the bottom
bridgman: for opensuse, of course
bridgman: is talking to himself again ;(
idletask: No you're not
bridgman: I know ;)
idletask: I'm reading this URL too, it may contain tips
bridgman: I mean yao_ziyuan seems to have vanished, squirted out of the universe like a watermelon seed
bridgman: ptooie !
idletask: My first plan was to buy and X1950, since this was a known supported card
idletask: But I couldn't find "less" than an rv790 (HD4890)
bridgman: until a month or two ago that was probably the best move
idletask: And to my great surprise, it worked out of the box!
bridgman: you have a 4890, don't you ?
idletask: Yes
bridgman: yeah, your timing was pretty good ;)
idletask: Which lspci reports as an rv790
idletask: Now, this IRQ firmware turns out to be a pain, really :(
bridgman: yeah, process changes in the open source world tend to be a bit bumpier than in the proprietary world
bridgman: no team of consultants
idletask: 2.6.31.x with x < 2 didn't need it, and it appears that future kernels won't be that kind anymore if you compile radeon dri built in :(
bridgman: but I would argue that's a good thing ;)
idletask: Maybe, but not when it introduces that much of a regression
idletask: Take over your config from 2.6.32.x to later versions and 3D stops working
idletask: Not good
bridgman: I'm not sure what the plan is - I thought the bin files were going to end up in /lib/firmware and be redistributed that way, but maybe that's still considered "part of the kernel"
bridgman: in fairness, that's "take your kernel from 2.6.32.x to a bleeding edge development version and..."
bridgman: released versions don't do that
idletask: bridgman: while this may not turn out to be a problem from distributions ultimately, it is for individuals like me who want a customized kernel
bridgman: I think you can blame Christmas for this one
idletask: And even distributors will have a hard time to cope with this change, I gather
bridgman: that seems highly unlikely; they're already doing initrd's for other hw vendors
idletask: That's the problem - initrds
bridgman: unlikely anyways, maybe not highly unlikely ;)
bridgman: that's the solution
idletask: They hinder performance
bridgman: what is the problem ?
bridgman: how do they hinder performance ? someone mentioned that earlier but never explained why
bridgman: it's a ramdisk
idletask: Well, you need to load them into memory to access them
idletask: And then you need to switch roots
bridgman: ok, milliseconds out of your life never to be regained, next ?
idletask: Without an initrd, I have a full fledged Gentoo booting in less than 7 seconds from grub to xdm
idletask: With an initrd in the way, add 5 seconds
yangman: pft. booting ;)
bridgman: 5 seconds is higher than I would expect
bridgman: but I agree people are funny about boot time
idletask: bridgman: not without reason
bridgman: never quite understood why though
idletask: bridgman: think mobile devices
idletask: bridgman: which are a huge market for Linux in general
idletask: bridgman: you wouldn't want your PDA to take 30 seconds to boot, would you?
bridgman: my blackberry takes more than that today
cxo: Cold boot, from power button to gnome desktop, 14s, Ubuntu-9.10, 3.2Ghz AMDx2
bridgman: almost a minute
idletask: bridgman: right now, my mobile phone takes longer to start than my desktop PC. What the... :p
bridgman: resumes in about a second though
bridgman: isn't a big believer in these newfangled phones ;)
yangman: so, how's the evergreen stuff going? weird corner cases everywhere as usual, or unusually smooth? ;)
idletask: Evergreen is only ever a name for me, and which seems to be another ATI/AMD chipset... What is is exactly?
chithead: evergreen is r800
chithead: code for it supposedly exists but is in legal review
idletask: chithead: does that refer to the next-gen radeon chipset? Is it on sale already?
yangman: idletask: the 5xxx line of RadeonHDs. they've been on sale for a couple months now
chithead: yes, the radeon 5850/5870 are on sale since late september
idletask: OK
idletask: I haven't seen any reference to them right now
chithead: evergreen are not properly supported in linux yet, not even by the proprietary driver
bridgman: search on initrd delay
bridgman: chithead; what's the issue with the proprietary driver ?
chithead: for example, no eyefinity support
bridgman: didn't think we even planned to release that yet.. (?)
bridgman: AFAIK it'll hit windows & linux at the same time
chithead: bridgman: http://www.amd.com/us/products/technologies/eyefinity/Pages/eyefinity.aspx it works on windows, footnote 4 says that plans exist for linux support
chithead: with catalyst 9.12 "hotfix" it is supposed to work in crossfire even (windows only)
bridgman: wishes that Santa had brought high speed internet
chithead: interestingly amd demoed eyefinity on linux before evergreen release. but the drivers they used seem not to be available to the public
idletask: bridgman: you don't say
idletask: chithead: I'd like to bother you with a question if you don't mind
bridgman: chithead; you're right, didn't realize we had released it on Windows
idletask: Hey!
idletask: Surprise surprise
idletask: I _can_ wine Dungeon Siege in a window now
chithead: I hope there will be a midrange card with 6 dp outputs too. I might buy one as soon as it is supported by open source drivers
idletask: It's unbearably slow but I can do so
idletask: There has been progress!
idletask: This means I can try Oblivion in a window as well
idletask: It does remain unplayable but this is more than I could achieve just two days ago
cxo: i wonder if alienarena works now
cxo: it was too slow before
idletask: nexuiz has never had a problem here
idletask: Neither did quake3
idletask: (except that the latter turned off composition)
cxo: neuiz and alienarena are too slow on my HD4870
idletask: Even nexuiz??
idletask: nexuiz just flies here
chithead: nexuiz works ok on my rs780
idletask: So does it here (rv790)
chithead: I have some problems with spring still, because the driver lacks s3tc support
idletask: chithead: that's vocabulary which I just don't have - "sprint still"? "s3tc"?
chithead: "spring" is a 3d rts engine
idletask: And... err... rts? :p
cxo: idletask, i get 12fps with compiz and 25fps without. maxed out settings, 1280x1204
cxo: ^1024
chithead: rts=real time strategy. spring works ok on r500 with latest drm http://www.phoronix.com/forums/showthread.php?t=21150 but not r600 yet
idletask: chithead: any game that I might try out?
idletask: (using that "spring" engine)
chithead: idletask: most popular are balanced annihilation and complete annihilation mods. visit http://springrts.com/ for installation instructions
chithead: developers seem to pay most attention to nexuiz and openarena, so these are the most likely games to work
chithead: supertuxkart and xmoto are also worth a try
idletask: chithead: those are great tips
idletask: While, ultimately, I wish all my Windows games to work flawlessly under wine, I'm also looking for "pure" linux testbeds for 3d stuff
idletask: Nexuiz is a compelling example
idletask: But gallium3d holds great promise for the future (may wine implement d3d over it, rather sooner than later)
chithead: I think the holy grail is etqw but it will still be some time until that works properly
idletask: chithead: for me, the holy grail would be a program that says: this and that feature is not implemented/software implemented/hardware implemented
idletask: Even by the detail
idletask: But that would be some tool for developers at least
idletask: Well, the holy grail given the general state of 3d right now - ie, "mostly there, but there is this and that left to do"
idletask: I have run the Mesa demos (I could even get one to hang the display but not the kernel)
idletask: But no Mesa demo exists that says, "this is hw/sw/no"
idletask: And, from what I could understand, Mesa is the best place to have such a "tell it all" program
Demo1: hello
Demo1: need help
BioTube: Demo1: state your issue
Demo1: My Xorg.0.log on ubuntu http://pastebin.com/d798c5f42, on gentoo http://pastebin.com/d663a7604 On ubuntu visuals runs fine, on gentoo not so good, maybe there are solution?
BioTube: define "not so good"
Demo1: delaying
Demo1: for example compiz
BioTube: card?
Demo1: Radeon 9550
BioTube: all I can suggest is updating libdrm, mesa and ddx
Demo1: how to update?
BioTube: I don't know gentoo
Demo1: mes ive updated
amarks: Demo1: use the x11 overlay for -9999 ddx/mesa/libdrm
Demo1: how?
Demo1: amarks: ?
idletask: Demo1: fill in /etc/portage/package.unmask
Demo1: idletask: and thats all?
Demo1: idletask: ?
soreau: You would need to add the x11 overlay of course
amarks: Demo1: did you add the x11 overlay with layman?
Demo1: amarks: nop
Demo1: How to?
soreau: Demo1: Adding gentoo specific overlays is a bit beyond the scope of this channel. Perhaps try one of your distro support channels
Demo1: soreau: i think someone can help
Demo1: soreau: here too
erilliam: I've noticed my graphics card (HD 4350) getting quite hot (uncomfortable to touch in fact). Is there anyway I can check the temperature? And is the opensource driver (in fedora 12) doing any sort of power savings? Would I be better to be using fglrx?
erilliam: I'm having problems where it seems the graphics shut off, but the computer still keeps working. (All monitors will just say 'No signal' -- but I can still ping / connect to the machine )
Demo1: erilliam: i think temp you can check if u have some temp indicators on graphical adapter
Demo1: erilliam: and then with some desklet
taiu1: agd5f: there seems to be some additional alignment requirements for cubemap mipmaps
taiu1: agd5f: could you confirm this for r600 ?