bloopletech: Quick question - I have a Redwood chipset; if I install the open-source drivers, I won't have 3d acceleration, but I'll still be able to play videos etc. right?
maligor: you won't have 2d acceleration either afaik
Ke: redwood is some r800?
maligor: midrange r800 I guess
maligor: or entry/midrange
Ke: what r800 would desperately need is powersave, not hw acceleration
Ke: at least on my card the noise and power usage are pathological
airlied: I think evergreen powersave is already in d-r-t
Ke: yay
AstralStorm: for me, power saving is a secondary concern
AstralStorm: more important is that missing GLSL 1.2 :/
Ke: well you won't have pleasant multimedia experience unless you really crank up sound volume
bloopletech: My current problems are that I get horizontal tearing on fast moving scenes using fglrx, something ATI knows about apparently but hasn't fixed yet
AstralStorm: Ke: get good earphones/headphones
AstralStorm: ;p
Ke: =o)
bloopletech: I should have been more specific - it's a Mobile card, the "ATI Mobility Radeon HD 5650"
Ke: bloopletech: try #ati
bloopletech: So noise isn't a concern - at least AFAIK
bloopletech: Ke: thanks
Ke: (though probably won't be able to help you anyways)
AstralStorm: Ke: actually, I'm living with a swscaler right now
AstralStorm: because whatever the scaler implemented in gl and xv etc. is not good enough
AstralStorm: lanczos 3 or mplayer's cubic is fine
AstralStorm: I've been yammering about it before
bloopletech: Ok, found some decent info on wikipedia - looks like 'Redwood' is in the Evergeen family
maligor: bloopletech, the lack of vsync in video has been in fglrx ever since they added vide playback ;P
bloopletech: maligor ;(
bloopletech: If I want to try the open source drivers, is it worth running of git HEAD if I'm using Evergreen?
Ke: you might want drm-next oslt
Ke: for kernel
bloopletech: Sorry? Still a bit new to this area of dev
Ke: the driver has 4 parts, drm in kernel, libdrm in userspace, xf86-video-ati and mesa
bloopletech: ah
bloopletech: Starting to make sense now
bloopletech: OK well I might try getting this to work
Ke: though setting up the kernel is non-trivial
bloopletech: Ke: I have experience compiling and installing stuff, but not a lot around the kernel
bloopletech: Is there a wiki page or anything on status of evergreen support?
bloopletech: I see WIP for a lot of things on the RadeonFeature page, but I was wondering if there was anything more specifc
papillon81: there is obviously a bug in mesa concerning GLX_USE_GL. it does not accept it anymore and thus OSG is no longer working
airlied: bloopletech: evergreen has no accel, modesetting works, not much else to report
bloopletech: airlied, ok, looks like there is some work on xv
airlied: nope not without accel
bloopletech: aw ok
papillon81: airlied: should i report this mesa bug somewhere else?
airlied: papillon81: what do you mean it doesn't accept it?
airlied: glxinfo uses that so I can't believe its broken
papillon81: airlied: i bisected mesa to find the commit that caused it: http://cgit.freedesktop.org/mesa/mesa/commit/?id=ad7d36e91933a49ce3f6ed2c14ae585a8a84fdf5
papillon81: airlied: i get Error: Not able to create requested visual.
papillon81: GraphicsWindow has not been created successfully.
airlied: papillon81: oh can you try a patch?
airlied: http://fpaste.org/5hGL/raw
papillon81: airlied: sure. will just need to rebuild OSG in it's vanilla form
airlied: papillon81: I think that patch is correct, I'll post it to the list anyways
papillon81: airlied: working here :)
papillon81: airlied: where exactly will you post the patch?
airlied: papillon81: mesa-dev
papillon81: airlied: thx, i found it. and thanks for the help and quick fix
hiveNzin0: hi, does someone have an ATI Radeon HD 5850 here ?
Ke: 5870 close enough?
hiveNzin0: I don't know, it's just to know if the open source driver works with HD 5xx0
Ke: nope
hiveNzin0: okay so It's not my "fault" that doesn't work :p
Ke: well you'll get kms, but no hw accel
hiveNzin0: I thought I had done something wrong in the kernel config or others
hiveNzin0: I use the closed one for now, I'm gonna wait my graphic card is supported, thank you =)
hiveNzin0`aw: is away
icewaterman: hi, i just noticed, that my radeon hd2600 is still slower than the hd1600
icewaterman: so r6xx is not fully complete, right?
glisse: far from it
okias: MostAwesomeDude: Hi, glxgears not sefault anymore, but show only blue screen, is that expected? (I guess segfault was fixed by Marek Olsak in lastest commits)
strange: hey guys after reinstall from the box from jaunty to lucid (ubuntu) with a ati 9200 Pro performance has gone, i can no longer see the xbmc gui full screen
glisse: agd5f: on r6xx-r7xx it seems we need to write at least one param for vs shader, right ?
evil_core: anyone saw zombie colors in ut2k4 in KMS classic?
evil_core: I didnt had them before
glisse: agd5f: nevermind it's about texture & addr needing to start on 16byte aligned
evil_core: and performance has dropped signicifantly in KMS and in UMS a bit too
Zaba: hm.. is there any information somewhere about what is done and what is still lacking in the r600 driver?
[Enrico]: Zaba: there is this, not very detailed may be http://www.x.org/wiki/RadeonFeature
Zaba: hm.. not detailed, indeed
chithead: there is also a page somwhere which shows the completeness of the opengl implementation, but I cannot find it right now
Zaba: I also wonder about gallium3d and r600
Zaba: it seems to be 'WIP' according to http://www.x.org/wiki/GalliumStatus
Zaba: but what I wonder about is, what kind of improvement can one expect once/if r600 is ported to it?
[Enrico]: Zaba: well better performance, more features
Zaba: sounds promising, indeed; is there some git branch for it?
BioTube: cgit.freedesktop.org/~glisse/mesa
Zaba: how well does it work at the moment?
BioTube: haven't tried it an a while, but last time it froze my system
Zaba: okay. Well... I periodically manage to freeze my system with the current r600 drivers from git, too.
Zaba: that is, soft-lockup the GPU
Zaba: but that's probably more related to the kernel
rmrfslash: is 3D hardware accel working in the latest radeon driver?
Zaba: works for me on a radeon hd 4200
_ds_: Works for me on an X300, but then it's never really not worked :-)
Zaba: quake3/UT99/nexuiz run quite smoothly, well, the latter not as much, but still
Zaba: tried springrts the other day, but it segfaults inside r600_dri.so when luashaders are enabled in it, and works but the map lacks textures otherwise, and 3d trees lag and look awful.
Zaba: but it's comparable to a geforce4 ti 4200, which my old system used to have, which I used with nvidia's drivers
Zaba: although I guess this card is supposed to perform better..
_ds_: Zaba: the answer to rmrfslash's question is "yes", given that there's no limitation to specific hw :-)
Zaba: well.. right..
Zaba: anyway, speaking of said segfault in r600_dri.so. I wonder, how would I go about getting a meaningful backtrace?
Zaba: I have the debugging symbols for the driver split, but even when I load them with symbol-file, it doesn't find any...
twnqx: glisse, around for a small question?
Zaba_: hrm
twnqx: would you expect the patch "drm/ttm: ttm_fault callback to allow driver to handle bo placement V6" to leave a usable kernel on its own?
twnqx: or would all the ttm patches from the merge be needed at once?
glisse: twnqx: it shouldn't break driver but it wouldn't be use without the other
glisse: but i didn't well tested this
twnqx: glisse: http://img688.imageshack.us/img688/663/img0348ul.jpg
twnqx: that's what happens on suspend for me with only that patch
glisse: works with the other ?
twnqx: suspends, but doesn't resume as per https://bugs.freedesktop.org/show_bug.cgi?id=27822
twnqx: well, resumes, but X is not usable any more, i can SSH in and reboot though :P
glisse: i will try to see if i can reproduce
glisse: but i fear without the hw i can't do much
twnqx: the patch itself looked rather small... gonna see if i can break it down further to see what exactly causes the bug
twnqx: just wanted to know if all the patches should work on their own or if they are too tightly coupled (e.g. the callback function having an int return value instead of void)
twnqx: time for some coffee.
vsrinivas: is it okay to point my crtc base address at a buffer located in system memory (via the gart)?
vsrinivas: is there not enough bandwidth from system memory for that to be reasonable?
glisse: vsrinivas: should work but highly unrecommanded
vsrinivas: oaky.
glisse: vsrinivas: may i ask what are you trying to do ?
vsrinivas: the reason i'm asking all these perverse questions - on plan9 some users of the radeon driver are complaining about pretty bad performance with our radeon driver
vsrinivas: someone traced it to the ton of framebuffer readbacks we do.
vsrinivas: i was wondering if there were things we could do to speedup readback, rather than trying to make devdraw not readback so much...
vsrinivas: (is why i was asking about the same game DownloadFromScreen does yesterday) :)
vsrinivas: though that would almost certainly be the better answer...
glisse: vsrinivas: my feeling is that all pixmap should in cacheable memory
glisse: and that we should keep a copy of front buffer in cacheable memory
glisse: issue is that today keeping a copy of front is wasting a lot of memory
glisse: also on AGP this would need more code to be able to use the PCI gart along the AGP one
glisse: all this is needed because X is not easy to accelerate
zigovr: hi all
raggiskula: Hi
raggiskula: question: I just upgraded to ubuntu 10.04 and after that my sceen drivers went nuts
vsrinivas: oh, hmm; does X readback often enough that a copy of the frontbuffer would be worth it?
vsrinivas: (are there any other reasons for a whole copy?)
zigovr: I have a HP pavilion with a HD3200 gfx card, it's working pretty well with the ati radeon driver, but for one point : I noticed it is heating quite a lot (it heats less with the proprietary driver, by about 10C)
raggiskula: hardware: T60 thinkpad laptop. Radeon X1400
zigovr: apart of that, video (with no tearing) , suspend, compositing, all is working impressively well
zigovr: any idea if the radeonhd driver might be appropriate for my model ?
raggiskula: the main display is ok, but the extended one is quire disorted
raggiskula: here's a photo of it
raggiskula: http://i.imgur.com/edtfr.jpg
mjg59: zigovr: Power management is currently being worked on
mjg59: radeonhd is pretty much unmaintained at this point. Most development is in radeon.
zigovr: ok thanks mjg59 , so I will be patient, temperature is still in acceptable ranges
zigovr: is there already something testable btw ?
glisse: vsrinivas: application like firefox keep reading back front buffer
glisse: especialy when scrolling
vsrinivas: oh joy
vsrinivas: do any of the other drivers keep a full copy?
vsrinivas: how would you update the copy when you have the card do fills over regions / etc?
otaylor: glisse: are you sure ffox does readback on the *front* buffer
otaylor: glisse: I think it does readbacks on pixmaps in some cases, but can't see when it would read the front buffer
_ds_: Fun… with freshly-built mesa (9d336056), just had oolite spewing lots of ints. I fiddled with its shaders setting (setting to full, then back to simple), it basically stopped responding, so I killed it (hmm, should have attached gdb), and now I'm getting a segfault. Backtrace at http://paste.debian.net/71644/
glisse: otaylor: last i looked it did that a lot when scrolling
glisse: or maybe it was pixmap
glisse: otaylor: btw i can't run gnome-shell on f13
glisse: never mind today update fixed it
glisse: i will look tomorrow at sync issue
glisse: otaylor: do you have a test case for that ?
otaylor: glisse: well, the only sync issue left on radeon is that clutter is falling back to opening /dev/dri/card0 directly...
otaylor: glisse: so, to do better, you need to implement GetMSC and ScheduleSwap in the DDX, then test that with anything that uses SGI_swap_control and see if it's synced to vblank
glisse: otaylor: ok so gnome-shell should work properly otherwise right ?
glisse: i will try to spend time on implementing those this week, i need to look at intel and at the extension to see what's this is doing
otaylor: glisse: As far as I know.
otaylor: glisse: Haven't been doing too much testing on my r300 AGP laptop since I'm still getting system crashes there, but it seemed to be working well with the mesa/clutter/xserver I built on friday on a r500 system
rnoland: adamk: ping
rnoland: were there perfomance issues with mesa 7.6.1 that i missed...
rnoland: agd5f: can you confirm that bit 0 in the pcie gart is to disable snooping?
AndrewR: mesa master not buildable for me .... (up to commit 559d124ed3f2070c73c5cb9c3b506ddd9cfaeb02)
AndrewR: gmake[2]: *** [main/api_exec_es1.o] Error 1 and before this - tons of lines .... more than my terminal set to show at once.
agd5f: rnoland: yes. bit 0 is disable snoop, 1 is relaxed ordering, and 2 is write enable, 3 is read enable
agd5f: rnoland: same on rs4xx/rs690/rs740 gart tables
AndrewR: even if i disabled gles1 and gles2 ... (make distclean && ./autogen.sh + my_long_config_line)
AndrewR: agd5f, hello
agd5f: AndrewR: hi
AndrewR: for some reason my desktop don't want to crash with debug kernel .... so, i tried to look why my rv280 fail fbo_firecube, updated mesa and ...lazily listen to music. BTW, xmms + visualizer can make nightmare for anyone who want to keep cached copy of frontbuffer. xmms's main window constantly update itself, up to 60 fps ....guess any visualisation will make process even more intensive.
rnoland: agd5f: hrm, ok i swear i had disabled snooping before and seen a small improvement with write-combining....
rnoland: now i was seeing lagging....
evil_core: Mesa master is broken due to ES changes
_ds_: Hmm. Just noticed lots of "[drm:r100_cs_track_texture_check] *ERROR* Texture of unit 0 needs 16777216 bytes but is 4096" in the kernel log.
_ds_: http://paste.debian.net/71667/
_ds_: (same build as I reported on earlier; kernel 2.6.33.2)
evil_core: I got it in the past too
evil_core: _ds_: does it happen in UMS
evil_core: ?
_ds_: Don't know. I don't really want to reboot this box right now :-)
_ds_: (plus there's fiddling with renaming files in /usr/lib/dri…)
adamk: rnoland: Pong.
adamk: rnoland: Not sure if that was directed at me, but I'm not aware of any...
adamk: Obviously there have been serious improvements since then.
Ronis_BR: did anyone fail to compile mesa git these days?
mattmatteh: using kernel 2.6.33.2 with kernel mode setting and i am missing /sys/class/backlight/radeonbl0
mattmatteh: not sure if here is the right place to ask. and i am not sure what provides that, i assume the old radeonfb driver did
agd5f: mattmatteh: kms needs to be wired up to support it on mac cards since it's not provided by the firmware/acpi
mattmatteh: agd5f, ohh
mattmatteh: agd5f, /sys/class/backlight/radeonbl0 is only on mac ?
agd5f: mattmatteh: MrCooper had a patch for it a while back. It should be in the dri-devel archives
agd5f: mattmatteh: yes
mattmatteh: ahh, i thought that was a path used by all
mattmatteh: agd5f, that patch would be applied to the kernel ?
agd5f: yes
mattmatteh: agd5f, i just googled and found some irc logs, where would i look for MrCooper patch ?
agd5f: mattmatteh: dri-devel archives. marc.info for example.
mattmatteh: ok
mattmatteh: googles again
agd5f: mattmatteh: http://marc.info/?l=dri-devel&m=126026164329261&w=2
rnoland: adamk: are you running -CURRENT?
Loplin: Hello. I've updated to xorg server 1.7.6 from 1.6.5 and my multi monitor config stopped working. I've started redoing it, but I'm not sure what is wrong. It will now mirror properly, but it won't put them side by side until I manually do `xrandr --output VGA-0 --right-of DVI-0`, in spite of having `Option 'RightOf" "DVI-0"` in my xorg.conf. I just started using radeon as opposed to radeonhd, and recall having to use RRoutput
Loplin: My xorg.conf is at http://pastey.net/135977
agd5f: Loplin: you need to set your monitor identifier names to the names of the output, e.g., DVI-0 rather than Monitor-0
adamk: rnoland: I am, yes.
adamk: From a few weeks ago.
rnoland: adamk: hrm, ok do you have r207069?
agd5f: Loplin: and VGA-0 rather than Monitor-1
Loplin: agd5f: so, in the device section, Option "DVI-0" "Monitor0" ?
adamk: In the Monitor section, give them those Identifiers.
adamk: You can map them in the device section, if you'd rather, but that's just extra lines that you don't need if you just name the monitors DVI-0 and DVI-1 in the Monitor section.
agd5f: Loplin: Option "Monitor-DVI-0" "Monitor-0"
adamk: Meh.
Loplin: Ah
agd5f: Loplin: also, the Option "RightOf" takes a montior name
agd5f: so that needs to match the monitor identifier
Loplin: agd5f: That's what I thought, but man xorg.conf says otherwise, and so did someone in #xorg
Loplin: I'll just name everything related to the first monitor DVI-0, and the second monitor VGA-0, and it should get it whether it is an output or a monitor
agd5f: yes
agd5f: Loplin: the xorg.conf man page is correct. at least as far as orientation is concerned:
agd5f: Option "LeftOf" "monitor"
agd5f: This optional entry specifies that the monitor should be positioned to the left of the
agd5f: monitor of the given name. (RandR 1.2-supporting drivers only)
Loplin: agd5f: heh, our man pages are different then: Option "LeftOf" "output"
Loplin: This optional entry specifies that the monitor should be posi‐
Loplin: tioned to the left of the output (not monitor) of the given
Loplin: name. (RandR 1.2-supporting drivers only)
adamk_: agd5f: Is there some advantage to doing that mapping in the Device section rather than just naming the monitors DVI-0 or DVI-1 in the MOnitor section?
adamk_: rnoland: I'm using csup, is there some way to find out what revision?
agd5f: adamk_: no. it's just for compat with old xorg.conf monitor entries
rnoland: adamk_: not an easy way really.... i bet you don't have that....
adamk_: Well all I can really tell you is that it's from April 8th... Older than I originally thought.
adamk_: Now I need to find out why my server here won't boot.
adamk_: bbl.
rnoland: agd5f: should it be safe to set the mmio to write-combine? or does it have the be uc?
agd5f: rnoland: you don't want the reg aperture wc
rnoland: agd5f: ok, i have it uc now... but it looks like what was happening with snooping off, was that it thought we were behind.... and so it spent a lot of time in wait_for_idle.
rnoland: re-enabled snooping and that issue goes away and it's all smooth and fast again...
rnoland: all of the pages backing the gart are WC... so i'm a bit unsure why the reg access seems to lag behind...
Loplin: Okay, so I now have in Identifier "DVI-0" in my first monitor section, Option "RightOf" "DVI-0" in my second Monitor section and in my device Section I have tried, various versions of Option "Monitor-DVI-0" "DVI-0", including Monitor0-DVI-0, Monitor-0-DVI-0, Just DVI-0, and omitting the option all together. Most of that just turns my second monitor off until I manually adjust with xrandr.
Loplin: Am I still missing something obvious? http://pastey.net/135984 is my current xorg.conf
adamk_: Loplin: Not sure if it would cause problems to have it, but you can remove line 131.
ezzieyguywuf: are there any know problems with radeon and WoW in wine?
ezzieyguywuf: the game crashes for me as soon as I enter the game world, and I've been told in #winehq that this is probably a video driver problem
[Enrico]: ezzieyguywuf: radeon driver.... which card ?
ezzieyguywuf: [Enrico]: Ati fireGL V5200
evil_core: ezzieyguywuf: read topic there
evil_core: ezzieyguywuf: what distreo does you got?
ezzieyguywuf: evil_core: gentoo. read the topic where?
[Enrico]: ezzieyguywuf: mhm iirc wow requires the GLSL so you need to use the gallum3d radeon driver, not the classic one
ezzieyguywuf: gallum3d radeon, hm I'll see if this is available in my package management system
hnsr: doesnt classic also support GLSL?
evil_core: ezzieyguywuf: no, read links in channel topic
ezzieyguywuf: evil_core: ok will do
evil_core: hnsr: not for r500
hnsr: oh
evil_core: anyway UMS rulez for r500 and MS really sucks there
evil_core: hnsr: and dont replace your current drivers/libs, you can have both installed and choose at runtime*when running application)
evil_core: s/hnsr/ezzieyguywuf/
ezzieyguywuf: evil_core: wait, what?
ezzieyguywuf: the link you mentioned says to remove fglrx
evil_core: ezzieyguywuf: its free radeon channel, fglrx is #ati
evil_core: ezzieyguywuf: and fglrx doesnt support your card anytime longer
ezzieyguywuf: evil_core: I'm saying, that could count as my "current drivers/libs"
[Enrico]: ezzieyguywuf: in gentoo you have to build mesa with gallium USE flag
evil_core: ezzieyguywuf: you know whats DRM, DDX and Mesa?
ezzieyguywuf: [Enrico]: excellent, I will do that thanks
[Enrico]: ezzieyguywuf: of course you alse need KMS so better if you use the .33 kernel
ezzieyguywuf: evil_core: no, I don't
evil_core: .33 sucks, d-r-t is better
ezzieyguywuf: [Enrico]: I JuST upgraded to .32
evil_core: so bad
[Enrico]: ezzieyguywuf: oh well it might be good the .32 :)
evil_core: .32 ad many problems with suspending and resuming for me
evil_core: ezzieyguywuf: I guess its T60p?
ezzieyguywuf: ok, there is still fglrx.ko in /lib/modules/
ezzieyguywuf: will this pose a problem?
evil_core: shouldnt
evil_core: the o nly problrm could be libGL and libglx probably
evil_core: anybody using new xserver and got half-life based game?
[Enrico]: ezzieyguywuf: if fglrx will not be loaded no
evil_core: KMS is good in theory, UMS in practice ;)
mattmatteh: agd5f, do you know what kernel version that was for ? 2 hunks failed http://dpaste.com/190142/ and the rej http://dpaste.com/190144
mattmatteh: agd5f, i just tried .34-rc6 and got 2 of 7 hunks failed
agd5f: mattmatteh: whatever was current at that time
mattmatteh: :P
mattmatteh: ok, thanks
soreau: The static-when-3D-happens-on-right-part-on-screen-with-rv350 is in lucid kernel too :P
evil_core: soreau: some interesting patches for r500?
soreau: evil_core: huh?
evil_core: soreau: was thinking you are talkign about some patches
evil_core: am Iloosing performance because using radeonhd?
evil_core: 3D performance
muffin2: I want to run 3d games with x1250
muffin2: what can I do?
muffin2: I want to do it on ubuntu
agd5f: evil_core: 3D has nothing to do with the ddx. it's all handled by mesa
evil_core: q3a runs perfectly in UMS
evil_core: but OpenArena is very laggy
evil_core: both in UXGA
evil_core: I wonder if its not sound related problem
muffin2: hi
muffin2: After removing fgxrl, radeon can run 3d games
muffin2: But it's slow as hell
evil_core: libGL: OpenDriver: trying /opt/xorg/lib/dri/tls/r300_dri.so
evil_core: libGL: OpenDriver: trying /opt/xorg/lib/dri/r300_dri.so
evil_core: libGL error: failed to open drm device: Permission denied
evil_core: muffin2: use UMS
evil_core: wats the problem in mine case?
muffin2: evil_core : what's ums/
evil_core: I got DRI only from root in KMS :/
muffin2: evil_core : Googling ums displays lots of many things altogether.
muffin2: evil_core : you need to tell me what it is
evil_core: whats card?
evil_core: glxinfo |grep render
evil_core: I was thinking you know what you are doing
muffin2: X1250
evil_core: but if not, maybe you dont have accel at all ;)
evil_core: glxinfo | grep render
muffin2: direct rendering : Yes
evil_core: should be more
muffin2: OpenGL renderer string : Mesa DRI R300 (RS600 7942) 20090101 x86/MMX/SSE2 NO-TCL
evil_core: like renderer
muffin2: It took some time to type
muffin2: That's all
evil_core: its AGP card?
muffin2: It's X1250
evil_core: but AGP or PCI-E?
evil_core: you got very old system/drivers
muffin2: It's X1250 on laptop. but I don't know
muffin2: X1250 on LG E300 laptop
glisse: it's neither it's IGP
muffin2: I don't know agp or pci-e
muffin2: evil_core : how can I find it out?
glisse: muffin2: it's neither AGP or PCIE
evil_core: glisse: but IGP also uses AGP or PCI-E, amd I wrong?
glisse: well it's more like PCIE
muffin2: should I pastebin /var/log/Xorg.0.log?
glisse: IGP can directly use system memory
evil_core: so I dont have PCI-E, but IGP in T60p?
glisse: but for GART stuff they are using PCIE i think, thought maybe they optimize the memory controller
glisse: evil_core: no
glisse: IGP are RS*** chipset
evil_core: IGP dont have their own memory?
glisse: on laptop you either have IGP or dedicated GPU
evil_core: or only for FB?
glisse: some IGP can have their decicated memory
glisse: but most of them just steal system memory
muffin2: Mine shares system memory
muffin2: But it's got a GPU
evil_core: mine got 256MB + 256MB TurboMemory and shows 512MB under windows
evil_core: can I use it also under linux?
muffin2: What is UMS?
muffin2: I should buy a laptop with lots of graphic card memory.
evil_core: muffin2: thats what you are using right now
glisse: i don't know what is behind the marketing name of turbomemory
muffin2: evil_core : user mode setting?
evil_core: yes
muffin2: I asked what it's an acronym for.
evil_core: muffin2: I was thining you want explanation how graphic stack looks under linux ;)
evil_core: you got really old system, if you want performance use fglrx, or d-r-t, etc, like in link
muffin2: evil_core : by default, radeon driver is very slow.
evil_core: but you should probably upgrade to sometrhing more recent at 1st
muffin2: evil-core : my system is not very old
muffin2: I bought it on 2008 aug.
muffin2: It's LG E300
evil_core: it is, its not suported by AMD any longer
muffin2: man
muffin2: I should give up
evil_core: no
muffin2: and do something else useful
evil_core: I got identical card
muffin2: what?
evil_core: no
muffin2: evil_core : what do you suggest I do?
evil_core: I was talking with r500 owner some time ago ;)
evil_core: muffin2: are you experienced with linux?
muffin2: Should I remove xorg.conf and put xrandr resolution commands in gdm startup file?
evil_core: you got old istro
muffin2: evil_core : I have ubuntu 9.10
muffin2: And I am a little experienced.
evil_core: no, it doesnt matter for performance
muffin2: I can modify xorg.conf
evil_core: 9.10 is latest?
muffin2: no
muffin2: 10.04 is latest
muffin2: Ubuntu lucid sucks. It doesn't work on my laptop
muffin2: 10.04 == lucid
evil_core: muffin2: if you need perofmrance on old OS, then fglrx is the way
muffin2: evil_core : tell me more about how to fix this
muffin2: fglrx doesn't work
muffin2: Should I remove radeon to make fglrx work?
evil_core: what means lucid doesnt work on your laptop?
muffin2: evil_core : after I enter "installation" menu, it displays a blank screen.
evil_core: dunno if 9.10 fglrx supports your card
muffin2: fglrx doesn't.
muffin2: Or
muffin2: fglrx and radeon can't live together.
evil_core: you can install bleeding edge DRM, Mesa and DDX
evil_core: but I dont know your card at all
muffin2: evil_core : how did you make it work?
evil_core: you got instructions in topic, dunno if its installable under so old system like your
muffin2: not lucid
evil_core: I am using bleeding edge free radeon driver, not fglrx
muffin2: evil_core : do you use radeon?
Janhouse: needs reboot
evil_core: yes
evil_core: but not from repositoroeis
muffin2: evil_core : where is it from?
evil_core: http://wiki.x.org/wiki/radeonBuildHowTo
evil_core: but I dont know your card, dunn owhat you want to achieve
evil_core: dunno how well its supported, I am nto best person to talk about RS cards
evil_core: I know only R200 and R500
muffin2: It's r300 as far as I know.
evil_core: glisse: its R300 what he got?
evil_core: muffin2: I can help you install it, but cannot promise that you will gain performance significantly
evil_core: muffin2: if you want real 3D performance, than you must use old distro with fglrx or windows
evil_core: muffin2: what is too slow in your opinion?
muffin2: glest is too slow
muffin2: It's a 3d game that would run smoothly on windows.
muffin2: Forget it
evil_core: muffin2: you will not acheive windows performance using free radeon driver
muffin2: kernel version is lower than 2.6.32
Loplin: Okay, after trying some more options, I noticed the error "(EE) RADEON(0): Cannot position output VGA-0 relative to output DVI-0 without modes" in my error log. I added a modeline to both Monitors but it still gives me this error and the second monitor just mirrors the first. If I try to use Option "Position" "0 0" and Option "Position" "1921 0" in the two monitors, it just turns off my second monitor. I'm still trying to get my second Monitor positioned
glisse: yeah it r3xx 3d core iirc
Loplin: http://pastey.net/136000 is what my xorg.conf looks like now, but I can't figure out what is wrong. It works the way I want if I manually do 'xrandr --output VGA-0 --right-of DVI-0`
evil_core: muffin2: I am building glest
evil_core: I know whats that, saw it few years ago
muffin2: evil_core : what kind of linux do you use?
evil_core: a nyway, can somebody help me with getting DRI2 w/lo root?
muffin2: do you use gentoo?
evil_core: no, its PLD, not recommend for inexperienced users
evil_core: but I am using custom kernel, DDX driver and Mesa
evil_core: so distro doesnt do much there
evil_core: except we got xserver 1.8. etc
evil_core: muffin2: maybe try with d-r-t, mesa/DDX master
evil_core: muffin2: you can install them with different prefix w/o messing main system
evil_core: and easily switch between them
muffin2: evil-core : I just give up on this.
muffin2: I'm too tired
AndrewR: evil_core, strange, it works for me, from user (KMS on rv280, slackware -current)
muffin2: evil_core : how did you learn all that?
AndrewR: i'm in group video .... and file permisson set as crw-rw-rw- root/video for card0 and crw-rw---- root/video for controlD64
evil_core: muffin2: because I bought laptop thats not supported by fglrx and I hate windows
evil_core: muffin2: and I dont want another laptop :P
muffin2: evil_core : I know the motive now. where did you learn those?
evil_core: wheres card0?
evil_core: muffin2: sitting in #radeonhd, in #radeon and reading aring
AndrewR: evil_core, usual Section "dri" Mode 0666 EndSection in xorg.conf .. may be something in udev rules, but i think i never touched any udev files for dri
evil_core: but in the past I was using mach64 and #dri was useful
AndrewR: evil_core, ls -la /dev/dri/ it has card0 and this controlD64
evil_core: I was changing game code that hanged GPU, and dri DEVS after seeing patches for games(mainly commeted codes) were fixing mach64 driver
evil_core: crw----rw-+ 1 root video 226, 0 May 2 22:24 /dev/dri/card0
evil_core: crw-rw----+ 1 root video 226, 64 May 2 22:24 /dev/dri/controlD64
evil_core: bso permissions are wrong?
evil_core: I am in video group
AndrewR: evil_core, if setting rw for video group will not cure your KMS .... may be there was something else. just try ....
AndrewR: evil_core, i'm not sure what people think is OK for security stuff
evil_core: AndrewR: DRI worked in UMS before reboot
evil_core: ouch, I rebuilt from scratch DDX and drm
evil_core: I would forgot about that ;)
evil_core: AndrewR: anjyway it should work, because everybody got rw to cvard0 and video to controlID64
evil_core: and I am in video group
AndrewR: evil_core, oh, and if your distro is Fedora - they may have SElinux or something like this .... no idea if it interfere with dri .... (i'm not sure in what order permissions are applied, may be kernel just check for group, and if your grop have no rights - skip user checks ?)
evil_core: maychmod a+rw /dev/dri/*
evil_core: its PLD
evil_core: I got d-r-t w/o selinux
evil_core: i recompiled DDX/drm(with make distclean
evil_core: and chmoding didnt helped
AndrewR: evil_core, may be i has some option in libdrm ... one maoment
AndrewR: evil_core, no .... (there was one about udev, but i drop it without any bad effect)
AndrewR: *droped
evil_core: open("/usr/lib/libdrm_radeon.so.1", O_RDONLY) = 4
evil_core: read(4, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0@\16\0\0004\0\0\0"..., 512) = 512
evil_core: fstat64(4, {st_mode=S_IFREG|0755, st_size=13668, ...}) = 0
evil_core: mmap2(NULL, 16648, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 4, 0) = 0xfffffffff77c9000
evil_core: mmap2(0xf77cc000, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 4, 0x2) = 0xfffffffff77cc000
evil_core: close(4) = 0
evil_core: mprotect(0xf77cc000, 4096, PROT_READ) = 0
evil_core: mprotect(0xf7011000, 8192, PROT_READ) = 0
evil_core: munmap(0xf7014000, 217192) = 0
evil_core: open("/dev/dri/card0", O_RDWR) = -1 EACCES (Permission denied)
evil_core: geteuid32() = 1000
evil_core: getuid32() = 1000
evil_core: $ ls -l /dev/dri/card0
evil_core: crw-rw-rw-+ 1 root video 226, 0 May 2 22:24 /dev/dri/card0
evil_core: bash-4.1$ cat /dev/dri/card0
evil_core: cat: /dev/dri/card0: Permission denied
evil_core: theres '+'
evil_core: what it means?
AndrewR: heh, guess it was pastebin material ....
mancha: i am going to invent a way to post large scroll on a website
mancha: oh wait, that's already been invented!
evil_core: ouch, sorry
AndrewR: drwxr-xr-x for /dev/dri ? (dir permission)
evil_core: $ ls -ld /dev/dri
evil_core: drwxr-xr-x 2 root root 80 May 2 22:24 /dev/dri/
evil_core: recompiling with --disable-udev didnt helped
AndrewR: evil_core, not sure what "+" at the and of crw-rw-rw-+ mean .... may be some "root-only" bit?
evil_core: but paste your permisions
evil_core: did you built with libkms?
AndrewR: http://pastebin.com/xzuzTSaQ - not sure about libkms ... i never asked for it, but may be got it by default
evil_core: udev helped
evil_core: disabling, and recpmpiling xf86-video-ati
evil_core: but mine system is udev, not mknod based anyway :/
Draconx: evil_core, the + means that the file has ACLs associated with it, which are probably causing your grief.
Draconx: I think getfacl /dev/dri/card0 will print them out...
Godmode: is porting the r600g code to the newer API supposed to be difficult or is it just me?
evil_core: hmm..now gallium doesnt work at all
evil_core: glxinfo[29202]: segfault at 9c ip 00000000f706a54c sp 00000000fff07290 error 4 in radeong_dri.so[f7058000+38a000]
ezzieyguywuf: are there known problems with opengl and the radeon driver?
MostAwesomeDude: Yes. It's not perfect.
MostAwesomeDude: Could you describe your problem?
evil_core: In file included from ../common/drisw_util.c:30:
evil_core: ../common/drisw_util.h:39:20: error: mtypes.h: No such file or directory
evil_core: In file included from ../common/drisw_util.c:30:
evil_core: mesa master
ezzieyguywuf: MostAwesomeDude: yes. When I enter the game world in World of Warcraft via wine, the game crashes. The guys in #winehq seem to think its a video driver problem
MostAwesomeDude: ezzieyguywuf: Entirely possible; what kind of crash is it?
ezzieyguywuf: MostAwesomeDude: would it help if I pastebined the output from wine?
MostAwesomeDude: ezzieyguywuf: Sure. What kind of crash, though? Does it quit unexpectedly?
evil_core: MostAwesomeDude: will you help also me with half life?
evil_core: mesa master is broken, got ot #mesa or something?
evil_core: it doesnt built
ezzieyguywuf: I get a windows popup that says the game has crashed, and that its probably the game's fault. Then I can still see my avatar nad stuff, though its frozen, and my only recourse is to ctl+c wine
svenstaro: MostAwesomeDude: will you help me with crysis?
MostAwesomeDude: We're talking about it on the ML. Back out the last day's changes and it should be fine.
MostAwesomeDude: svenstaro: Dunno if we provide all the stuff needed for it to trun.
ezzieyguywuf: http://dpaste.com/190178/
svenstaro: MostAwesomeDude: meh sorry I was jk
ezzieyguywuf: MostAwesomeDude: ^^ wine's output
MostAwesomeDude: svenstaro: :3
svenstaro: MostAwesomeDude: just make KMS performance suck less and I'll be completely happy
MostAwesomeDude: svenstaro: 'k.
svenstaro: MostAwesomeDude: That one was serious :)
MostAwesomeDude: ezzieyguywuf: From that, it would appear that it isn't our bug.
ezzieyguywuf: MostAwesomeDude: hrm. what would you suggest I do then? head over to #winehq again?
evil_core: diff --git a/src/mesa/drivers/dri/common/drisw_util.h b/src/mesa/drivers/dri/common/drisw_util.h
evil_core: index 2b54b22..9c3d01c 100644
evil_core: --- a/src/mesa/drivers/dri/common/drisw_util.h
evil_core: +++ b/src/mesa/drivers/dri/common/drisw_util.h
evil_core: @@ -36,7 +36,7 @@
evil_core: #ifndef _DRISW_UTIL_H
evil_core: #define _DRISW_UTIL_H
evil_core: -#include "mtypes.h"
evil_core: +#include "main/mtypes.h"
evil_core: #include
evil_core: #include
evil_core: [it fixes mesa master
MostAwesomeDude: ezzieyguywuf: File a Wine bug. If they have questions, they can ask us.
MostAwesomeDude: evil_core: Dude. You should know better.
ezzieyguywuf: MostAwesomeDude: will do.
Dr_Jakob: evil_core: scons?
evil_core: http://pld.pastebin.com/cB4TMhxK
evil_core: no, code is bropken in master
svenstaro: evil_core: Please dont paste into here
evil_core: svenstaro: ok, I will trye ;)
evil_core: but can you apply it upstream?
ezzieyguywuf: does anyone in here by chance know what chipset my ati firegl v5200 is ?
airlied: ezzieyguywuf: rv530
ezzieyguywuf: airlied: thanks
evil_core: b44c52870dc2eec9a41e984d4dceca22ea83d334 breaks mesa
evil_core: airlied: maybe you can apply that stupid patch?
evil_core: radeong is broken currently?
Godmode: heh
evil_core: Godmode: ?
Godmode: im just laughing at the scrollback.
zhasha: airlied, ping?
evil_core: http://www.free3d.org/
evil_core: is it a joke?
Godmode: glxgears as a benchmark? hahah!!
evil_core: http://wiki.x.org/wiki/radeon
mattst88: looks like they're just _really_ uninformed.
evil_core: theres link to it
evil_core: See RadeonFeature and RadeonProgram for driver feature and supported program lists. For more 3D information see Radeon 3D acceleration Portal.
zhasha: evil_core: I think they're serial
evil_core: and link redirects to it
mjr: super serial
evil_core: yeah, but users are redirected there from offical radeon site
evil_core: it everything in "status" section
evil_core: and we later wonder peoples goes there and by "terribly bad performance" they understands less fps in glxgears than on older card
zhasha: evil_core: to be fair, r300g did once render so slowly that glxgears was a valid benchmark
evil_core: Currently Recommended Card for Best 3D Performance: X800
evil_core: but its not gallium related
mattst88: the "glxgears is not a benchmark" meme is missing the point.
zhasha: anyway, airlied or glisse: should I update to rawhide (devel installation) if only just to try out a newer kernel, because the bug regarding SVideo and possibly VGA-out still persists in F13
mattst88: it's not a _good_ (read: comprehensive) benchmark, but it is a benchmark pretty clearly
zhasha: mattst88: ideally it benchmarks SwapBuffers
evil_core: whis is missing in r500 ;)
evil_core: ouch, pageflipping*
evil_core: but lack of pageflipping doesnt affect performance in real apps
zhasha: evil_core: if it's not using pageflipping, then it's just blitting, either way, 500 blits per second is perfectly acceptable
evil_core: I know that after disabling buffer swapping I didnt got more fps
evil_core: in q3a
_ds_: FWIW, I see 1720fps and lots of CPU usage with the default window size, and 92fps with next to no CPU usage when full-screen. (rv370)
zhasha: I know the implications, believe you me. Page flipping is still desired though
evil_core: for tearless and power saving
evil_core: but I doubt it will help performance
zhasha: I wrote a "blitter" that would map a surface, read back into floats in [0;1], then convert it to integers in [0;255] and finally upload it to X. If you can, you might as well avoid blits
zhasha: it can use a large amount of CPU time if it falls back
zhasha: I'm sure there's some esoteric X11 feature that will make it fall back :P
evil_core: if fallback to buffer copying?
zhasha: the reason I wrote that was because I didn't know how to implement a DRI2 handler in my case
evil_core: http://carme.pld-linux.org/~evil/radeon/r5xx_pageflip/
evil_core: what is hanging GPU, drm or DDX code probably?
Godmode: did you write that?
evil_core: Godmode: I wrote it using agd5f, glisse and some other peoples ;)
evil_core: its groupwork
evil_core: but it hangs GPU when q3a is started and pageflip is enabled in xorg.conf
Godmode: hm
evil_core: Godmode: important logs are in .txt file
Godmode: i see
Godmode: i dont use r500 do i have no idea
Godmode: *so
evil_core: Godmode: what card dou you have?
evil_core: I am going sleep anyway
Godmode: RV730, or RV740 forget which one Radeon 4650 with a terrible half ass fan
evil_core: good night all
Godmode: good night!
evil_core: RV are RS replacements?
Godmode: RS is mobile i think
Godmode: RV is desktop.
Godmode: im not sure though
evil_core: glisse told me that RS were IGP
Godmode: hes right then
evil_core: Godmode: but is it IGP also?
Godmode: mine is PCI E
Godmode: i dont think RSs are ever PCI Es
evil_core: http://www.amd.com/us/products/technologies/ati-hyper-memory/Pages/ati-hyper-memory.aspx
evil_core: is HyperMemory supported by free driver?
Godmode: heh
evil_core: why mine 256MB V5200 shows under windows as 512MB?
kmg: hey, i've got an x1400 running with r300. when i try to run a specific Wine app i get the following: drmRadeonCmdBuffer: -22. Kernel failed to parse or rejected command stream. See dmesg for more info.
evil_core: or maybe linux shows 256MB only
Godmode: evil_core, mine shows up as a 2GB card
Godmode: its off by a GB :p
evil_core: its hypermemory?
kmg: dmesg reports: [drm:r100_cs_track_check] *ERROR* [drm] No buffer for z buffer !
Godmode: what is with this hypermemory thing?
evil_core: but I got FireGL, so dunno now :/
evil_core: http://en.wikipedia.org/wiki/HyperMemory
Godmode: yeah i know what it is
evil_core: something that shouldnt be in hi-end cards
Godmode: but like... ugh....
kmg: oh hey, evil_core, i was googling around for this error and found a log where you were talking about it in this channel.
evil_core: http://forum.thinkpads.com/viewtopic.php?f=30&t=24163
evil_core: its hypermemory
evil_core: kmg: yeah, remember that got somethinmg like that, check dmesg
evil_core: kmg: isnt that some old gfx card?
evil_core: UMS or KMS?
kmg: yeah it's old and support isn't that great.
evil_core: but what?
dark: Are HD 5850, Hd 5750, HD 4770, HD 4350, .. models included in the specs released by AMD? I'm trying to check on http://dri.freedesktop.org/wiki/Radeon , but the 'hardware documentation from amd' seems to cite chipsets (r3xx 3d registers, etc), not those models.. (I want to know if, for example, "Ati Radeon Hd4350 1gb 1 Gb Ddr2 Sapphire" has the spec released)
evil_core: and what game
kmg: evil_core, i don't know what mode setting i'm using. i haven't really messed with anything, i'm using fully updated ubuntu.
kmg: evil_core, and i'm trying to run lightwave 3d, 3d modeling / animation app.
evil_core: kmg: show me link to this log, I forgotten what problem it was
kmg: http://people.freedesktop.org/~cbrill/dri-log//dri-devel-2010-01-13.log
kmg: "evil_core> gallium is totally broken now"
kmg: i think you're talking about r500 though
kmg: should i try and follow http://wiki.x.org/wiki/radeonBuildHowTo and rebuild everything?
mattmatteh: pokes MrCooper about ppc radeon patch
evil_core: kmg: sure, you will get better performance and less problems then
kmg: okay, i'll try that. i'm sure it'll take me like forever, and i'll probably have problems along the way, which i'll bug the channel about.
kmg: thanks
evil_core: kmg: and dont replace original files in system, use separate prefix, than you can switch between dri drivers on the fly(when starting app)
evil_core: kmg: wait
kmg: okay
kmg: i can prooobably figure that out
evil_core: http://pld.pastebin.com/cB4TMhxK
evil_core: apply it on mesa if you will get error with mtypes.h
evil_core: dunno if they applied it upstream
kmg: oh, it's a patch to fix an error, if i get it?
kmg: thanks
evil_core: kmg: its applied upstream anyway
kmg: okay cool
evil_core: kmg: http://carme.pld-linux.org/~evil/radeon/, lookover building_scripts and running_scripts
kmg: i will
evil_core: kmg: http://carme.pld-linux.org/~evil/radeon/dirhack/
evil_core: new updated version
zhasha: airlied: if I may remind you, I'm getting really tired of not being able to use my VGA port. Do you have anything I need to try out before jumping to conclusions, can you tell me where the offending code is, or are you on as bare a bottom as me?
airlied: zhasha: no idea whats wrong, is there a bug open to track it? otherwise keep losing things ;-)
zhasha: airlied: it's probably related to the S-Video bug I opened
zhasha: here's essentially what happens. S-Video is always "connected" according to the driver. When I actually connect S-Video and enable the output, garbage ensues. Now VGA broke around the same time, however it's not perma-on. The thing with VGA is that it will output the correct image, but either flickering like all hell, or garbled up somehow. Eventually, if I disable/reenable enough times, it will work correctly, at complete random
zhasha: https://bugs.freedesktop.org/show_bug.cgi?id=25386
airlied: zhasha: can you bisect to when it went wronger?
zhasha: I'm assuming that means bisect the kernel, got a good guide for that, or will I be rebooting a centillion times? I can tell you the RPM in which it went wrong
airlied: oh that might help
zhasha: I had rawhide running all the way up to F12 was released
airlied: try booting with radeon.new_pll=0 maybe
zhasha: it worked up until then, and the RPM that went into F12 stable is the one that broke it
zhasha: airlied: radeon.new_pll=0 does nothing
zhasha: interestingly enough, debugfs says I have dri/0 and dri/64, where dri/64/name is empty
zhasha: is this normal?
airlied: yeah
zhasha: I'll install rawhide if necessary. I'm set up for it anyway. I just want this bug way the hell out before F13
agd5f: zhasha: does ums still work correctly?
zhasha: agd5f: I honestly haven't checked. does radeon.modeset=0 still work?
agd5f: zhasha: yes
agd5f: if it does, open a bug and attach the output of avivotool regs all for both the ums and kms cases, and I'll fix it tomorrow
zhasha: agd5f: fairly good point made there. It works perfectly under ums
zhasha: VGA at least, and S-Video is not incorrectly detected
agd5f: zhasha: IIRC, in ums load detection is disabled for tv-out by default
zhasha: in any case, VGA is busted under KMS, but works under UMS
zhasha: sorry, this TV is too new (and not my TV) to have SVideo, and I don't have any capacitors lying around (not my house)
agd5f: zhasha: what system is this? what chip? r5xx? r6xx? r4xx?
zhasha: RV515 (M54) aka Radeon Mobility X1400
agd5f: zhasha: both dacs work fine on the rv515 card one I've got here
agd5f: You'll have to attach the avivotool logs
zhasha: agd5f: any specific way I need to run avivotool?
kmg: radeon: stable or development?
kmg: how far ahead is dev from stable?
soreau: compiz-0.9 crashes when I enable/disable some plugins in ccsm http://sprunge.us/gaFc but I'm not sure why it seemingly crashes in the driver. Also this doesn't happen with compiz 0.8
vsrinivas: hmm, on teh radeon, the BITBLT_MULTI packet can take dst coordinates in the -8191 to +8191 coordinate range; what if the visible area covers that entire size... is there any way to bitblt to offscreen areas then:?
MostAwesomeDude: Not gonna happen, as you're not going to have any single pixmap covering more than 4kx4k.
agd5f: zhasha: avivotool regs all
agd5f: zhasha: although I think I know what might be happening
zhasha: agd5f: any specific point at which you want me to run that?
agd5f: zhasha: under ums when it's working and under kms when it's not working
agd5f: avivotool regs all > working.out
zhasha: so, clean boot, connect nothing and dump regs?
agd5f: zhasha: no with the vga enabled
zhasha: okay, but what about isolating SVideo?
agd5f: not relevant
vsrinivas: MostAwesomeDude: oh, that's useful. :)
agd5f: vsrinivas: offscreen pixmaps have different base addresses
agd5f: the coordinates are relative to the blit src/dst addresses
vsrinivas: ok. thnaks.
zhasha: agd5f: http://zhasha.com/regs.tar.gz
agd5f: zhasha: does this patch help? http://people.freedesktop.org/~agd5f/pm-drt/0004-drm-radeon-kms-atom-disable-the-encoders-in-encoder.patch
agd5f: disable TV out before you try and enable VGA
zhasha: I'm not sure how to do that
agd5f: zhasha: xrandr --output S-video --off
zhasha: you want me to do that in KMS?
agd5f: or DIN or whatever your tv-out connector is
agd5f: yes
zhasha: well, it appears I can't turn S-video off, or it is already off
zhasha: it's reported as connected, but not enabled
zhasha: my only TV out connector is S-video. I supposedly also have a DVI connector, but it's not on the laptop itself
zhasha: oh, and S-video likes to turn itself on when running xrandr, seemingly at random
zhasha: and it was supposedly off when I did that dump
zhasha: regarding that patch though, I'm not set up to compile my own kernel right now
DanRose: There's no open source driver for a RV380 [Radeon X600] is there?
zhasha: DanRose: 2D or 3D?
zhasha: (both are a yes, but they're different drivers)
mattst88: national library of australia...
DanRose: 3d. What's wrong with my domain?
zhasha: DanRose: mesa3d, r300
DanRose: and 2d?
mattst88: DanRose, no, nothing, just interesting. :)
zhasha: xf86-video-ati
DanRose: I'm an interesting guy.
DanRose: Thank you!
zhasha: that's pretty much what every distro will install for you
DanRose: gentoo :)
mattst88: DanRose, just set VIDEO_CARDS="radeon" in /etc/make.conf, and compile X like normal
DanRose: done it, it works, just slow.
DanRose: just checking out the options really.
zhasha: what exactly is slow?
DanRose: googleearht
zhasha: does glxgears report direct rendering, and what does it say your renderer is?
zhasha: glxinfo*
DanRose: OpenGL renderer string: Software Rasterizer, which is what got me wondering
zhasha: try LIBGL_DEBUG=verbose glxinfo
agd5f: DanRose: depending on your distro you may need to install the linux-firmware package. or if you are running a 64-bit OS, you'll need to install the 32 bit 3d driver for googleearth since it's a 32 bit app
DanRose: OpenGL renderer string: Software Rasterizer OpenGL version string: 2.1 Mesa 7.8.1
DanRose: thanks
zhasha: what? no, I need the actual debug info, where the error is reported
DanRose: no error.
zhasha: what X driver are you using then?
zhasha: vesa?
DanRose: Driver "radeon"
DanRose: so there you go eh? I hadn't looked that far.
soreau: DanRose: Can you pastebin your X log?
agd5f: DanRose: check dmesg and your xorg log
DanRose: a lot of extra crap in there
DanRose: http://paste.pocoo.org/show/209175/
zhasha: (EE) RADEON(0): [dri] RADEONDRIGetVersion failed to open the DRM
zhasha: [dri] Disabling DRI.
zhasha: there be your problem
DanRose: aha! So how do I fix that? Is that the missing firmware?
zhasha: directly above it: '[drm] failed to load kernel module "radeon"' - meaning you're apparently missing the kernel portion (drm) of the driver
DanRose: brb
zhasha: could be that
zhasha: could be you never compiled that drm driver
soreau: thats the most likely case
soreau: need to enable drm>radeon in the kernel
DanRose: <*> ATI Radeon [*] Enable modesetting on radeon by default - NEW DRIVER
DanRose: It's on.
DanRose: damn.
DanRose: bblater
soreau: Can you pastebin dmesg output?