Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

Radeon IRC Logs For 2008-11-16

Search This Log:


damentz-afk: is away: Gone away for now
airlied: AndrewR: pong
damentz: is back.
damentz-afk: is away: Gone away for now
soreau: eww, away message
AndrewR: airlied, rv280 kernel modesetting currently broken?
airlied: AndrewR: it should in theory work with modeset=1
AndrewR: airlied, kernel part work, but xf86-vide-ati (from modesetting-gem-cs branch) give me 'no modes'
AndrewR: * xf86-video-ati
airlied: AndrewR: post the xorg log
AndrewR: http://pastebin.com/d5e3565e2
AndrewR: it was worked around october, 21
AndrewR: i use your new rebased patch against 2.6.27
airlied: wierd... looks like its not picking up the output
airlied: pastebin dmesg
airlied: I can test on my r1/2xx tomorrow at some point
AndrewR: airlied, one moment ...
AndrewR: airlied, http://pastebin.com/mf103f29
airlied: AndrewR: you running any bits from modesetting-gem?
airlied: the API was broken.
airlied: you need to revert some of the cleanups from Jako.
airlied: I should probably branch a libdrm with what Fedora is using for now.
AndrewR: yes, libdrm modesetting-gem current + your modesetting-gem-cs DDX
airlied: yeah should probably try libdrm from 1c0d8632
airlied: I need to do a branch without Jakobs cleanups.
airlied: or create a patch on top of my Fedora branch.
AndrewR: i will try older libdrm ...
SimpleAONE: http://hdrlab.org.nz/radeonhd-driver/
airlied: /win 25
airlied: oops
AndrewR: airlied, with libdrm from 12 november it works, but crashes under kwin (KDE 3.5.10, composite enabled) if i move window... also if mplayer will try to use xv - it will crash, too. Also i have problem with my arrow keys ..so - reboot.
nobody888: has someone a working xorg.conf for tv-out ?
soreau: You can use xrandr to enable tv-out
nobody888: no I cant:(
nobody888: I can'get a xserver to start on my dvi-out or on my tv-out
soreau: Do you have your drivers installed correctly?
nobody888: only if I only connect my monitor on dvi and nothing else, then I can start an xserver
soreau: After starting X, you can use xrandr
nobody888: but then xrandr can't find the SVIDIO output:)
soreau: Well, it's probably not called SVIDIO..
nobody888: xrandr -q shows only dvi
soreau: Can you pastebin /var/log/Xorg.0.log?
nobody888: root
nobody888: ups
nobody888: wrong window
janboe: Hi, I am a newbie to write driver for radeon. But I have a strong interest in it. Is there some kind one tell me the start point?
soreau: I would say grab the source and get it compiling first :)
nobody888: http://pastebin.com/d64697172
soreau: nobody888: You have Radeon X1550 and I'm not sure how well the open driver supports that hardware yet
nobody888: I also tried a x1300
nobody888: thats the same
janboe: thanks. I already have the code and I am reading the code. But I do not understand the PLL clock programing section.
soreau: And as a side note with x1.5 you probably want to be using exa rendering
soreau: janboe: I figured you'd say something like that ;)
soreau: janboe: But you've come to the right place. I am not experienced on writing these drivers, but just ask your question in detail and wait for someone who knows
nobody888: soreau: I have no picture on tv-out, so I don't need rendering at the moment I tink:)
glisse: janboe: pll are for modesetting
soreau: nobody888: (--) RandR disabled
glisse: there is somethings explaining this on xorg wiki
janboe: Is there some document for my radeon 9550?
nobody888: how do I enable it?
soreau: nobody888: You have aiglx disabled
glisse: pll theory apply to all hw
glisse: than you have specific hw tweak
janboe: glisee: thanks. Yes. I am interesting in modesetting.
nobody888: soreau: I have not loaded any module, it's a very simple xorg.conf at the moment
soreau: nobody888: And seems you're not even using the open drivers..
soreau: Pastebin your xorg.conf
soreau: well at least it makes mention of fglrx in your x log
nobody888: http://pastebin.com/d672b3752
soreau: nobody888: If you want to use the open ati driver, you will need to remove any traces of fglrx
nobody888: I did
nobody888: I thought
nobody888: rpms are all removed
soreau: nobody888: The first thing you need to do is enable aiglx and exa in xorg.conf
soreau: Also, those tv lines you have in there are being ignored
nobody888: So
nobody888: I added this to my xorg.conf:
nobody888: Section "ServerFlags"
nobody888: Option "AIGLX" "on"
nobody888: Option "exa" "on"
nobody888: EndSection
nobody888: then I plugged the svhs cable off, so that ony my monitor is connected via dvi
nobody888: the I rebootet the system
nobody888: connected the svhs cable again
nobody888: Now I can start the xserver on my monitor, but xrandr don't see the SVIDO Ouput
nobody888: only Dvi-0
soreau: Well you got the aiglx part right, but I'm not sure exa is supposed to be in that section
soreau: Remove that exa line and put this in your Device section:
soreau: Option "AccelMethod" "EXA"
nobody888: ok
soreau: nobody888: And remove the line: Option "DRI" "0"
nobody888: ok now my xserver don't start anymore
nobody888: because I removed the Option "DRI" "0"
soreau: Pastebin /var/log/Xorg.0.log again
nobody888: drmOpenDevice: node name is /dev/dri/card0
nobody888: drmOpenDevice: open result is -1, (No such device or address)
nobody888: drmOpenDevice: open result is -1, (No such device or address)
nobody888: drmOpenDevice: Open failed
nobody888: ok
nobody888: http://pastebin.com/d7ec8a58a
soreau: (WW) RADEON(0): R500 support is under development. Please report any issues to xorg-driver-ati@lists.x.org
soreau: heh
nobody888: that's not funny:)
soreau: Well, I don't know what else to tell you
nobody888: hmm
soreau: fglrx is another option for you, but that does not support xrandr
soreau: though it does support tv-out for your chipset
nobody888: fglrx have I tried
glisse: nobody888: drm isn't loaded on your computer
nobody888: but
glisse: you likely have fglrx kernel module lying around
soreau: thinks that is very likely
nobody888: videos where choppy sometimes
soreau: nobody888: Yup. That's fglrx for ya ;)
MostAwesomeDude: nobody888: Kernel 2.6.26? Did you remember to blacklist or rename fglrx.ko?
nobody888: no kernel 2.6.27 I didn't blacklist fglrx.ko
MostAwesomeDude: 'k.
nobody888: i think I removed it
MostAwesomeDude: You need either "blacklist fglrx" in whichever modules autoload your distro does, or you need to make sure that there's no fglrx left.
nobody888: i will blacklist it
MostAwesomeDude: My favorite:
MostAwesomeDude: $ find /lib/modules/`uname -r` -name fglrx.ko -exec rm {} +
nobody888: I will try that too:)
nobody888: how do I load drm?
MostAwesomeDude: "modprobe radeon" should work just fine.
MostAwesomeDude: If you've had fglrx loaded, you *will* need to reboot with it blacklisted or removed before you can use our DRM, FYI.
soreau: MostAwesomeDude: Nice :D
agd5f: nobody888: there's no tv-out support for your chip yet
MostAwesomeDude: soreau: Once one learns find, one never goes back to "ls | grep". :3
MostAwesomeDude: I'm actually finding that I'm a very competent programmer. Except when it comes to fog, of course.
soreau: lol
nobody888: agd5f: thanks, that is the news I don't want to hear:), fuck
soreau: coulda did without the profanity :p
nobody888: do you know a pcie ati card, where the tv-out is supported?
soreau: I'd say it's better to know a feature is not yet supported than spending countless hours of miserableness trying to get something to work that wont in the first place
nobody888: hehe I'm trying since friday:)
agd5f: nobody888: it's only supported on r1xx-r3xx at the moment
nobody888: fine
nobody888: with fglrx I get choppy video
nobody888: with radeon driver I get no tv-ot
nobody888: with nvidia I get tearing
nobody888: with nv I have no tv-overscan
nobody888: I love Linux
soreau: My radeon 9600 works great with either radeon or fglrx :)
nobody888: :)
nobody888: but it not a pcie card
soreau: yup
nobody888: and you have never choppy video?
nobody888: with fglrx?
soreau: I don't know what you mean by that exactly
soreau: With video driver x11 it is slow, with opengl it is great, but not while running compiz (flickers)
bobbens: one of my box has a 9600, it did pretty poorly with fglrx
bobbens: but last time i tried was 2 years ago :)
bobbens: I use my X550 which works wonders with radeon, r3xx stuff is pretty nice, if only for GLSL :)
soreau: bobbens: fglrx has improved over the years but it's still not perfect
bobbens: soreau: i stopped bothering with fglrx a while back, just waiting patiently for radeon to mature more and reach the long awaited opengl 2.0 spec :)
nobody888: with fglrx,
nobody888: I get such errrors every 3-5 minutes, while playing videos:
nobody888: AddAudioData():p2: Audio buffer overflow, audio data lost!
nobody888: and then the video stops for 5-10 seconds
soreau: I test both drivers fairly routinely and I am surprised by the developments of the open driver
soreau: nobody888: You're probably using the wrong video out driver
nobody888: yes
nobody888: I know
nobody888: but I use mythtv and it uses opengl or xv and both are not nice
nobody888: with x11 and mplayer it seems to work
damentz: is back.
damentz-afk: is away: Gone away for now
damentz-afk: is back.
jcristau: damentz-afk: please turn off the verbose away
damentz-afk: fine
soreau: thank you
SimpleAONE: does it also happen in linux that the videomemory when its 256mb or more is divided into blocks ?
glisse: SimpleAONE: this is pci bar size
adamk: Heh... KMS, at least in the F10 preview, has some issues... I have two monitors, an DFP connected directly, and a CRT connected through a KVM. It adjusted the resolution of the panel to 1280x1024, and the CRT is still at 640x480. And, of course, the LCD is only actually using what appears to be 640x480, leaving the rest of the monitor blank.
adamk: The installer is completely unusable because once X started, both screens are at 640x480...
chithead: adamk: if you have an r400, try to enable atombios
adamk: I haven an r500... Still worth enabling?
chithead: r500 should always use atombios, only for r400 it is optional
chithead: maybe the kvm does not properly forward ddc information
adamk: Oh, I know it doesn't. I have to specify a modeline in my xorg.conf file for it.
adamk: For the CRT, that is.
adamk: But I'm thinking KMS should be a little more graceful.
adamk: Or maybe Fedora needs to make their installer usable at 640x480 :-)
chithead: iirc, fedora also has a text mode installer
adamk: Well I still think KMS should handle the situation better. Having a 640x480 console when the screen resolution has been set to 1280x1024 is pretty bad.
chithead: it could be that before kms is claimed to be production-ready, users would be able to inject modes with fbset or similar
chithead: and besides, the code is there, so why don't you implement it yourself?
adamk: I never said I was a programmer. Just an end-user reporting an issue. Isn't that what this channel is for, among other things? :-)
airlied: adamk: if it can't detect the monitor it uses 640x480, though I need to change that to at elast 800x600
airlied: the console justs uses the lowest common denom, however X should be smarter when it start.
adamk: airlied, It's possible that X will be smarter after the install. The F10 install keeps crashing on me, so I can't really check yet :-)
dmb: i wonder if there is a way to force x to use 2 colors :P
dmb: even on a modern card
spstarr: wasn't that removed? :)
spstarr: black and white
glisse: airlied: hhhmm compiz is allmost working
glisse: well only screenshot is broken
TobiasTheCommie: glisse: i thought compiz already worked
glisse: TobiasTheCommie: dri2 != compiz working :)
glisse: for getting compiz you need to implement the dri2 Settexbuffer extension
TobiasTheCommie: yes
TobiasTheCommie: but
TobiasTheCommie: 5/27/08 06:29 pm
TobiasTheCommie: "So I now have compiz running without any corruption or wierdness."
glisse: and you also need to fix all the bugs that come along
TobiasTheCommie: from airlied's livejournal
TobiasTheCommie: http://airlied.livejournal.com/60180.html
glisse: TobiasTheCommie: not with new command stream submission
TobiasTheCommie: ah, oki
TobiasTheCommie: :)
glisse: and not with dri2
TobiasTheCommie: actually, i never understood what specific benefits dri2 have over dri
glisse: TobiasTheCommie: no more sarea, no cliprect, ... simpler & cleaner to sumup
glisse: but it only really matter for dev, users mostly don't see the differences
TobiasTheCommie: ah, oki
TobiasTheCommie: well, if it makes it easier for you devs, it gives us a better driver :)
TobiasTheCommie: so is the dri2 still only r300? or is it r300-500?
glisse: few bits are missing for r500 in mesa
glisse: in r300_cmdbuf.c
TobiasTheCommie: oki
TobiasTheCommie: sounds good though
zhasha: Hey glisse, do you still have the source for that Gallium3D skeleton driver?
TobiasTheCommie: well, ima go grab some shut eye now
glisse: zhasha: most of it should be in my repo on fdo
zhasha: glisse, which one?
glisse: mesa
glisse: zhasha: this week i will work on gallium
zhasha: found it, thanks :) and lovely news
glisse: bringing up to date winsys bit and push out my hw accel bits
soreau: \o/ 4 glisse
zhasha: this is from the 0.1 branch though
zhasha: I clearly don't understand the gallium infrastructure :/
glisse: this repo is very old
airlied: glisse: if userspace on classic bufmgr works we should merge it to master then.
airlied: and disable the GEM bufmgr, by default, until we have it stabilised
spstarr: rmh3093: machine working?
airlied: the quicker we can stabilise the classic paths and r100/r200 the better.
glisse: airlied: i need to retest classic and fix clearmask regression
glisse: which iirc was the only regression i had
glisse: at least according to pigglit and openarena,q3,.... :)
glisse: also i have a leak with dri2
glisse: some obj are not unref enough after they are close
glisse: i suspect that dri destroy drawable is not call
airlied: I'll need to check the bufmgr covers all the bugs I've been fixing.
airlied: mainly the aperture space checks.
glisse: well i cheated their :)
glisse: i use a fixed value
airlied: well we need two fixed values :)
glisse: while i believe we should get it from the kernel
glisse: yup 2
airlied: the latest radeon-gem-cs has my updated DDX checks.
glisse: but i use 1 one, i cheat :)
airlied: play with 32MB VRAM cards makes it easy to test.
airlied: someone gave me a webapge with a 1200x8000 pixmap
glisse: i rebased my ddx on yours yesterday iirc
airlied: we need to port DDX to the new bufmgr also
glisse: well i have done half of that in my ddx
airlied: once F10 is released I will have more time to integrate.
glisse: need to think on what kind of stuff i want to add to libdrm-radeon before finishing it
airlied: don't think too much, I don't want lots of unused options :)
glisse: well it's mostly about which remaining functions in radeon_bufmgr_gem.c of ddx we want in radeon_bo_gem.c of libdrm-radeon :)
airlied: the force gtt and set limits are probably a bit hacky :)
glisse: also need to use inline for cs write and test against macro for speed comparaisons but should be api transparent mostly
airlied: have to try and fix EXA today to see if it stabilises anyone..
airlied: then find the misrendering text bug.
spstarr: will test
spstarr: if you're pushing to koji if not then git's fine
glisse: airlied: btw do you know when rpmfusion for fedora 10 will be online again ?
glisse: :)
airlied: glisse: is it not online now? I don't use it that often though.
glisse: nop
airlied: is there not an rpmfusion-devel
airlied: or rawhide
glisse: i installed an fd10 after loosing another hd yesterday
airlied: that'll be the F10 one until F10 is release
glisse: and no mp3 or video
airlied: wow you lose more harddisks than me
glisse: well i have lot of them too
glisse: somethings like 50 hd
glisse: never counted them
glisse: my thesis data is sometings around 1To
glisse: :)
zhasha: Wow glisse, you're quite the popular one
glisse: zhasha: lot of 80g/40g drives
zhasha: glisse, I was referring to the recent DRI2 porting :)
zhasha: glisse, what happens with the DDX driver when the gallium driver is done?
airlied: zhasha: nothing
airlied: gallium is only mesa side
zhasha: airlied, but I just read that they're putting video acceleration into gallium
airlied: zhasha: nothing to do with the DDX though.
airlied: you can accelerate video on the client side also or using indirect rendered video
zhasha: I thought video acceleration was an integral part of XvMC and thus the DDX drivers responsibility
airlied: XvMC is just like DRI
airlied: so the DDX might need some XvMC support code.
zhasha: I feel I really need to brush up on the structure of things
airlied: which we don't have now.
zhasha: So does that mean that regardless of galliums capabilities, the DDX driver will forever need to support the 2D aspects of things?
airlied: zhasha: there will always be a DDX yes.
airlied: if may eventually in some dream future, be quite small though however I'm not seeing the dream come true just yet.
airlied: if Gallium got an Xrender API then in theory EXA could be removed from the DDX and it could just interface to the gallium driver loaded for AIGLX.
airlied: however I'm not sure when that might ever happen and if it'll ever perform well enough
zhasha: Is the radeon DDX driver dependent on mesa r300?
spstarr: the DDX isn't dependent on mesa at all
zhasha: Remind me again why we have 2 different drivers for one device?
airlied: zhasha: where?
airlied: Mesa is a client side library loaded by 3D apps.
airlied: its not a driver as such
zhasha: But it does communicate with the graphics card
spstarr: in my dream world there would be one unified driver for 2D and 3D none of the need for distinction anymore.
spstarr: but, newer video cards there's no more '2D' on them you need 3D to render 2D
airlied: spstarr: why?
airlied: spstarr: you don't actually understand anything though.
airlied: zhasha: it sends command to the kernel to send to the graphics card yes.
airlied: but it doesn't run in the same place as the DDX.
zhasha: I should get hands on with the code (gallium)
spstarr: I hope wayland becomes something more
airlied: spstarr: stop hoping aloud please, phoronix is there for a reason :)
spstarr: hehe
mentor: mesa is an implementation of OpenGL rather than a driver
mentor: ?
spstarr: airlied: in my speculation it almost seems as if krh has 'given up' on the current X display server itself, if that was the case, I don't blame him
airlied: spstarr: no he hasn't given up.
airlied: spstarr: wayland is an experiment
airlied: mentor: the word driver is overloaded.
spstarr: i don't think anyone should be afraid of it though, his experiment
airlied: spstarr: but I think most people missed the point of it.
airlied: and are ranting cluelessly about things they don't understand
marcheu: I think we should stack up dri on top of wayland to provide gallium acceleration !
zhasha: I'd like to see a more stable Xserver, but that's about it
airlied: marcheu: there we go. dri3 surely
spstarr: well what I do understand is trying to ram modern graphics methods into 20 year old X display servers (im not taking the X11 protocol) is sorta painful and we're all seeing that result.
airlied: marcheu: on wayland5 with gallium4d
airlied: spstarr: you don't understand very well though.
spstarr: yeah, you're right on that :-)
marcheu: wow gallium4d do you need KMS for that ?
zhasha: KMS2
airlied: marcheu: no it requires a new branch of the kernel called Airluedux
z3ro: marcheu: nope, gallium4d just needs a tab of Acid. ;)
airlied: dman I fail at spelling my own name.
spstarr: tee hee
marcheu: airlied: is that linux for the drunk ?
airlied: marcheu: yeah it melds Linux and x.org in a new way that makes you get drunk faster
airlied: its sorta like a snakebite
airlied: hmm snakebite...
marcheu: snakebite gets you drunk ?
z3ro: never tried that... apparently my friend tried one, and ended up blacking out.
zhasha: off topic much?
airlied: marcheu: mixture of lager and cider :)
spstarr: oh I know it's a C++ display server !! *run*
airlied: surely the display server should be in java or mono by now.
zhasha: I don't see whats wrong with a C++ display server
airlied: we would't have any memory leaks to worry abouyt
spstarr: smirks
spstarr: XOrg#
zhasha: xorgmm
marcheu: airlied: hah good think I hang here today, learnt something
zhasha: jxorg
z3ro: feels stupid. I found out why my laptop would report duplicate symbols when loading some drivers... I forgot to mount /boot :P
z3ro: well, actually I forgot to compile the ext2 driver, so /boot wouldn't get mounted.
spstarr: heh
kapace: hello, i was building mesa, and found myself with this error: http://rafb.net/p/w4VDB695.html
airlied: you need a newer libdrm
kapace: ok
kapace: ok so i am able to compile mesa now, but i get this error with glxinfo: Error: couldn't find RGB GLX visual or fbconfig
airlied: kapace: you might need a new X server.
kapace: :|
spstarr: looks at koji
airlied: spstarr: give thew new X server in koji a go on AGP with EXA enabled
airlied: see if it locks up less...
spstarr: sure thing, good timing
spstarr: -4 not ready et
spstarr: yet
airlied: -5 should be there
airlied: hmm wonders what happened
spstarr: -5 building
airlied: -5 should be finished, I think koji is messed up.
spstarr: you have a drm fix for radeon but -111 failed
airlied: lucky it did I have more fixes....
spstarr: i'll grab that kernel when you push
spstarr: Istanbul is good now that i can even record 3D and show corruption easier
spstarr: getting Xorg...
spstarr: enabling composite
spstarr: testing
airlied: spstarr: was EXA unstable without composite?
spstarr: yes
spstarr: i can try to crash it sec. w/o com
airlied: so test that first :)
spstarr: kwin has made fixes to workaround the bug I logged
spstarr: so i do not crash enable/disabling composite anymore
spstarr: (only VT switch but thats going away with KMS enabled)
spstarr: if i drag my plasma widget on desktop it would lockup system with EXA and no composite.. trying
spstarr: wow..fast
spstarr: the only thing I notice w/o composite is the show redraw of xchat from offscreen when minimizing/restoring
spstarr: 1.5 second delay on that, i dont know if thats due to GDK and it's redrawing itself
airlied: might be worth profiling
spstarr: xchat itself? i can attach
spstarr: tries firefox
spstarr: same for firefox
spstarr: the actual widgets take longer to redraw the window appears instantly itself
spstarr: lemme record
airlied: sysprof or oprofile might help.
spstarr: looks at how to use sysprof
spstarr: Istanbul doesnt show the effect
spstarr: oh nice a gui for it
spstarr: wow
spstarr: cd /lib/modules isn't cleaned up with all those kernels
spstarr: 2.6.25.10-91.fc9.i686
spstarr: purges
spstarr: oh .. virtualbox leftovers
spstarr: oh sysprof is abandoned :/
spstarr: the gui
spstarr: er, that was for here
spstarr: I have AccelDFS on only
spstarr: oh thats so neat it shows me EVERY userspace/kernel call
airlied: hmm get a sysprof :)
spstarr: in an xml file.. sure
spstarr: got it
airlied: you may need some -debuginfo installed
spstarr: got some of those .. checking which ones im missing
spstarr: now to find a program to read the XML
airlied: sysprof.
spstarr: if i had the gui building :)
spstarr: it wants glib 2.0 got devel, gtk+ 2.x devel got that
spstarr: gdk-pixbuf 2.x i thought this i sincluded in gtk+ 2.x devel , pangoft2 (checking) and libglade 2.x i got devel
airlied: libglade-devel
spstarr: dont got that...
airlied: sorry
airlied: libglade2-devel
spstarr: dont have that
spstarr: the lib only
spstarr: pass ok got gui
spstarr: airlied: the offenders are
spstarr: ProcRenderDispatch
spstarr: ProcRenderComposite --> CompositePicture
airlied: wierd I'd expect one function in the self column
spstarr: exaComposite -> ExaCheckComposite
spstarr: exaComposite is worst
spstarr: which side left or right?
airlied: self column
airlied: on left side
spstarr: the worst there is memmove()
spstarr: from kernel
spstarr: that is minimize/restoring Xchat window
spstarr: callers from that is
spstarr: RADEONDownloadFromScreenCP
spstarr: 23.58 (23.90 total)
spstarr: from that tree:
spstarr: RADEONDownloadFromScreenCP -> memmove() -> irq stuff -> drmCommandNone() -> drmCommandWriteRead...
spstarr: RADEONCPGetBuffer, RADEONCPFlushIndurect, drmCommandWriteRead -->ioctl()
spstarr: roughly
spstarr: find / ...
spstarr: same path
spstarr: flushing going on
spstarr: worse offenders inside memmove() calls
spstarr: RadeonBlitChunk
spstarr: some kernel syscalls which are ioctls mainly
spstarr: RADEONGetPixmapOffsetPitch
spstarr: drmCommandNone's kernel calls which drm ioctls
spstarr: with
spstarr: on left side again: second offender is radeon_commit_ring (8.59)
spstarr: radeon_do_wait_for_idle (5.79)
spstarr: delay_tsc (5.17)
spstarr: pixmap_op (4.23)
spstarr: and last dixLookupPrivate (2.15) the rest are < 1.99
spstarr: delay_tsc... i dont have TSC timer on this laptop
spstarr: [ 28.125112] Clocksource tsc unstable (delta = -184378429 ns) (it would turn it off?)
spstarr: i can force it off however it it's using it
spstarr: brb rebooting forcing tsc off..
spstarr: trying again
spstarr: no tsc used delay_loop instead
spstarr: memmove (29.56) -> RadeondDownloadFromScreenCP 29.50
spstarr: whatever drmCommandNone is doing with delay_loop that is causing 5.13 alone
spstarr: pixmap_op is 5.69
spstarr: radon_do_wait_for_idle 6.93
spstarr: radeon_commit_ring 10.02
spstarr: pixmap_op's most offending call in that tree is pixmap_region_substractO (1.92)
spstarr: itself is 2.10
spstarr: dixLookupPrivate 2.78 and most offending is exaGetPixmapOffset (0.74)
spstarr: tied exaPixmapIsOffScreen (0.74)
spstarr: airlied: any of that useful?
stoned: hi
stoned: 'radeon' driver doesn't support 3D acceleration in r5xx cards? I have x1900xt and I use the radeon driver on debian sid and I have no accel
stoned: please advise
ssieb: stoned: what version of the driver do you have?
ssieb: it's very hard to tell what the version number of the cards is, but I think I have an r5xx type card and get 3d acceleration
ssieb: it's not the greatest yet, but it works reasonably well for most things
spstarr: can't believe writing to the GPU causes so much delay something has to be wrong somewhere
stoned: sorry
stoned: eyeglassliquidators called me
stoned: its freakin 12:19am
stoned: I'm not going over there until morning, let the server stay down
stoned: I haven't slept in 36+ hours ffs.
stoned: ok enough
stoned: !@#$@#
stoned: ssieb, I let me check
stoned: Version: 1:6.9.0-1+lenny4
spstarr: 'eyeglass liquidators' hmm
stoned: xf86-video-ati 6.9.0
airlied: spstarr: will try and have a look later.
spstarr: ok
stoned: I get free stuff all the time
airlied: just building 780G test box.
stoned: the owner is a buddy of mine he gave me a pair of Bvlgari yesterday
stoned: :D
stoned: so here is a weird issue. When I use my x1900xt with radeonhd, monitor says no input signal, when i use it with fglrx, same thing. It only works with 'radeon' driver at 1680x1050 and even in windows xp it says no input signal on anything other than native vga mode
stoned: I dunno what the problem is
stoned: but apprantly I am stuck with 'radeon' driver cuz thats the only one that works
spstarr: thats a good thing :)
stoned: so how can I get 3d accel on x1900 with radeon driver
stoned: is this possible?
airlied: stone get Xorg 1.5 + mesa 7.2
stoned: how
airlied: or get a distro with all the bits already in it.
stoned: I have debian sid
airlied: use experimental
airlied: it should have all the bits.
airlied: pull xorg server and mesa from it
stoned: xserver-xorg-core (2:1.4.2-9 [amd64, arm, hppa, i386, ia64, kfreebsd-amd64, kfreebsd-i386, mips, mipsel, powerpc, s390, sparc], 2:1.4.2-8 [alpha, armel, hurd-i386, m68k])
stoned: yes sid
stoned: experimental is 1.5?
stoned: ok
stoned: ok wish me luck in screweing up my system
stoned: airlied, which packages do i need?
stoned: Setting up libgl1-mesa-dri (7.2-1) ...
stoned: Setting up xserver-common (2:1.5.3-1) ...
stoned: Setting up libgl1-mesa-glx (7.2-1) ...
airlied: stoned: yup and -ati driver
stoned: Setting up xserver-xorg-video-radeon (1:6.9.0+git20081012.c0e6cb6d-1) ...
airlied: yup all look good.
stoned: now I restart X?
airlied: I'mm not sure what debian names are like
airlied: you may need a kernel
stoned: radeon package is a wrapper for ati or soemthing
stoned: vice versa
stoned: same thing
airlied: yup same thing
stoned: Setting up xserver-xorg-video-ati (1:6.9.0+git20081012.c0e6cb6d-1) ...
stoned: now to restarting X?
stoned: do i need any certain options inside the xorg.conf?
stoned: $>glxinfo
stoned: name of display: :0.0
stoned: Error: couldn't find RGB GLX visual
stoned: whats gonig on?
stoned: oh i have to load glx extention
stoned: direct rendering: Yes
stoned: !yay!
stoned: \o/
stoned: |
stoned: /\
stoned: its rather slow
stoned: but at least it works
stoned: now i can check out some of these happy penguin games