pdowling: hi all
impostor: I just installed the radeonHD driver from fedora's development repository. i then enabled it in xorg.conf (by changing the Driver entry to "radeonhd"). However, when I restart the X server, it fails, leaving me with a command shell. My /var/log/Xorg.0.log reads: "(EE) Failed to load module '"radeonhd" (module does not exist, 0)". Any help with this?
ndim: impostor: The rawhide package?
bunk: hi, anyone here who could help me getting DVI output on my 690G working (analog works fine)?
egbert: bunk: which board do you have?
bunk: Gigabyte GA-MA69GM-S2H
egbert: ok. that's the 'other' one.
egbert: bunk: we've got a report on the asus m2a-vm.
egbert: where dvi isn't working atm.
egbert: i've got such a board - sticking it into a case atm.
bunk: this asus would have been my next choice...
egbert: bunk: what problem do you see?
egbert: bunk: i would have preferred the gigabyte. it doesn't come with a riser card for hdmi.
bunk: no picture (sometimes something like a quarter of the background picture fullscreen for half a second and immediately disappearing)
egbert: this way i could have fit everything into a slimline case.
bunk: I don't care about hdmi, but the asus doesn't look good for using a big and silent CPU cooler
egbert: bunk: ok, this could be related to electrical values on the dvi line. i'm going to look into it.
bunk: thanks :)
egbert: bunk, please watch the commits and ping me if you see something go by which looks like a fix for your problem and let me know if it works for you.
bunk: egbert: which commit? in xf86-video-radeonhd?
bunk: egbert: BTW: logfile is at http://www.fs.tum.de/~bunk/tmp/Xorg.0.log
egbert: bunk: yeah.
bunk: that was with the 1.1.0 from Debian unstable, but latest git from saturday had the same problem
egbert: bunk: ok. i'm curious what i will see on the asus.
bunk: commit bef7f09c6de1b4cf2192dbc46bddb9634c0bd058 is funny, I can imagine the problem :)
egbert: bunk: yeah, people thought we can guess from the log what they can't see :p
plectrum: egbert: ping
egbert: plectrum: what can i do for you?
plectrum: you don't have time now to debug VGASave right?
egbert: ah, ok, if people use a non-telling nick and don't identify themselves :pp
egbert: plectrum: i've sent you email. i need to run rather soonish.
plectrum: ok just checking
egbert: i can't promise but i can try to ping you later today.
plectrum: sounds good
gusta1: is there a plan to release 1.2 any time soon, or are you waiting for 3d-stuff?
egbert: gusta1: i'd assume 1.2 will be before 3d.
egbert: we are still looking into rv620/635.
gusta1: 1.0: Nov 30, 1.1: Dec 21, so 1.2 should be Jan 12 ;)
egbert: gusta1: there was xmas in between, this slowed us down :p
gusta1: egbert: oh, yes I forgot about that ;)
gusta1: i'm really waiting to try the acceleration..
udovdh: on rv630 :-)
udovdh: bunk, DVI_D works fine on this gigabyte hd2600pro
bunk: udovdh: it's always nice to hear that it works for other people when something doesn't work for you...
udovdh: well, I had some small trouble too
udovdh: so it might be around the corner for you
udovdh: or you could get a cheap hd2600
udovdh: I had to use xrandr to make the analog -> DVi switch
udovdh: without rebooting
udovdh: because there was no signal at first
bunk: udovdh: I don't want to stick a card into the board
bunk: X says:
udovdh: ok... just wondering what off-chip thing could be the cause of your issue
bunk: (II) RADEONHD(0): Output VGA_1 disconnected
bunk: (II) RADEONHD(0): Output TV_SVIDEO disconnected
bunk: (II) RADEONHD(0): Output DVI-D_1 connected
bunk: (II) RADEONHD(0): Output DVI-D_1 using initial mode 1920x1200
bunk: (II) RADEONHD(0): RandR 1.2 support enabled
udovdh: but no picture, right?
udovdh: did you try a lower resolution?
udovdh: here 1680x0150 works
bunk: I once tried 1600x1200 without success
udovdh: is there a resolution limit for dvi?
udovdh: where dual-link should be used?
udovdh: I am no expert in that area
udovdh: bunk, did you send logs to the devs?
udovdh: `X -logverbose 7`
udovdh: address is in the log
bunk: 1920x1200 should be no problem with single link DVI
CornedBee: where do you devs get your r500 information from, by the way? is there an info dump from ATI on that somewhere?
bunk: I told egbert the URL to my Xorg.0.log, and when he has further questions I'm sure he'll ask me here
udovdh: CornedBee, amd gives the info sometimes
CornedBee: yeah, but is there a specific document about R5xx chips published?
udovdh: hmmm. ask the devs, I saw a link pass by here some days ago
michaellarabel: Link for specs? http://www.x.org/docs/AMD/
Goga777: is there any news about opening specs for hardware video acceleration - avivo HD and uvd ?
udovdh: uvd etc is the hard part
udovdh: opening up some docs
michaellarabel: Opening R600 UVD right now is a likely a "no" since it can't be opened without putting DRM at risk. R500 video though should be possible.
udovdh: without giving out enough info for breaking drm
Goga777: DRM - is it base problem for spec's opening ?
michaellarabel: For opening up the video side that is a problem, though the base problem for everything is AMD just having enough time to sanitize/complete the documentation and get cleared by legal.
marcheu: agd5f: "AMD GPG" ? :)
michaellarabel: GPG == Graphics Product Group
marcheu: oh, I thought it should be GPU
marcheu: but so be it, GPG :)
soc: hi libv ..
wdah3: hello... what is tcore, please?
michaellarabel: Tcore is basically what AMD software engineers use internally before the silicon for new GPUs is available.
wdah3: michaellarabel: so it's some kind of programming interface for GPUs?
sytse: mm, my wincfg just crashed my X server
CornedBee: wdah3: no, more of an emulator, I think
wdah3: aye, thanks
egbert: bunk: i still need to install my rs690 box.
egbert: bunk: 1920x1200 is ok for single link using reduced blanking.
egbert: there seems to be another problem which needs to be tracked down.
gusta1: seeing megacorp's "going open" gets increasingly sad. after half a year, AMD releases the docs. again? now own their own web site... "i have my own web site now" like '95...
gusta1: i certainly hope there's a reason behind this. but when application development is hard to set a time table for, it seems even harder for legal-staff..
soc: anything interesting released lately?
gusta1: yes, gpu docs
soc: which ones?
gusta1: the half-year-old ones
gusta1: are relased today
gusta1: well, at least today is when their webmaster put them online ;)
Temujin_: the world of open-source gpu development is insane. it's got black holes and time warps it seems
gusta1: yes, just like the real world
gusta1: oops *universe* i mean
CornedBee: what docs were that?
gusta1: http://www.x.org/docs/AMD/ those ones
gusta1: are here http://ati.amd.com/developer/open_gpu_documentation.html now
CornedBee: i see
gusta1: yeah, it's amazingly interesting
gusta1: I almost can't go to bed now when some random megacorp put some random already-released documents on my web page
gusta1: stops irony and prepares for sleep
gusta1: on *their*
Honk: You SHOULD be concerned about what you first said.
Honk: That'd a bad security problem :]
CornedBee: on the plus side, it's at least explicit that the information in the M56 document applies to all R5xxs.
CornedBee: That answers a question I had earlier today.
agd5f: tcore is basically a userspace driver suite for testing hw: things like the command submission, asic init, etc. it can run on real hw or simulators.
KOPRajs: hi, could someone please get me clean answer on if there is any 3D support for RV535 chip in either radeon or radeonhd opensource drivers?
agd5f: KOPRajs: not yet
KOPRajs: neither in r300_dri nor in radeonhd?
KOPRajs: ok, thank you very much
marcheu: I was thinking, maybe there should be a global page on fd.o somewhere, with all our drivers and what they can do respectively
marcheu: I have a feeling that 90% of the end user questions revolve around that
agd5f: marcheu: yeah
KOPRajs: I found many info on the web, but some of them was confusing and many were outdated
CornedBee: there are end users of radeonhd? :)
egbert: CornedBee: ?
CornedBee: I mean, does anyone beyond those interested in its development actually use the radeonhd driver?
marcheu: I think so, according to phoronix quite a bit of people indeed
KOPRajs: ok, so thank you for a finally clean answer :)
CornedBee: well, that's cool
CornedBee: didn't expect it
marcheu: IIRC they have more people than nouveau for example :)
marcheu: (both being under development)
egbert: CornedBee: yeah, there should already be a bunch of people. from the traffic we get.
KOPRajs: and good luck with the driver so I can get rid of fglrx soon on my X1650 :)
KOPRajs: "soon" :)
gusta1: I got rid of crapware fglrx months ago
egbert: of course it is hard to distinquisch between 'normal users' and those interested in development.
gusta1: or you might need 3d...
gusta1: egbert: or those who're just sick of fglrx (99%)
KOPRajs: I do need 3D :)
KOPRajs: for playing NWN e.g. ;)
CornedBee: speaking of which
gusta1: lots of people "need" compiz too
CornedBee: something I'd like to see is along the lines of
egbert: gusta1: fglrx can do more. but sometimes you have to pay a price.
CornedBee: "What can I do to help out?"
CornedBee: beyond using and reporting problems ...
CornedBee: gusta1: like the damn thing totally not working at all
gusta1: egbert: yeah, and if that price is higher than no 3d and no xvideo, then the switch is simple ;)
KOPRajs: actually I got very poor XV with fglrx
CornedBee: yeah, me too
gusta1: I could see movies on my 1280x1024 monitor with no probs (ranging from crap videos 300x200 to HD). now, I have to watch any video on my 1024x768 monitor.
CornedBee: effectively unusable
gusta1: and I'm on a C2D
CornedBee: I mourn my old 9800
gusta1: actually, things works almost better when showing videos in fullscreen in Media Player Classic on WinXP in VirtualBox. that's scary as hell
bunk: gusta1: radeonhd + mplayer + scaling videos to 1920x1200 works fine for me
gusta1: uh ok..
gusta1: you on super-optimized gentoo or somthing?
gusta1: I use debian sid and a 2.13 GHz C2D
bunk: Debian unstable with 2.4 Ghz single core Athlon
gusta1: that's weird..
gusta1: radeonhd 1.1.0?
CornedBee: x11 driver in mplayer?
gusta1: and swscaler
CornedBee: haven't tested video on this box yet
bunk: gusta1: 1.1.0 and now current git
gusta1: it's as sweet as a cactus
bunk: CornedBee: yes, with -zoom
sytse: that's odd
gusta1: bunk: and it worked well with 1.1? I assume current git is faster with EXA/XAA optimizations
sytse: wine crashes my X server when I have XAA accel disabled (git master)
bunk: I can't swear, but Ithink my first try was with 1.1.0
sytse: but with both EXA (which sadly makes my rxvt-unicode unsufferably slow) and ShadowFB it doesn't crash
bunk: (but I might be wrong with that)
gusta1: i think it might have been with current git, bunk. at least I hope that does the trick
gusta1: after all, your res is 75% higer than mine, and 2.4Ghz Athlon doesn't justify the lead over 2.13Ghz C2D
bunk: it makes mplayer a CPU hog (between 60% and 70%), but that's not a problem
gusta1: well, maybe the swscaler in mplayer is pro-amd, who knows
gusta1: 1024x768 makes my mplayer a ~100% cpu hog, and any res higher than that, obviously lags (de-syncs)
gusta1: I hope radeonhd 1.2 is released soon (with the optimizations)...
sytse: oh, that's precious.. when disconnecting my monitor and running wine, the X server crashes, but with my monitor connected it works
xAFFE: I'm also using radeonhd and mplayer just plays everything fine :). my current res is 1650x1080. I should also note when watching fullscreen, the load is gointo to 2-3, but also depends on the vid and how it is encoded.
marcheu: gusta1: the optimization you're calling for is basically an implementation of Xv, which requires the 3D engine on those cards
gusta1: I thought the opt's in git were blit opts which, theoretically, could improve the swscaler in mplayer, but maybe i'm just plain wrong
marcheu: well, they can only help so much
soc: development of 2.6.25, future of the x-server: http://www.heise.de/newsticker/meldung/103282 (german)
gusta1: I need 1024x768 -> 1280x1024 speedup :P
gusta1: xAFFE: you using git or 1.1.0?
sytse: xAFFE: works fine here, too, 1364x1024, -vo x11 -zoom
xAFFE: gusta1: git
gusta1: sytse: git or 1.1?
sytse: gusta1: git master
sytse: current git master
gusta1: something tells me git is quite a _lot_ faster than 1.1
gusta1: sytse: thanks
sytse: gusta1: I'll test if XAA actually works with a monitor attached ;-)
gusta1: heh, yeah that's a nice bonus
sytse: yup, works
sytse: it must call some randr function or something like that, that crashes the X server when no monitor is attached
gusta1: btw, is there a maximum resolution in X or in radeonhd?
sytse: but that's the DVI maximum resolution, it's a hardware limitation
gusta1: i.e.? I'm thinking of buying two 1600x1200 monitors to use side-by-side, which ends up with 3200x1200...
sytse: and radeonhd also has a fixed max resolution, but I guess that was simply a safeguard, not a real limitation iirc
gusta1: but if that's beyond a limit then such a buy isn't very clever
sytse: oh, it should do that -easily-, but check the ml :)
gusta1: ok i will
gusta1: maybe my card doesn't... hmm..
sytse: 3200x1200? That's just 1600x1200 twice, and it supports that..
Zhenech: should work, as long its not ONE screen *g*
gusta1: uh, checking some .xls from sapphiretech.com for my x1650 and it says: "3D Graphics Resolution: 2560x1600"
gusta1: it's one screen, but two monitors
gusta1: "big desktop" / "xinerama" / [insert name here]
gusta1: hmm maybe that was for one output... i'll check this out tomorrow.. nite all.
Zhenech: i meant one monitor
gusta1: though so ;)
Zhenech: the tech foo should be for one physical monitor too
gusta1: but can you use 3D on two monitors/outputs?
gusta1: i.e. one "big 3d"?
Zhenech: uhm, /me has a radeon 7000
Zhenech: and I can use 3d on both screens, if I use 16bit colors
Zhenech: (16mb memory is just TO small for more)
gusta1: ah.. but that's pretty cool..
Zhenech: and I'm on radeon, not radeonhd (which does not support 3d [yet])
gusta1: right. i'm on readonhd because I finally made it work with my dual head settings (old non-edid monitor and stuff)
gusta1: well, gotta sleep. have meeting with the tax authorities tomorrow, wohoo... nightie.
Zhenech: tax, uhuhuh
egbert: sytse: did you make a bugzilla entry about your problem with X crashing under wine?
egbert: sytse: if not could you do this providing a log file after the crash? (this should already contain some backtrace information).
egbert: sytse: since it happens when no monitor is connected wine seems to use xrandr to set a mode.
egbert: this is the only time when an unplugged monitor actually gets detected.
CornedBee: well, just tried video
CornedBee: works fine for a 624x336 xvid
CornedBee: but only unscaled
CornedBee: xine starts framedropping and mplayer lags when trying to scale up to 1280 width
CornedBee: Athlon64 @ 2 GHz
sytse: egbert: not yet
sytse: egbert: I'll investigate it tomorrow, when I have more time
sytse: egbert: the reason I'm using a non-connected monitor is because it fails to be detected because of my failure to add some 75 ohm resistors ;-)