rx__: hmm.. changelog? :)
airlied: rx__: announce mail ..
rx__: ah there it is
rx__: really big changelog too
rx__: wonders what awaits the 6.8.x series
airlied: rx__: bugfixes to 6.8 :)
rx__: are you going to fosdem?
airlied: rx__: nah too far..
airlied: rx__: my gf sets guilt limits on conf attendance ;-)
rx__: i'm really curious about amd's talk and glisse's as well
arekm: damn, 6.8 broken for me :-/ bad dpi
arekm: (II) RADEON(0): Setting screen physical size to 444 x 277
airlied: arekm: is your monitor lying?
arekm: airlied: no, it's thinkpad, remember adding support for dcc type 6?
arekm: -(II) RADEON(0): Output: LVDS, Detected Monitor Type: 0
arekm: -(II) RADEON(0): Detected non-DDC Monitor Type: 2
arekm: huh, previously it used EDID
arekm: carme.pld-linux.org/~arekm/xorg-diff-1.txt diff between previous (167) and current (6.8) driver
airlied: arekm: oops I wonder did we break ddc type 6..
airlied: agd5f: di you break type 6 ;-)
arekm: fix welcome ;)
bgoglin_: agd5f: did you see the git-bisect at the end of http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=465864 ?
arekm: do_wait: drmWaitVBlank returned -1, IRQs don't seem to be working correctly.
arekm: Try running with LIBGL_THROTTLE_REFRESH and LIBL_SYNC_REFRESH unset.
arekm: uh. these are not set anyway
MrCooper: arekm: it could also be related to the driconf setting vblank_mode, or happen just once on app startup
arekm: MrCooper: thanks, setting 0 via .drirc makes app run (well, a bit longer and then locks GPU probably)
MrCooper: which app?
arekm: google earth
MrCooper: k, unfortunately I can't test that easily due to lack of powerpc binaries
airlied: Temujin_: hey.. so rs690 is beta quality, as in I've run glxgears and openarena..
Temujin_: anyone try the new 6.8 driver with rs690/X1200 yet?
airlied: it should support the outputs fine..
Temujin_: airlied, is the DVI 2D stuff working okay?
Temujin_: ok, you read my mind
airlied: Temujin_: I haven't tested mine yet..
Temujin_: the last time i tried this driver, it was on Ubuntu and it was a pain building new mesa, drm, etc.
Temujin_: it also appeared that 2D accel wasn't working
Temujin_: but I'm using Arch now, so I'll give it a shot when they update
Temujin_: thanks for the update
airlied: Temujin_: yeah its a bit of a pain to update.. 2d accel works better if you have kernel support..
airlied: but I'm trying to merge that at the moment.
Temujin_: yeah, I was using a custom 2.6.24 kernel on ubuntu
Temujin_: hell, maybe i'll build it from source later today, arch makes it pretty easy
Tigerchen: is radeon with this release able to drive a x1400 (M54)
MrCooper: should be
Tigerchen: thanks I'll try it
Tigerchen: seems to work.
agd5f: bgoglin_: does this patch help: http://www.botchco.com/alex/xorg/debian465864.diff
bgoglin_: agd5f: I'll ask the submitter, thanks
agd5f: bgoglin_: also, can he attach a full log if possible?
agd5f: airlied, arekm: could be
quicksilver: I see new release of open source ati driver on phoronix, whoop
quicksilver: there is some stuff about TV out in the changelog
quicksilver: does it now support TV out on those cards which previously needed the gatos patch?
orkid: which version?
agd5f: has for ages. no theatre support though
agd5f: arekm: send me your bios and I'll take a look
agd5f: I added code to use the bios provided masks for lcd ddc types, but perhaps those are wrong...
PSYCHO___: question: X700 and EXA are friends now?
agd5f: PSYCHO___: what do you mean? exa generally sucks performance-wise regardless of the hw
PSYCHO___: ah that explains a lot
PSYCHO___: ok i won't use exa
PSYCHO___: so another problem :)
PSYCHO___: if I don't use EXA but XAA after suspend/resume first Ogl app hang Xsession
PSYCHO___: no log or trace left
PSYCHO___: it only consumes 100% cpu
PSYCHO___: even don't leak
glisse: agd5f: do we restore the bit i founded years ago ?
agd5f: glisse: what bit?
glisse: agd5f: one bit un clock iirc which if not set lead to easy lockup on r300
agd5f: oh, yeah that should be there still
agd5f: although I think that gets set in screeninit, so it may need to be set in the resume path as well
glisse: just wondering if it was restored from suspend/resume, even though this isn't a magic bit which fix all lockup :)
agd5f: PSYCHO___ and anyone else having resume gl hangs: http://www.botchco.com/alex/xorg/restore_dynclocks_on_VT_switch.diff
PSYCHO___: why this is not in main tree if it fix that?
agd5f: PSYCHO___: I don't know if it will fix it
agd5f: just a guess
PSYCHO___: so i have to modify my nice ebuild :D
quicksilver: agd5f: great news.
quicksilver: agd5f: excuse stupid question, but what's theatre support?
agd5f: external tv encoders. we only support integrated tv-out right now
quicksilver: cool, that's fine.
quicksilver: I have a 9500 (I think?) which I've been using gatos tv-out on for ages.
agd5f: quicksilver: should work
quicksilver: how recent an Xorg installation would I need?
agd5f: at least xserver 1.3 and ati 6.8.0 for pal, 6.7.19x for ntsc
PSYCHO___: how the fun begins, reboot, hibernate, resume soo long :)
agd5f: PSYCHO___: good luck :)
PSYCHO___: oh good news
PSYCHO___: patch worked, my new kernel worked and new alsa is also working ;)
agd5f: PSYCHO___: cool
agd5f: thanks glisse :)
PSYCHO___: oh yeah, i don't remember when i last time had this kind of luck
sdw: anyone available to give a hand with getting an Xpress 200 working ?
sdw: the Xorg.0.log for starting up is at http://pastebin.com/m52baed3f
PSYCHO___: now i am little worried what is going to broken for ballancing this :P
sdw: i've trimmed out some of the things that i don't think matter
umtc: hello MrCooper, i'm trying out the new release right now
sdw: it looks like my Xpress 200 can't get the modes from an Acer AL1916W by DDC (a GeForce 7600 GT in the same machine with the same X installation can get them by DDC). any suggestions for what to look for ?
umtc: xv on zaphod mode is broken, but that's a known bug and i'll keep waiting for a fix
bgoglin_: sdw: is your monitor plugged on the VGA output ?
umtc: crash when moving the mouse to the right side of the primary screen is still present, and the screen stays in power off mode after quitting x. i'll wait for the next release
sdw: bgoglin_: yeah. there's only one port and it's connected via a VGA cable.
sdw: the drivers seems to think there are two ports, though, unless i'm misreading
bgoglin_: vga doesn't work well on xpress 200 iirc
sdw: odd. it worked before under Debian Lenny, but i didn't think to note which version of the driver.
sdw: i've just re-installed Slackware (12.0) and that's when i started having the trouble
bgoglin_: debian lenny has 6.6.3, it's very old, may work better in some cases but it is generally much better to use 6.8.0
sdw: Slackware 12.0 has 6.9.0, apparently
bgoglin: that's Xorg version
sdw: it's radeon module version 4.2.0, according to the X.0.log
sdw: ah, sorry. which version did you mean ?
bgoglin: the xf86-video-ati driver version
sdw: ok. i'll see if i can find that
sdw: ah. that's 6.6.192
sdw: i'll get 6.8.0, then. thanks.
agd5f: ddc on XPRESS chips doesn't work reliably, I'm trying to track down why, but I haven't had much luck yet
agd5f: umtc: known bug
sdw: agd5f: ok, thanks. i've just downloaded 6.8.0, so i'll let you know whether that's better than 6.6.192
agd5f: sdw: if you have the log where ddc worked, let me know
sdw: agd5f: will do. thanks. are you usally in here or on freenode ?
agd5f: sdw: I'm usually around
arekm: agd5f: heh, bios third time? ;-))
arekm: agd5f: 197 was of course working fine
agd5f: arekm: sorry, I can't remember which bios image is yours :)
arekm: agd5f: http://carme.pld-linux.org/~arekm/thinkpad-z60m-radeon-mobility-x600.rom
osiris_: agd5f: hmm, looks like I don't get true color on radeon (even though Xorg.0.log says it's true color and I set DefaultDepth 24 in xorg.conf)
osiris_: any suggestions?
agd5f: osiris_: not sure
MrCooper: osiris_: define 'not true color'
osiris_: MrCooper: dithering is easily visible
osiris_: MrCooper: with fglrx I have no problems
agd5f: osiris_: probably something we aren't setting correctly. can you get a reg dump of radeon vs fglrx using radeondump?
sdw: agd5f: are you around now ?
agd5f: sdw: sort of
sdw: i just wanted to let you know that the log of an Xpress 200 working with version 6.8.0 is at http://www.no-dns-yet.org.uk/Xorg.0.log
sdw: feel free to grab it at your leisure
sdw: that's DCC, by the way
agd5f: sdw: so it works with 6.8.0?
sdw: yeah. well, DDC does. i'll make sure resolutions are set up correctly when i get to work, but at least X starts and the Xpress talks DDC to the Acer AL1916W
agd5f: cool. maybe my i2c rework did the trick
sdw: possibly. i saw lots in the changelog about DDC. if you want me to try a version pre-i2c-rework, let me know which one
sdw: and post-i2c-rework
MrCooper: is off for tonight, bbl
Goga777: is there the future for video HD acceleration in radeon ?
rx__: what video hd acceleration exactly?
Goga777: I mean the support avivo HD and uvd in radeon for HD series cards
rx__: UVD is still up in the air for HD series cards
Goga777: why ?
sdw: agd5f: i'm going to disappear now, so you can drop an email to firstname.lastname@example.org if you want any testing done to see what fixed DDC
rx__: digital rights management
Goga777: no any HDTV video decoding in open source due to the DRM ?
agd5f: Goga777: drm is tied up in certain parts of the uvd decoding hw
agd5f: however, we should be able to implement some of the decoding using shaders
Goga777: will you work in parallel with radeonHD ?
PSYCHO___: about that resume lock, now it hangs gl app if runs littlebit later
agd5f: PSYCHO___: :(
PSYCHO___: i told it to my cat and she agrees it is really sad
agd5f: Goga777: the 3D stuff will be common
Goga777: is there all necessary specifications for 3D from amd ?
rx__: there are.. inside AMD ;)
rx__: there is no public documentation yet
rx__: it should come out 1Q08
Goga777: but you can continue to work under 3D implementation ?
rx__: legal issues aside i think agd5f can work on 3d implementation
libv: rx__: uvd is not up in the air, i think it is something we can all just write off already
osiris_: agd5f: radeondump hangs my machine (sysrq doesn't work)
rx__: well that saves the suspense ;)
agd5f: osiris_: :(
agd5f: Goga777: I'm working on the 3D documentation now actually :)
Goga777: good luck :)
agd5f: Goga777: there are two parts to video acceleration, decode and render. the drm uvd stuff represents just a small part of that
agd5f: I think we'll be able to get info on idct/mc out and with textures ans shaders we can implement a pretty decent support
rx__: that's good news
rx__: i was thinking a few days ago.. why is radeonhd called radeonhd if you can't use the hd part of the hw ;)
Tigerchen: hi. just wanted to let you know that radeonhd slightly outperforms radeon atm on x1400
Tigerchen: just feels a bit more fluid
rx__: outperforms in 2d?
agd5f: Tigerchen: using accel or shadowfb?
PSYCHO___: i just run opengl game
PSYCHO___: that turned of my external lcd
Tigerchen: agd5f: whatever is default
PSYCHO___: and after i quited that game on external i showed that what is on LVDS
PSYCHO___: but when i move mouse there it works like external one
rx__: if you used drm accel then it would have performed equally
agd5f: PSYCHO___: the game probably uses xvidmode to change the mode which isn't randr 1.2 aware
rx__: it's been known that shadowfb has a slight advantage because it doesn't have to do an extra copy
PSYCHO___: any way to fix this
PSYCHO___: not wanna restart xorg
agd5f: shadowfb is probably faster since 2D accel doesn't do much for modern desktops
agd5f: you really need full accel or none, any fallbacks to software kill performance
rx__: maybe i'm getting something confused :)
Tigerchen: rx__: well default is xaa on radeonhd i think
Tigerchen: but it wasn't measurable with glxgears or sth
rx__: you need r500-support drm branch
rx__: i don't think that was ever merged into master
Tigerchen: just that radeon fealt a bit .... lazy
rx__: yea.. you need the mesa/drm r500-support branch to take advantage of xaa
rx__: wonders why it hasn't been merged yet
Tigerchen: rx__: i knew there was something, like back when i user r300 from cvs with my last notebook
Tigerchen: wonders why he always has to build stuff from cvs/svn/git....
rx__: because you know too much ;)
agd5f: rx__: it has been merged
agd5f: it's in drm master and queued for kernel
agd5f: merged ages ago
rx__: checks log again
rx__: Merge branch 'master' into r500-support
rx__: Merge branch 'r500-support'
rx__: one of the first few commits of the year.. how did i miss that ;)
Tigerchen: hangover from new year
rx__: good point
Tigerchen: ok so I'm up to build drm from git... again
rx__: yeah master should suffice :)
Tigerchen: wow is git fast atm.... 35KiB/s
arekm: agd5f: have you looked at my rom?
Tigerchen: and i think radeon might leave the card in some "dirty" state. when i killed my x and started it again with radeonhd i noticed a total distortion of my display which doesn't happen on a clean boot directly into x with radeonhd
agd5f: Tigerchen: probably some element of tiling
agd5f: arekm: not yet, haven't had a chance
Tigerchen: agd5f: i don't know just remembered it and wanted to let you know
agd5f: Tigerchen: does "Option" "ColorTiling" "False" help?
Tigerchen: agd5f: can't test just yet, git-checkout of drm takes some time
Tigerchen: eceiving objects: 84% (127831/150540), 39.77 MiB | 13 KiB/s
Tigerchen: rx__: well i built drm and mesa from git and restarted x and now i get unbelievable 300 fps in glxgears (1k before)
Tigerchen: but /dev/dri/card0 exists now
PSYCHO___: glxgears is not benchmark
Tigerchen: i know
PSYCHO___: try some game
Tigerchen: but a nice rough estimate
PSYCHO___: or video
PSYCHO___: wouldn't say so
PSYCHO___: i got 500
PSYCHO___: and play doom3
rx__: good ;)
orkid: ot but are there any FPS games out there that have been made available for free?
orkid: there was a big uproar about quake1 but it seems like you still need the game (for the wads)
PSYCHO___: now should be appropriate to say that i first reboot to win xp
PSYCHO___: just kidding
rx__: orkid; nexuiz
orkid: yup those are open source, but any that are closed/free?
rx__: america's army
orkid: like older commercial games?
rx__: check the list
PSYCHO___: i prefer to search in portage, so I mustn't make my own ebuild :P
PSYCHO___: so lazy
orkid: ok didn't notice tx
PSYCHO___: fill bug write ebuild close bug
agd5f: drm support does not equal 3D support
rx__: what work is there to be done for 3d docs?
airlied: rx__: cleaning.. legal ppl..
rx__: agd5f speaks legalese? ;)
rx__: oh cleaning
agd5f: rx__: well there is no programming guide :)
agd5f: I'm putting it together
agd5f: that's part of why it's taking so long, legal stuff aside
rx__: looks forward to this 'writing a radeon driver for dummies' 10,000page ebook
Tigerchen: another bug: closing lid, reopening, funny blinking whatever-stuff, switching to text and back to x restores it
rx__: funny blinking whatever-stuff? eh?
rx__: does it turn to white?
Tigerchen: looks like very strange chars the size about the chars in no-fb-console
Tigerchen: pinkysh and yellow and striped and blinking
Tigerchen: i made a screenie (that doesn't really show what i see but shows something....
Tigerchen: *waiting for imageshack*
Tigerchen: reality does look even better
Tigerchen: but i don't have the nerve to get my digicam atm
agd5f: Tigerchen: probably the bios is arbitrarily programming the hw when you hit the lid switch, I attempted to fix that, but I don't have an atombios based laptop
airlied: does.. but no idea about lid switch..
dli: I tried gtkperf benchmarking, noticed GtkDrawingArea - Text is several times slower than http://www.phoronix.com/scan.php?page=article&item=981&num=3
airlied: dli: its all done with CPU.. so if your CPU is worse etc..
dli: airlied, no, other numbers are all faster, so is overall time
airlied: dli: so some ops are done with gpu might be faster, some with cpu might be slower wetc..
dli: airlied, let me try Option "AccelMethod" "XAA", it's EXA now
agd5f: arekm: ddc types 5 and 6 fixed
mo0n_sniper: what is the state of a radeon xpress mobility x1150 with the new driver version?does it has 3d?
damentz: agd5f: exa has more cpu overhead on 3d applications right?
damentz: or have you guys solved that?
agd5f: damentz: ?
damentz: well i mean, is exa architectually slower when it comes to expensive 3d rendering?
damentz: i tried warsow with exa and xaa
damentz: xaa seems to be more smooth, otherwise the fps is the same
damentz: but i think its just the slow cpu i have on the laptop
agd5f: damentz: exa has nothing to do with 3D, it's 2D and render only
damentz: so using exa instead of accel should not effect 3d?
agd5f: for 2D stuff ye, XAA is faster than EXA
damentz: xaa is faster than exa with 2d?
damentz: wait i think ou got that reversed
damentz: anyway, i'm seeing if i can get exa + compiz on this old card
damentz: it runs fine on my radeon 9600, and infact has no performance regression with compiz fusion running
damentz: but on my laptop, i get artifacts around in the shadows of windows
rx__: for the most part xaa is better than exa
damentz: you mean for the moment?
damentz: i thought exa was going to completely replace xaa
damentz: for complexity issues or something
rx__: iirc, i think it has to do with taking advantage of the hw
rx__: so yeah for the moment
damentz: exa is already fast at 2d rendering on my laptop
damentz: though Xorg 7.0 had a very fast exa renderer
damentz: everything was wicked fast
damentz: then 7.1 came around and it was slow
damentz: and now its fast again now
agd5f: damentz: that's the plan, but it's not there yet