Radeon IRC Logs For 2008-3-15

rx__: oh nice
rx__: another doc release
simpsonc: So, I think I’ve got the first rough draft of the shader translator for r500.
simpsonc: All it does is set a handful of flags in the very first instruction to pass the pixels through unmodified, but that should be enough for glxgears.
simpsonc: I’m still made of fail when it comes to hooking the rest of it up, but if airlied was serious about helping me sometime next week then we may be in business before the end of spring break.
simpsonc: Mm. Sleep. Going to the birthday party tomorrow (today), so see you all in ~20 hours.
arekm: can audio be transmitted over dvi-d — hdmi cable?
arekm: I guess not but …
PSYCHO___: 01:05.0 VGA compatible controller: ATI Technologies Inc Radeon X1200 Series
PSYCHO___: this is x500 correct?
rx__: arekm; according to wikipedia hdmi can carry audio..
rx__: oh.. you mean a dvi-d – hdmi converter
rx__: nm
rx__: it should be x700 w/o shaders or something like that
rx__: crippled x700
icewaterman: PSYCHO___: no this is x1200 integrated chipset and will someday work with radeonhd
icewaterman: PSYCHO___: it is some x5xx chipset
icewaterman: though uhm, sorry r5xx
PSYCHO___: ah i also wrote x500 not r500 🙂
libvde: arekm: sure
libvde: arekm: because the cable doesn’t matter
libvde: arekm: it’s what on either side of the cable 🙂
libvde: arekm: you need an rs690 with hda enabled or an r6xx
libvde: plain r500 just doesn’t have the soundcard that feeds data into the TMDS signalling
edgecase: Backtrace:
edgecase: 0: /usr/bin/X(xf86SigHandler+0x7e) [0x80c780e]
edgecase: 1: [0xb7fbc420]
edgecase: 2: [(nil)]
edgecase: Fatal server error:
edgecase: Caught signal 11. Server aborting
edgecase: is gdb the next step?
icewaterman: hm, any idea, why i cannot use 1280×1024 even though both my graphics card and my monitor support it?
icewaterman: the proprietary driver detects and uses it
icewaterman: hm, also my x850 is extremely slow
fir3_: hi
fir3_: has amd released all the necessary documentation for the X*** series to write an opengl2.0/3.0 driver yet?
icewaterman: fir3_: even if that were the case (which afaik isnt), it would still be a long time for the drivers to be available (they have to be written after all)
icewaterman: fir3_: i suspect at least 2 years to pass until there is an open source opengl 2.0 driver for that series.
icewaterman: fir3_: thats why i bought a cheap radeon X850 which works way better but also has its issues
edgecase: i think a lot of docs are already released, and there’s rapid progress
icewaterman: any idea how i can compile the newer radeon releases for ubuntuß
icewaterman: configure complains about missing packages, that are unavailable on ubuntu or named differently
simpsonc: Birthday party, see you all later.
icewaterman: oh great, i upgraded to version 6.8 and now instead of the default picture size beeing to small it is now too big
icewaterman: now it exceeds the monitors capabilities
icewaterman: how can i fix the following message: no matching device section for instance (id of secondary port – which is unused atm)
Mad_Grenadier: I just got a Radeon 9800 Pro AGP card in my system (ubuntu 32bit) Is there anything special I need to do to install it or can I just go do the ATI site and download its linux drivers and run them in the terminal
Mad_Grenadier: (I only ask because of my bad experiences from my 8500 and cause of my low bandwidth)
icewaterman: Mad_Grenadier: ubuntu comes with proprietary drivers already
icewaterman: Mad_Grenadier: on most cards, the open source driver is having sever glitches
icewaterman: Mad_Grenadier: https://wiki.ubuntu.com/BinaryDriverHowto/ATI
Mad_Grenadier: Dont I need to download the newest ones from the site to play games and such (UT04,Doom3, ETQW, etc)
icewaterman: Mad_Grenadier: dunno
icewaterman: try the ones from ubuntu first
icewaterman: they are a lot easier to setup
Mad_Grenadier: k
Mad_Grenadier: Looks like they are working, is there anywhere I can go to tweak the settings and stuff?
edgecase: is git radeon driver compatible with radeon_fb ?
partymola: hello everyone
partymola: I am having an issue with a “Radeon XPRESS 200 Series” in FreeBSD, using the radeon driver
partymola: I have uploaded a video of it at: http://uma.frikiplan.com/radeon/
partymola: also the xorg.conf file I am using, a Xorg.0.log and the list of pci cards in my system (it’s a laptop)
partymola: berniyh: did you have the same problem i had… or not? :S
icewaterman: partymola: i wouldnt count on getting help here
icewaterman: tried it for days and 1 person asked for further logs, when i posted it, nothing but silence.
partymola: then where? O_o
edgecase: i would try your distro suport groups first
partymola: ok
edgecase: ubuntu is pretty good for that stuff
partymola: so bad i am using freebsd… lol
edgecase: oh i got you confused with someone else
icewaterman: partymola: i think you are just totally screwed 🙂
icewaterman: but if that helps, i am on ubuntu and cant get my radeon to work either
icewaterman: at least it might make you feel better 🙂
partymola: i got this working on Ubuntu with fglrx
partymola: but i don’t like it
partymola: the Xv support is very bad
partymola: and… well… i need the laptop to run on freebsd
icewaterman: partymola: my card works with fglxr as well, much better in fact
icewaterman: partymola: i am going to try getting the open source driver working today, but if that fails i’ll get back to fglrx
partymola: compared to nvidia’s driver… fglrx is the biggest piece of crap ever
edgecase: well radeon devs are working for a reason
edgecase: to get an opensource driver that’s better
icewaterman: partymola: at least it works, radeon driver is worse for all beyond 9200
icewaterman: edgecase: well, i admit not an easy task, but right now they are very far away from that
edgecase: icewaterman, what was your problem, it sounds like you need to update your xorg.conf file to the new 6.8
icewaterman: edgecase: no, my problem is exactly the same with 6.8
edgecase: i’m using all < 9200 so i like radeon OS driver
icewaterman: both 6.7.197 and 6.8 fail the same way
edgecase: config error about missing device section?
icewaterman: edgecase: well i used that one in the past and they worked quite well
berniyh: partymola: no, i haven’t seen that, but i experienced something similar, but for me it only happens sometimes (when i disable and enable the lvds with radeontool)
partymola: lvds? radeontool?
berniyh: lvds = internal lcd screen
edgecase: should use xrandr now yes?
berniyh: edgecase: if possible, yes
icewaterman: edgecase: (WW) RADEON: No matching Device section for instance (BusID PCI:1:0:1) found yes, i think thats because it wouldnt let me disable ddc and use the modelines i specify
edgecase: oh right, radeontool to turn off backlight
edgecase: icewaterman, pastebin your xorg.conf
edgecase: and lspci
berniyh: partymola: with radeontool you can do various things, like switching of the lvds backlight, switching of the external screen etc.
berniyh: partymola: is that really the xorg.conf, that you are using?
icewaterman: edgecase: http://rafb.net/p/7nJHbC22.html
icewaterman: the thing is, i can configure all i want, it will never disable ddc and use the modelines (now commented out)
berniyh: (!!) RADEON(0): MergedFB support has been removed and replaced with xrandr 1.2 support <– where does that come from, if you didn’t enable mergedfb in your xorg.conf @ partymola
partymola: wait!
partymola: i think i found the problem
partymola: (EE) RADEON(0): [dri] RADEONDRIGetVersion failed to open the DRM
icewaterman: basically my problem is that radeon driver tries to be smart (in a very stupid way)
berniyh: partymola: no, that only disables drm, you can ignore that, afaik
icewaterman: it detects no section for my monitor on the second head (there is none, because there is no such monitor) and then thinks: “hey, you didnt configure it, lets try ddc)
berniyh: partymola: this is what i’ve seen sometimes: http://www.berniyh.net/DSC00001.JPG
partymola: let me see
partymola: YES
partymola: but i get that always 🙁
werdz: berniyh: is that a laptop with (k)ubuntu?
werdz: after a suspend?
berniyh: werdz: no, gentoo + kde4 (svn)
werdz: (sorry i haven’t been readin ghte conversation)
werdz: ah, ok it looks like what the ubuntu resume scripts do to my x700
werdz: (disabling everything in them fixes it)
berniyh: and it isn’t related to suspend, it happens sometimes when i disable and enable lvds with radeontool
berniyh: and that way it will be fixed again
werdz: weird
partymola: berniyh: how you do remove the effect? playing with lvds settings while X are running?
berniyh: i think it also happened once after a forced dpms off
edgecase: icwaterman, why do you use BusID option?
berniyh: partymola: install radeontool (is that available on bsd?) and run “radeontool light off && radeontool light on”
berniyh: until it is fixed
partymola: ok, let me see
partymola: i have the tool
partymola: i found out that
icewaterman: edgecase: basically its a relic and because X –configure created it
partymola: if i run “radeontool stretch on”, X freezes
berniyh: hm
edgecase: icewaterman, if you only have 1 video card, you can comment that out
partymola: then i turn it off
partymola: and it unfreezes
icewaterman: edgecase: thing is, i have 1 card with 2 outputs
edgecase: doesn’t matter
icewaterman: i only wanna use 1 but xorg doesnt care
berniyh: partymola: then don’t use stretch 😉
edgecase: comment it out
icewaterman: edgecase: ok, i’ll try
edgecase: there’s a new xorg.conf format for radeon
partymola: berniyh: when i have it off, i see the ugly effect
partymola: having that on, there’s that prob
berniyh: oh
partymola: i mean, when having on, it freezes xD
partymola: completely stopped
berniyh: partymola: if it is really reproducable, you might want to contact one of the devs, maybe you hit a bug, that should be sold out
partymola: where do i find a devel? 🙂
berniyh: i had the problem, that it happened randomly and i didn’t find a reliable way to reproduce it
berniyh: i think airlied and agd5f are the devs here, but unsure, tbh
icewaterman: edgecase: as expected it didnt make a difference
partymola: hmmm
edgecase: pastebin the error msg
berniyh: (i mean, unsure if there are more ;))
partymola: it seems like the prob has something to do with stretching
partymola: the effect changes as i activate , deactivate, or turn only vert or horiz
icewaterman: edgecase: need to restart xorg first. wait a second
edgecase: error msg in /var/log/X.org.log.0 ?
berniyh: partymola: what about that? (!!) RADEON(0): MergedFB support has been removed and replaced with xrandr 1.2 support <– where does that come from, if you didn’t enable mergedfb in your xorg.conf @ partymola
partymola: i don0’t even know what MergedFB is LOL
icewaterman: edgecase: there is not much of an error message, just the resolution 1280×1024 doesnt work at all
berniyh: you really shouldn’t use mergedfb
edgecase: icewaterman, doesn’t work how, monitor shuts off?
edgecase: you could try Modes “1280×1024” etc
edgecase: don’t put the @60
partymola: wait, wait, i think I have found out something interesting
icewaterman: edgecase: it simply uses another resolution lower (either 1024×768 or 1280×768
edgecase: does the monitor support DDC ? unless it’s very old, it should
partymola: i think radeon detects my laptop screen as a tv? O_o
edgecase: i would remove these from Monitor section too:
edgecase: HorizSync 30.0-70.0 VertRefresh 55.0-87.0
edgecase: let DDC do it’s job
icewaterman: edgecase: tried already, doesnt change anything
berniyh: partymola: did you try to run X with the refreshrate settings commented out?
partymola: let me do it
berniyh: Section “Monitor”
berniyh: HorizSync 36-52
berniyh: VertRefresh 36-60
edgecase: Modes “1280×1024” “1024×768” “800×600” “640×480”
partymola: same problem
partymola: ok, gotta add those modes
icewaterman: edgecase: brb
edgecase: icewaterman, that’s the normal way
edgecase: partymola, is this an LCD?
partymola: native resolution for this screen in Vista is 1280×800 though
partymola: it’s an lcd, yes
edgecase: laptop?
partymola: yes
partymola: check : http://uma.frikiplan.com/radeon/startx.log
edgecase: i’d guess scaler problem
icewaterman: edgecase: ddc doesnt get the native resolution for my display (which is 1280×1024)
edgecase: partymola, something’s wrong, your DRM doesn’t start?
partymola: i have drm module loaded in the kernel…
edgecase: paste xorg.conf
edgecase: you need modules loading in there too
partymola: again
edgecase: i don’t have link to your old one
partymola: check : http://uma.frikiplan.com/radeon/xorg.conf << you can see what modules i am loading there
edgecase: try taking out the entire Module section
partymola: ok
edgecase: uh what OS you using?
partymola: freebsd 7.0
edgecase: what Xorg version?
partymola: loaded without drivers. same problem
edgecase: 1.4
partymola: 1.4.0
edgecase: same error about DRM?
edgecase: in any case, it looks like it’s not detecting that you’re using a laptop panel,
edgecase: so it’s not setting up the scaler
edgecase: which is required for internal LCD panels
partymola: (EE) RADEON(0): [dri] RADEONDRIGetVersion failed to open the DRM
partymola: [dri[ Disabling DRI.
partymola: etc.
edgecase: try using the BIOS bootup resolution, the scaler will already match that
icewaterman: edgecase: (II) RADEON(0): Not using default mode “1280×1024” (hsync out of range)
icewaterman: (II) RADEON(0): Not using default mode “1280×1024” (vrefresh out of range)
edgecase: or set bios to nozoom mode
icewaterman: those two probably prevent the mode from beeing used
edgecase: icewaterman, paste Xorg.log.o
edgecase: partymola, let me see latest Xorg.0.log
partymola: edgecase: have you noticed the errors (EE) RADEON(0): No devices found! , and “(EE) RADEON(0): Using VGA default” ??
partymola: finished output detect: 0
edgecase: yeah that’s what i’m saying, it can’t find LVDS info
partymola: maybe I can force the dectection of the LCD panel?
edgecase: to setup the scaler
partymola: 🙁
edgecase: possibly, paste whole Xorg.0.log
partymola: ok
partymola: gotta upload it right now
edgecase: some laptop bios, have an option to stretch the 80×24 text mode to fill the screen, or not
edgecase: try setting it opposite to current setting
icewaterman: edgecase: http://madebyme.homeunix.org/users/matthias/test/Xorg.0.log.txt
partymola: will take a look
partymola: but in any case, i think this bios doesn’t allow it 🙁
partymola: http://uma.frikiplan.com/radeon/Xorg.0.log << latest Xorg.0.log
edgecase: icewaterman, so what’s the issue again? you using widescreen monitor on DVI-0 output?
edgecase: might need to specify widescreen modes in Modes
icewaterman: edgecase: no, it is 4:3 but the resolution it is willing to give me is only 1280×768
icewaterman: radeon will not give me the native display resolution
airlied: icewaterman: do you have a monitor section?
icewaterman: airlied: of course
icewaterman: i have even the correct modelines from monitor manufacturer (as auto-detection didnt work out)
partymola: edgecase: i can’t change that setting in the BIOS, will try the Vista utility :/
airlied: icewaterman: so set the refresh rates in the moniotr section.
airlied: icewaterman: properly.. so they don’t limit the mode selection
icewaterman: icq
icewaterman: oops
icewaterman: airlied: the thing is they do NOT limit it
icewaterman: airlied: just radeon CLAIMS they do
airlied: icewaterman: have you worked out the maths then?
airlied: icewaterman: the code is generic in the X server now..
airlied: icewaterman: so isn’t just radeon code..
airlied: post your conf and log somewhere..
icewaterman: airlied: but radeon driver is the only one to fail on this setup, all other drivers work it out
airlied: icewaterman: all other drivers being fglrx?
icewaterman: http://madebyme.homeunix.org/users/matthias/test/Xorg.0.log.txt <- the log
edgecase: partymola, your Bios Connector Table seems to be missing the internal LCD LVDS port
icewaterman: airlied: well, even vesa
icewaterman: and fglrx
airlied: icewaterman: so you are connecting the monitor to tv-out?
icewaterman: vesafb also works fine (just console though). for now i have disabled it because it until xorg works
icewaterman: airlied: no
partymola: edgecase: so…. getting the terminal to have the native size would make it work?
airlied: icewaterman: so your conf file is wrong..
airlied: post it..
icewaterman: airlied: http://rafb.net/p/vMNhMI60.html
icewaterman: this is my current one
airlied: so in the Device add…
airlied: gimme a sec to dig it out.
edgecase: partymola, i have (II) RADEON(0): Port4: DDCType-0x0, DACType-0, TMDSType-0, ConnectorType-7
edgecase: (II) RADEON(0): Port2:
edgecase: Monitor — AUTO
edgecase: Connector — LVDS
edgecase: DAC Type — None
edgecase: TMDS Type — None
edgecase: DDC Type — 0x0
edgecase: partymola, if you don’t have an LVDS, laptop panel will have issues. i don’t know why you don’t have it tho
airlied: Option “monitor-DVI-0” “aticonfig-Device[0]”
icewaterman: airlied: all options that are commented in the device section i have already tried.
icewaterman: airlied: where?
partymola: laptop was cheap, but no that cheap LOL… I swear it comes with a screen LOL
airlied: icewaterman: in the device section
icewaterman: ok
airlied: icewaterman: use DVI-1 if the monitor is connected to the other port
icewaterman: airlied: i added that option with no effect, i’ll try dvi-1
edgecase: arlied, i think icewaterman’s not getting DDC detecting his monitor
icewaterman: airlied: Option “monitor-DVI-0” “aticonfig-Device[0]” <– what sense does that make?
icewaterman: i mean this way i tell the device that the monitor-dvi-0 is the device
icewaterman: i’ll try it with monitor instead
airlied: thats correct..
partymola: edgecase: i am going to be out for a hour or so (dinner time)
airlied: icewaterman: yes aticonfig-Monitor[0]
partymola: i will be back later… with the stretching disabled, i hope 😉
icewaterman: airlied: still nothing
edgecase: arlied, i think partymola’s LVDS panel in’t detected, no bios (II) RADEON(0): Panel ID string: 1024×768
airlied: icewaterman: can you post the new log file..
airlied: edgecase: I see that, wierd..
airlied: edgecase: the BIOS doesn’t report any LVDS>
edgecase: maybe his fbsd isn’t mapping VBIOS segment?
airlied: no it finds the BIOS alright..
icewaterman: airlied: the monitor does heavily flicker during xorg start
airlied: if I’m reading the right logfile..
edgecase: heh too many log files
edgecase: one customer at a time, please!
airlied: edgecase: it sounds like we don’t see his chip as a mobility..
airlied: which I’m not sure it is..
edgecase: oooh only mobility has LVDS out + scaler?
airlied: edgecase: yes, usually..
icewaterman: airlied: i have updated the logfile
edgecase: ah i wasn’t aware of that, good to know
airlied: edgecase: and for mobility we usually add an LVDS output if none exists..
edgecase: arlied, someone mentioned new docs, i’m curious what
icewaterman: airlied: http://madebyme.homeunix.org/users/matthias/test/Xorg.0.log
airlied: edgecase: just some r300 3D regs..
airlied: icewaterman: try making the hsync bigger
icewaterman: airlied: in the modeline?
airlied: I can understand vesa working because it ignores most things..
edgecase: arlied, i’m curious, will radeon_fb conflict with 6.8.0 or git master?
airlied: icewaterman: just for the monitor..
airlied: icewaterman: the HorizSync line..
airlied: edgecase: radeonfb may or may not.. its pot luck..
icewaterman: airlied: radeonfb doenst work at all on that card
edgecase: partymola, check out Option “LVDSProbePLL” “boolean”
airlied: edgecase: the main users for it are ppc machine..
airlied: partymola: when you come back are you building from git?
edgecase: ok just wondering if radeon driver explicitly dies when radeon_fb is loaded at start
edgecase: partymola, Option “DefaultConnectorTable” “boolean”
airlied: edgecase: nope it just tries to work, and maybe fails later 🙂
airlied: edgecase: I’ll see if I ccan generate a patch for him
icewaterman: airlied: airlied now it even tried 1280×1024 but at 85Hz instead of 60
edgecase: you mentioned a kernel modesetting driver for intel, but it’s not fbdev? seems like a lot of chips have modesetting code already in fbdevs
icewaterman: airlied: at least when i forced it with xrandr
edgecase: icewaterman, remind me again what your problem is?
icewaterman: edgecase: 1280×1024 (native resolution) will not work with radeon & my tft
edgecase: it’s not a widescreen?
icewaterman: edgecase: no
edgecase: what make/model is it? what connector, DVI or VGA?
icewaterman: it is 4:3
edgecase: ok
icewaterman: vga connector.
airlied: icewaterman: you have an 85mhz mode in the conf..
airlied: icewaterman: you might want to remove it
edgecase: why doesn’t his DDC detect the monitor?
airlied: edgecase: I’m going with busted cable 🙂
edgecase: ok, so set the Vreferesh/hsync from the manual of the monitor
airlied: edgecase: fbdev can’t handle multi-head very welll..
airlied: edgecase: at least not usefully enough for what we want to do with it..
airlied: edgecase: but kernel modesetting drivers also provide fbdev API on top
airlied: hmm that chip doesn’t appear to be a mobility…
airlied: I wonder how the connected the panel to it..
icewaterman: airlied: where again did you say i have the 85Hz configured for 1280×1024?
icewaterman: ah in vertrefresh
airlied: icewaterman: you have two modelines for 1280×1024 one of those I think is an 85Mhz mode.
icewaterman: airlied: well, i am back to square one, once i reduce vertrefresh to 55.0-60.0 it will try 1600×1200
icewaterman: airlied: the monitor is not capable of 1280×1024@85 and the modelines are from the manual
airlied: okay then finally add Option “PreferredMode” “1280×1024” to the monitor section
airlied: icewaterman: I’m guessing the manual is lying..
airlied: they just cut-n-pasted it from somewhere else.
icewaterman: the manufacturer offered a “driver” for the monitor
icewaterman: airlied: ok, maybe
airlied: the fact that EDID doesn’t work is really speaking against their manual 🙂
icewaterman: airlied: i have an xorg.log from the fglrx driver and will use the modelines it detected
edgecase: icewaterman, you should try and get vertrefresh and hsync from the manual
icewaterman: edgecase: the values were from the manual
icewaterman: and i had to increase them
edgecase: arlied, i did git-clone from anongit… how do i “cvs update” on it?
edgecase: what is the highest vrefresh in the manual, from any mode?
airlied: icewaterman: oh post the fglrx log
airlied: edgecase: git pull
edgecase: git-log shows: commit b35c09a597c93a1d9f06ef0091c96822b0653f98
edgecase: Author: Dave Airlie
edgecase: Date: Thu Mar 13 18:42:29 2008 +1000
edgecase: that up to date?
icewaterman: airlied: http://madebyme.homeunix.org/users/matthias/test/fglrx-xorg.log.txt
airlied: edgecase: yup
airlied: icewaterman: hmm wierd, they’ve limited the virtual size somehow to 1400×1050 which is why you don’t get 1600×1200 jumping up.
icewaterman: airlied: i tried that as well
icewaterman: didnt help though
airlied: icewaterman: but 1280×1024@60, mode with the 108 looks correct.
airlied: icewaterman: you had a 135.00 1280×1024 mode which is possible he 85Mhz one.
icewaterman: ModeLine “1280×1024” 108.0 1280 1328 1440 1688 1024 1025 1028 1066 +hsync +vsync this one is from the manual
airlied: icewaterman: that one + preferred mode and tell me what happens.. and/or post new log
icewaterman: airlied: well, what happens is easy, kdm now starts in 1600×1200
icewaterman: which the monitor cannot display
icewaterman: uhm, 1600×1024
icewaterman: monitor then runs in failsafe mode and shows a crappy picture
airlied: icewaterman: with preferred mode?
airlied: icewaterman: post that log..
icewaterman: airlied: yes
airlied: hmm I wonder have you the server that doesn’t support preferred mode.
icewaterman: airlied: i updated the log
icewaterman: airlied: i also added virtual setting and it simply seems to ignore it, while it honors it with fglrx
airlied: icewaterman: put the preferredmode in the monitor section
airlied: wierd it used 1280×1024 in that logfile
airlied: I wonder where that mode is coming from..
airlied: icewaterman: okay change the modename in the modeline to 1280x1024_60
airlied: and preferredmode in monitor section to 1280x1024_60
icewaterman: airlied: whoooohooooo
icewaterman: airlied: that worked
icewaterman: you are my hero 🙂
airlied: icewaterman: I hate monitors :).. but good it worked in the end
icewaterman: airlied: it is mysterious though, because now i have all the settings from the monitor manual and the preferred mode setting and it works fine. the _60 seem to do the trick (whatever appendix will do the trick)
airlied: icewaterman: for some reason the standard 1280×1024 mode was getting picked over the modeline one
icewaterman: ouch, 3d performance is factor 10 below the radeon 9700 pro i had before
icewaterman: glxgears works fine though
icewaterman: maybe there is still some fglrx stuff lying arround triggering software opengl or so
icewaterman: or the r480 is always slow with radeon driver
airlied: icewaterman: check glxinfo..
icewaterman: direct rendering: Yes
icewaterman: glxgears does 5000fps, i know thats not a benchmark
icewaterman: but counterstrike 1.5 on wine is at 4-7fps, my old radeon 9700pro did 40-60fps
partymola: edgecase: with LVDSProbePLL set as true and false, i get the same ffect
partymola: let’s try the DefaultConnectorTable
partymola: edgecase: tried DefaultConnectorTable true and false too, and no change
partymola: no changes 🙁
icewaterman: airlied: ok, at least it is not slow in general, ppracer was @ 120fps before it crashed the box 🙂
partymola: airlied: any clue on what I can do to get my laptop screen working right?
icewaterman: work, tremulous is finally playable with that card
icewaterman: 50-90fps
airlied: partymola: it looks like a wierd ass mcahine, I’ve no idea how the panel is connected to the chipset..
airlied: partymola: have you tried fglrx? if that works a log might be useful
partymola: yes
partymola: it works
partymola: … in linux, it did
airlied: partymola: a log from that might help..
partymola: ugh
partymola: that’s hard to get, you know
partymola: there’s no linux in this machine anymore
partymola: and reinstalling the thing … ugh
airlied: partymola: one of two things is happening.. either your chip is a mobility and we don’t know it is… or they’ve connected the panel via an external chip or TMDS
partymola: are you looking for something in particular?
edgecase: arlied, DVO -> LCD panel ?
airlied: edgecase: DVO->LVDS chip->LCD
partymola: can we force it to be detected as mobility?
partymola: the computer is a recent model “Dell Vostro 1000”
airlied: partymola: we could but I don’t think it is one.. from reading the website..
partymola: and it only has an external VGA connector
edgecase: arlied, ok my crashing is due to 16 bit mode. when starting openarena X gets SEGV. I cloned my hdd from laptop with 7500 -> desktop with Radeon7500 QW, still crashes
edgecase: I can try an AGP 7000, but i think it might also crash in DefaultDepth 16
airlied: partymola: lets try forcing mobility
partymola: ok
airlied: partymola: have you got a git tree?
partymola: no
partymola: should I install all the stuff from git?
airlied: partymola: how are you building it?
partymola: i am building it directly from freebsd ports
partymola: it grabs the source code
partymola: from…
partymola: let me give you the url
airlied: partymola: okay any ideas how to patch that?
icewaterman: hmm, googleearth crashes Xorg
airlied: partymola: might be easier to clone the git repo
partymola: airlied: i have the source code in my harddisk
partymola: it downloads a tar.gz
partymola: compiles, and installs it
partymola: if you want, i can send you the .tar.gz file
partymola: it’s 900Kb
airlied: partymola: no I assume its 6.8.0
partymola: yes, it is
partymola: xf86-video-ati-6.8.0
airlied: partymola: easier to install git and build the git version for playing around
airlied: partymola: you only need the driver..
partymola: ok
partymola: so what should i download exactly?
partymola: i don’t want to install too much
partymola: to not mess up packages 🙂
airlied: git clone git://anongit.freedesktop.org/git/xorg/driver/xf86-video-ati
partymola: ok
partymola: give me a minute meanwhile git is compiling
edgecase: arlied, should I try an X300 now and see if it dies in 16bit mode?
partymola: and thank you very much for your help, airlied
airlied: partymola: then you run sh autogen.sh –prefix=/usr and see if you need some devel packages.
airlied: edgecase: 16-bit is always a bit dodgy as we don’t test it.
partymola: airlied: i have compiled the whole Xorg
partymola: so i guess i will have all the necessary packages 🙂
edgecase: i have an interest in using 16bit mode, is there anything I can do?
airlied: edgecase: track down what is going wrong :), or maybe find a version that worked and bisect the breaks.
edgecase: ok bisect it is
edgecase: oh, i’m using git master now, but i still only see 1 XV port
airlied: edgecase: maybe 6.6.3 or somewhere like that..
edgecase: i’ll use last ubuntu version as a base
edgecase: Adaptor #0: “ATI Radeon Video Overlay”
edgecase: do I need to do something to enable textured video?
airlied: edgecase: shouldn’t .. it should work..
airlied: I’ll bbl.. shower time
edgecase: anything i should see in logs?
edgecase: i’ll try some newer cards and see what happens
partymola: airlied: error compiling: http://pastebin.com/m19d75efe
partymola: oh, he went having a shower 🙂
partymola: edgecase: do you know why the compilation failed? 🙁
airlied: partymola: you have too old a version of the xorg macros I think
partymola: i have xorg 1.4.0 …
partymola: oh
partymola: compiled it
partymola: added the missing declarations 🙂
partymola: oops, didn’t work :/
airlied: partymola: what it do?
partymola: make: don’t know how to make atibus.c. Stop
airlied: did you run autogen.sh –prefix=/?
partymola: yes
partymola: my prefix is /usr/local
airlied: I’m just wondering where its finding atibus.c
partymola: ehm…
partymola: i think it may work now
partymola: i copied config.h from my local port
partymola: and now it’s compiling
partymola: ok
partymola: driver installed
partymola: let’s test it
partymola: same problem :/
airlied: partymola: well we haven’t hacked it yet
airlied: so thats no surpsring
partymola: ok
partymola: i am compiling it with version 6.18.0
airlied: so edit src/radeon_driver.c
partymola: to see if i am using the custom one or not 😉
partymola: ok, opened in vim
airlied: at line 1609 add info->IsMobility = TRUE;
partymola: ok
partymola: changed
partymola: make clean && make install
airlied: rebuild install and see what happens.
partymola: it worked!
partymola: perfect image :))))
airlied: partymola: nice.. can you post your log file..
partymola: yes
partymola: airlied: http://uma.frikiplan.com/radeon/Xorg.0.log
partymola: there you can see my 6.18.0 module lol
airlied: partymola: excellent, now to fix it properly..
partymola: actually it was an easy fix, eh 🙂
partymola: you have a table with device ids to decide if they are Mobility or not, or what?
airlied: partymola: yup I have a csv file that we use to generate tables..
partymola: if you need any info of my hardware, just ask it
airlied: partymola: should be fixed in git now.
partymola: nice 🙂
partymola: i will take a note of the process to not forget it 🙂
partymola: i hope the freebsd port maintainer updates the port of the driver asap 😀
edgecase: arlied, solid work!
edgecase: tried Radeon 9200 AGP, openarena dies in depth16
edgecase: XVideo works in 16bit, on 7500 and 9200
edgecase: now to try 7500M
edgecase: arlied, seems it’s only Radeon Mobility M7 LW [Radeon Mobility 7500] that’s got XVideo broken in depth 16
edgecase: does that make sense?
edgecase: seems Macrovision defaults to ON
edgecase: heh enabling S-video out fixes Xvideo
airlied: edgecase: it could happen.. the bios defines macrovision.
airlied: I don’t think we touch it.
edgecase: heh hands off or else
edgecase: hmm have to restart X and see if it’s really the S-video that fixes it
edgecase: oh snap
edgecase: xrandr –output S-video –auto <— fixes XVideo overlay
airlied: edgecase: that is wierd.
edgecase: progress is nice
edgecase: arlied, i’d test the patch you made the other day, interlace + ColorTiling bug, but i seem to have misplaced it
airlied: edgecase: its in git now.
edgecase: oh
airlied: edgecase: alex cleaned it up
edgecase: ah nice
edgecase: you guys are making my head spin
edgecase: i’m using git master, however it appears I still have an issue
airlied: edgecase: wierd..
edgecase: xrandr –newmode NTSCi 14.318 768 784 848 910 483 484 492 525 Interlace -VSync -HSync
edgecase: xrandr –addmode VGA-0 NTSCi
edgecase: xrandr –output VGA-0 –mode NTSCi
edgecase: causes LCD panel of laptop to go crazy-tiled
airlied: edgecase: hmm we should reprogram things properly.. but maybe not..
edgecase: i didn’t test your patch before, because i had trouble with it + 6.8.0. I’m using git now,
edgecase: is the anongit lagging behind?
airlied: nope.. its the same repo
airlied: edgecase: yeah we don’t go back and fixup all the crtcs..
airlied: edgecase: we might need a call to do that
edgecase: do you want Xorg.0.log ?
airlied: edgecase: nope, I just need a reminder to fix 🙂 file a bug..
edgecase: ok where is bug tracker?
edgecase: bugs.fd.o ?
airlied: edgecase: yes..
edgecase: product xorg…
airlied: should be a drivers radeon component
edgecase: yeah found it
edgecase: cc ?
edgecase: aight filed 2 boogs
airlied: edgecase: they all go to the list anyway..
edgecase: yeah, i guess i should subscribe by now
edgecase: next i will try Texturedvideo, on X300
edgecase: is there anything in log from textured video, that might indicate why it’s absend on r100/r200 ?
airlied: edgecase: not sure.. try with EXA enabled
edgecase: so i noticed, the atombios cards, mention framebuffer reserved by firmware,
airlied: edgecase: its just stack space it doesn’t really need to be in VRAM
edgecase: CPU stack?
edgecase: wierd
airlied: stack for the atom interpreter
edgecase: back in the days of static vram, you could save a crashdump over reboot on vid card, i believe
edgecase: just more VGA legacy removal musings
edgecase: Option “AccelMethod” “exa” ?
Terman: I’ve gut a bug with my T60 (X1400 mobility) and radeon: The display is 1400×1050. When I do xrandr –output LVDS –rotate right, the display rotates, everything gets a proper new place, but the topmost 350 pixel (in the rotated display) remain black (the windows are placed in there) – except the mouse, which is visible
Terman: is this a known problem? I’m using Option “AccelMethod” “EXA”
edgecase: arlied, exa didn’t help, would there be anything in the logs?
airlied: Terman: I’m not sure its a known bug ..
airlied: edgecase: maybe.. I’m not sure about the r100/r200 textured video codepathas
edgecase: i’ll try depth24, then X300
Terman: airlied: or any way how to proceed from here to pinpoint the bug tomorrow (not tonight, I’m rather tired)?
airlied: Terman: log a bug and someone might take a look.. I need to test my r500 again
Terman: airlied: I’m very happy with the current git-head. I can use it for my daily production work already (clone mode) and to watch videos 😉
Terman: airlied: I was playing around when I noticed the bug
damentz: hey airlied, i used a friends projector successfully with the xrandr cli tool
damentz: it just experimented with commands
damentz: though all i could do was get it to clone
damentz: i didn’t have any gui xrandr 1.2 tools i could use
airlied: damentz: normally for non-clone you have to set a virtual line
damentz: there was one called grandr but it crashed whenever i clicked on things
damentz: airlied: oh?
damentz: airlied: what gui tool are you most impressed by at the moment?
damentz: it seems that debian sid doesn’t include any of them
airlied: damentz: I don’t use gui tools yet, but the one redhat are writing is the one I’ll be using 🙂
damentz: alright, i’ll look forward to that
damentz: i tried grandr, but it crashed whenever i clicked on monitor placements and what not
airlied: there is one going into gnome that ssp is doing…
damentz: i’m guessing it was xrandr 1.0
damentz: oh ya, its supposed to be in ubuntu hardy heron, should be interesting
edgecase: hmm xrandr –newmode doesn’t like +CSync
edgecase: CSync works
edgecase: ooh sexy, pure RBG on the TV
edgecase: now to swap R <–> B
simpsonc: airlied: Just finished reading through the Red Book on frag shaders. Looks like if I just get it to exit with the first instruction, the shader should still do most of the work automatically, right?
airlied: simpsonc: you won’t get any output unless the last instruction writes to the outputs.
simpsonc: airlied: Right. It looks to me like I can make an instruction of type OUTPUT that has the LAST and OUTPUT flags set, and then just copy the RGBA data directly in one go.
simpsonc: I think.
simpsonc: Testing will be required, so I have to figure out the code paths.
airlied: simpsonc: yup, you should be able to do a single instruction that just copies RGBA to OUTPUT
simpsonc: But in the long run, only the frag shader parser will actually need serious work.

Radeon IRC Logs For 2008-3-15

Rig Lab Hardware

You Might Also Like

RadeonHD.org | Radeon IRC Logs For 2008-10-23

RadeonHD.org | Radeon IRC Logs For 2008-11-03

RadeonHD.org | Radeon IRC Logs For 2008-5-19