Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

Radeon IRC Logs For 2008-11-15

Search This Log:


damentz: gn8
SimpleAONE: hi
SimpleAONE: anyone here that knows a LOT about r200 and r250 etc.. ?
glisse: SimpleAONE: you should just ask your question
glisse: you will see if anyone can answer
C10uD: hello there
C10uD: i'm trying to play a game with wine
C10uD: but it starts only with fglrx
C10uD: i want to use radeon, but it complains i don't have a good graphic card
C10uD: or whatever
hifi: what graphics card you have
C10uD: ati x1600
C10uD: but with fglrx, it works
hifi: X1600 is a RV515 chipset
hifi: what versions of the radeon driver, xorg and mesa you have?
C10uD: ubuntu intrepid's so it should be 7.4, 7.2
C10uD: and lemme check radeon driver
hifi: does glxinfo say you have direct rendering on?
C10uD: yup
C10uD: mesa renderer r300 etc.
hifi: glxgears works fine with over 1k fps?
C10uD: yes, 3d works good
C10uD: no compiz enabled
C10uD: or whatever
hifi: the game might use some features of your graphics adapter that aren't supported (yet) on the open source radeon driver
C10uD: i see..
hifi: the feature set of OpenGL extensions is not complete and some fall back to software rendering I suppose
hifi: if you have driconfig installed you can check if you have the fallbacks disabled, when you do some programs might run faster
hifi: I think they are by default disabled these days
C10uD: uhm, so there might be a chance
C10uD: i'll check for this driconfig
C10uD: thanks
C10uD: :p
SimpleAONE: ok
SimpleAONE: i am part of a company doing the 3d drivers for AmigaOS4
SimpleAONE: we have the problem that on radoen 8500 the 3d seems to be broken but on radeon 9250 it works
SimpleAONE: so r200 fails, r250 and friends work
glisse: SimpleAONE: should works
glisse: SimpleAONE: how is it broken ?
glisse: lockup ?
SimpleAONE: eventually
SimpleAONE: the outpot on screen looks broken
SimpleAONE: it draws
SimpleAONE: but what it draws is completely wrong
SimpleAONE: like the coordinates are all wrong
glisse: SimpleAONE: maybe somethings related to tiling
glisse: iirc surface stuff changed a bit btw this 2 asics
SimpleAONE: a small example
SimpleAONE: take compositing
SimpleAONE: we have that aswell
SimpleAONE: i guess screenshots would make it more obvious
glisse: could be
SimpleAONE: do you have 10 minutes ?
SimpleAONE: then i can reboot put in the radeon 8500
glisse: yup
SimpleAONE: re
SimpleAONE: ok i have no idea anymore who i was talking after a rebooted...
glisse: SimpleAONE: with me
SimpleAONE: ok ! :)
SimpleAONE: thanks
airlied: make sure you emit texture stuff in pairs for original r200.
airlied: amongst other workaround my brain recalls.
glisse: airlied: btw i started to convert to libdrm-radeon in my repo dri2 branch
glisse: now i need to remove the dribufmgr wrapper so i can track bo ref/unref issue i have with dri2
airlied: oh nice... I found a ttm bug in my drm-rawhide tree, bad unlock path
airlied: I'm just trying to make my discardable bo flag work properly
airlied: so I can get proper fast-user-switch working
airlied: with GL on both servers.
glisse: airlied: its a bo flags or cs flags ?
airlied: bo flag.
glisse: also i have and issue with dri2 but it might be due to ddx, when i create a context it crash others pixmap contents :(
airlied: I allocate uffers with a no backing store flag
airlied: so far I only used it to make suspend/resume faster
airlied: but for 2 servers I need to use it for all evictions.
airlied: time for tv... zzzz.
glisse: ok good night
woden1: Is 2D stable for RV730?
woden1: I mean, the Radeon HD 4670...
glisse: woden1: no 2d accel for rv730
woden1: What is 2D accel?
glisse: hardware acceleration
woden1: Not really sure what that means. I just want to know if I can use the card and work with my computer.
woden1: Just with regular non-3D stuff.
glisse: woden1: yup you can but things might be slow like video
woden1: And the same limitation is with the Radeon HD 4830 ?
glisse: everythings above x1950 doesn't have yet any accelerations
glisse: otaylor: you play a bit with kms on rawhide right ? :)
glisse: otaylor: wanted to know if you ever had screen corruption with kms + 3d app like gears
otaylor: glisse: I haven't, but I haven't done too much with kms and 3d
otaylor: glisse: the 3d stuff I've been doing recently is all compositing and kms => no zero-copy tfp
otaylor: glisse: So I've had kms off for the last week or two
woden1: glisse: Is this true for radeonhd as well?
glisse: woden1: yes true for radeonhd as well
woden1: What is the largest issue holding back progress in development of the driver?
glisse: woden1: documentations aren't yet released
woden1: This documentation is expected soon?
glisse: no timeline, getting documentation out is cumberstone
glisse: sigh this dri2 stuff is painfull at each single step
glisse: i wish the ddx was not such a mess
wohnout: Hello I'm running Gentoo with open source drivers for my ATI X1400 (R500). Xorg-server 1.5. Mesa 7.2. Drm 2.4.1. xf86-video-ati from git. When I'm trying to run GoogleEarth I get this output http://rafb.net/p/d6S3CE79.html
glisse: wohnout: mismatch in your dri/libgl driver somewhere
glisse: glxgears doesn't work too right ?
wohnout: No they works great
glisse: wohnout: can you pastebin output of LIBGL_DEBUG=verbose glxinfo
wohnout: http://rafb.net/p/rxKdWd55.html
glisse: wohnout: googleearth is a 32bits app right ?
wohnout: yes
glisse: so your /usr/lib32/libGL and dri driver doesn't support your card
glisse: but ones in /usr/lib64/ does
wohnout: hmm i will try emerge emul-linux-x86-xlibs-20081109 there should be new mesa
wohnout: this 32b vs 64b problems are so annoying
adamk: One of the reasons I'm still running a 32 bit distribution.
wohnout: thanks for advice
damentz: gm
rmh3093: can someone help me
rmh3093: i cant get my second display working
rmh3093: this has worked in the past
spstarr: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
spstarr: 2526 root 20 0 406m 49m 6200 S 98.8 4.9 1:02.69 Xorg
spstarr: sighs
spstarr: wtf is it doing!
gentooer: spstarr, did you try profiling it?
spstarr: im gonna strace the daminin thing
spstarr: sec
spstarr: its trying to do some
spstarr: EIO (permission denied)
spstarr: reboots
spstarr: yeah
spstarr: since X didnt start on the right TTY it was stuck
spstarr: Here's hoping Wayland will replace the X display server altogether :)
spstarr: airlied: i see you've been looking at EXA on xorg ML
glisse: airlied: any idea on why RADEONInit3DEngine wouldn't do its jobs
icewaterman: hi, how do i configure dual-head with the radeon driver?
icewaterman: i can only find old howtos on that
glisse: airlied: nevermind my bad, as usual ;)
icewaterman: any idea on how to setup dual-head?
icewaterman: there is very little info even google offers only quite old setup howtos, that seem to no longer work
glisse: icewaterman: google: randr1.2 tutorial
icewaterman: glisse: thx
spstarr: rmh3093: ping
spstarr: rmh3093: got repo?
rmh3093: spstarr, naa waiting for the new server to get up
spstarr: ok
dmb: whats the tool you can use to spit out the registers in the graphics card?
dmb: i want to try to provide info about my crashes
benh: crap
benh: so X radeon whacks the chip when shutting down after X has relinguished the VT to the kernel
benh: an that locks up with my new radeonfb imageblit accel :-(
benh: I knew it was whacking SURFACE_CNTL though I though that was mostly harmless (cursor stuff maybe ?)
benh: but it looks like there's more to it
dmb: i wonder why secondlife works so much better with radeon then fglrx
revx_: on what card?
dmb_: ok, i have my computer in one of those softlocks
dmb_: how can i debug it?
dmb_: the mouse is jumpy, and i here audio from a program that was open out of my speaker
dmb_: very skippy
dmb_: spstarr: was it you that said I should see whats in the registers?
dmb_: i'm sshed into the box
dmb_: basically, the screen is just frozen, can still move around the mouse, can't switch vts, can't kill x, etc
dmb_: :/
SimpleAONE: anyone on here that knows the DRI sources really well regarding r200 ?
airlied: SimpleAONE: maybe...
airlied: much better to ask questions, its a big source base.
SimpleAONE: we have problems with our 3d driver on radeon 8500
SimpleAONE: which works on radeon 9250
SimpleAONE: i talked to someone this morning about it
SimpleAONE: wrote his suggestion to my programmer
SimpleAONE: he replied this:
SimpleAONE: What I *think* is happening is that the 8500 sometimes duplicates and/or
SimpleAONE: swallows vertex data. For example, if I draw two triangles it may happen
SimpleAONE: that one vertex of the second triangle is missing and a vertex of the
SimpleAONE: previous triangle is used instead. I didn't find out what this is the
SimpleAONE: case; I suppose that the default setup for the VAP is different, but
SimpleAONE: then, I do bypass the VAP/T&L unit and only use the rasterizer.
SimpleAONE: In other words, I don't have the slightest idea. The code works on all
SimpleAONE: other R200 derivates, just not on the R200 itself.
airlied: SimpleAONE: oh wierd.. my memory only holds a couple of R200 errata.
airlied: mainly around texture engine
SimpleAONE: airlied: got that in a doc somewere ?
SimpleAONE: airlied: you know the stars mesa demo ?
airlied: hehe.. not a hope ...
airlied: are you using the r200 vertex shader?
airlied: or tcl fixed function bits?
SimpleAONE: would have to ask my programmer
airlied: the texture stuff is in r200_texstate.c look for dbg = 0x6;
airlied: has an 8500 AGP somewhere but it hasn't been plugged in in a long time.
SimpleAONE: btw this is on a PPC system
airlied: ah even more fun.
SimpleAONE: i do know we do not use the DMA engine yet
SimpleAONE: but the push method iirc
airlied: ah just mmio programming.
airlied: does mesa work doing the same thing?
spstarr: airlied: you know what's funny, the radeon driver has working 3D, but not so good EXA, and nouveau has good EXA but no (working) 3D :)
airlied: spstarr: radeon's EXA is pretty good.
spstarr: not for agp =)
airlied: spstarr: nothing to do with EXA though.
airlied: spstarr: thats a bug on AGP, not a problem with EXA on radeon
spstarr: hmm
airlied: you need to stop drawing the wrong answers from the random words.
airlied: a bug in AGP is not a sign EXA on radeon is worse than EXA on nouveau
airlied: I do wonder if I fix the EXA bug I mentioned on the mailing list will it help AGP cards.
spstarr: it might
airlied: sending flushes etc to the card for no reason is probably pointless and may cause issues.
spstarr: and might add delay
Boxfire: so I am now working with the CRT from an RCA camera, just the eye piece CRT. smallest CRT I have ever seen. I want to try and use my laptop's tvout to output to it
Boxfire: this will be lots of fun
Boxfire: for my purposes it would be nice if I just cut the code down to whats needed to activate and modify the tvout part of the radeon
Boxfire: xpress 200M
Boxfire: hmm can I do it at the same time as having VESA video out? the DRM module is broken for me for some reason
Boxfire: are specs out for the tvout part of the card?
Boxfire: I need the lowest level control I can get on what goes out the tvout pins
Boxfire: well if anyone has any more information that whats apparent in the DRI/DRM source code, please let me know. My network is very unstable, so I will probably ask about this again later on the mailing list
AndrewR: airlied, ping ...