pinchartl: is anyone working on 3D support for the R600 serie (and more specifically the RV620) ?
pinchartl: or 2D EXA
bridgman: pinchartl: yes, there is a group of developers working in an NDA area on using the r6xx 3d engine (which we need for 3d, exa, textured video etc...) in parallel with clearing the last of the 6xx 3d documentation for public release
ndim: Hmm. I have a crasher. 3D accel, a certain torcs track, and suddenly the FPS drop from 20 to 10 to 3 and the box hangs.
libv: woliu: :)
Andre_h: How far have u been able to underclock your 48xx cards?
Andre_h: Or rather how low
markit: hi, I've a 2400pro, does radeonhd 1.2.1-2 (debian sid 64 bit) work? last time I tried, I had a black screen and had to reinstall nvidia crap
markit: (I'm not interested in 3D, just be able to work with KDE)
buggs: markit, http://www.x.org/wiki/RadeonFeature
Fuddl: markit: i've got a 3650 in my sid/amd64 machine. it works fine even without 2D accelaration
markit: urgh, 2400 should be 610chip, lots of green box missing also in 2D
Fuddl: markit: hm, since i didn't experience any problems with the version in the debian package i have no experience with any workarounds. just a stupid guess: did you try playing around with "AccelMethod" in xorg.conf? (see "man radeonhd") perhaps that could help
markit: Fuddl: do you use DVI output also?
markit: (is digital connector DVI? forgive my ignorance)
markit: Fuddl: I simple installed, changed xorg.conf to reflect new driver, rebooted, saw black screen and reverted to nvidia
Fuddl: markit: yes. the TV-output is the only thing i didn't try out, yet. bug DVI, 2xDVI and DVI+VGA work for me
markit: I should take a working pc side by side, to google and check for help while troubleshooting next time :)
Fuddl: markit: hehe, can be helpful in "dark" situations ;)
xtknight: have the docs for r6xx been released to the public or radeonhd yet so that implementation can begin?
Fuddl: xtknight: http://www.x.org/docs/AMD/ says "no"
adamk_: xtknight, Development has begun for some developers under NDA. The documents have not been released with an NDA yet, though.
Fuddl: markit: have you also tried radeon (not radeonhd)? both drivers should be able to work for your card
xtknight: so r600isa isnt really relevant
Fuddl: markit: but i experience problems in 2xDVI mode - the second DVI-output stays black on radeon
Fuddl: xtknight: nope, it's really only about the ISA as far as i read through it
markit: Fuddl: no, I will try next time, thanks
xtknight: oh Instruction Set Architecture, whatever that means ;)
markit: Fuddl: is radeon still a Free driver? or amd binary? I avoid closed sources driver
Fuddl: markit: ... i would never suggest non-free drivers! ;)
xtknight: radeonhd/radeon/avivo/ati are open source, fglrx is the binary one by amd
Fuddl: markit: so yes, it's GPL/MIT afaik
markit: Fuddl: I'd also prefer more Free (GPL) over less Free (MIT) licenses, I'm a sort of RMS fan :)
adamk_: markit, Then you'll be disappointed to know that the Xorg drivers are MIT licensed.
markit: adamk_ in fact :(
adamk_: Which is something I'm very very happy with.
Fuddl: markit: come on, what does the difference between damn free and not-so-damn-free software compared to a blob you cannot trust?!
markit: adamk_ as far as I know, in the past was a big weakness of them, since many companies keep their improvements for their own use, without contributing
markit: Fuddl: sure, I agree :)
markit: was just to do some talk :)
adamk_: markit, That's not a weakness... That's the very idea of the license. Letting developers use the best code available, whether they want to release their modifications or not.
markit: adamk_ I understand, but for me that want code to be Free, is a weakness that damages users (me). But don't want to flame :)
markit: hope radeonhd will catch up quickly a more wide support of ATI boards
markit: so thanks to all the developers here :)
markit: and keep the good work :)
Andre_h: goonight niggers
ndim: xdpyinfo says
ndim: dimensions: 1400x1050 pixels (370x277 millimeters)
ndim: resolution: 96x96 dots per inch
ndim: (WW) RADEONHD(0): No monitor size info, assuming 96dpi.
egbert: ndim: this is easy to explain.
egbert: ndim: x advertises the size and the pixel resolution not the dpi to the clients.
egbert: the client recaluclates the resolution.
egbert: so the driver figures out the dpi the server recalucates a size from it which is passed to the client.
egbert: if you take above numbers you will arrive at 96dpi.
ndim: egbert: OK, so someone is not detectin gthe panel size.
ndim: And that would be radeonhd.
agd5f: ndim: not all panels provide an edid, much less physical dimensions. depends on the oem
agd5f: for LVDS panels that is
ndim: (II) RADEONHD(0): Query for AtomBIOS Get Panel EDID: failed
ndim: IIRC I had that in former driver versions.
agd5f: ndim: yeah, your laptop probably doesn't provide an edid for the panel
ndim: starts looking for the last working version.
ndim: I am certain the font sizes were different with the driver from a few days ago
ndim: OK, that was a glitch.
ndim: After a reboot, everything works fine.
ic3man05: Hi, I'm looking to help with development for the radeonhd driver but have never coded a driver yet, though I do know c/c++, any information on where I can get started at?
ic3man05: I have a RS690 X1200 and HD4870 available to me
xtknight: ic3man05, there is some dumping and tracing you may be able to do
xtknight: that basically means you run the proprietary driver and mmiotrace(memory mapped IO trace) what the driver does. i am not sure if radeonhd needs that or not as i'm not a dev, just a suggestion
xtknight: and certainly you can test generic 2D functionality on those cards and see if it works as expected
airlied: ic3man05: it mainly is what you are interested in doing.
xtknight: is radeonhd developed fully by the open specifications or is there some reverse engineering going on like mmiotrace for tvout,etc,etc?
airlied: xtknight: all from specs.
airlied: I don't think we've really done any mmiotracing of it.
airlied: for 3D stuff we've done some stuff using revenge
xtknight: isn't there something that needs dumping for the atombios, like certain vendors whose cards don't conform to the specs(exceptions), etc
airlied: xtknight: not really, the connector tables are dumped for radeonhd to workaround problems.
xtknight: airlied, so basically people just need to implement what is in the open specs but the laymen cannot do much at this point(not like the nouveau project where many dumps were needed)
airlied: xtknight: pretty much, no need to lots of dumps when you have specs.
ic3man05: Since I haven't coded like this yet I was thinking just trying to fix open bugs then try to help actually develop whatever needs to be done
ic3man05: and of course testing it in the process
xtknight: ya well i am a bit of a programmer, i am just afraid i would mess something up because i haven't done driver programming before
ic3man05: xtknight, I feel I'm in the same position
xtknight: i did a little driver programming but it crashed windows several times haha :)
ic3man05: I could only imagine windows being a nightmare to code drivers for
xtknight: see when you're executing at that level (ring 0) you can't do anything wrong. else, you get a hard crash. that's what makes it a bit hard
xtknight: although i assume when they are making interfaces like this there is just a lot of janitorial work like copying structures and making functions to copy structures, since there are about a million of them
ic3man05: I think i'm going to take a look at the source tonight after work
ic3man05: if I have time
xtknight: look at the docs they have online
xtknight: http://www.x.org/docs/AMD/ many of the docs have stuctures and bitfields etc that have diff variables in them
ic3man05: cool, definately look at that later
prav33n: I am trying to adjust the backlight brightness of my Thinkpad T60 (ATI Mobility Radeon X1300) using xrandr
prav33n: Anyone knows the name for the property to be used with 'xrandr --set'?
airlied: prav33n: xrandr --verbose should show it
prav33n: airlied: Thanks for the tip
prav33n: airlied: I am not sure which one it is
yangman: there's a BACKLIGHT property
yangman: you may be running a version that's too old
prav33n: yangman: radeonhd version?
prav33n: yangman: xrandr version?
yangman: I don't recall when it was added. I run git version of radeonhd myself, and 1.2.3 xrandr
prav33n: yangman: Thanks for that info
prav33n: I guess I have to live with the dullish backlight till I get an updated driver in OpenSolaris
prav33n: I also have a wierd problem on my OpenSolaris
yangman: well, switching to another VT, adjusting it there, then switching back is still a reliable way to do it
prav33n: When Firefox is opening any page and showing the progress animation (small circle) rotating in the favicon position, Xorg starts using 100% core
prav33n: yangman: Unfortunately OpenSolaris doesn't have a vt
prav33n: yangman: I am new to OpenSolaris
prav33n: yangman: I was a GNU/Linux user before
MostAwesomeDude: ...I didn't know that the radeon/radeonhd drivers worked on OpenSolaris.
prav33n: MostAwesomeDude: It works great so far
prav33n: MostAwesomeDude: I am running a dual screen setup successfully at 3200x1200
prav33n: I see a few performanfce issues
prav33n: However I hope they would be resolved as the driver evolves
prav33n: It is a awesome driver compared to the crappy fglrx I was using for a couple of years
MostAwesomeDude: prav33n: Agreed. fglrx is a PAIN.
prav33n: Yeah, I even stopped bothering about 3D for a few months now
prav33n: I would rather wait for 3D support in radeonhd instead of going back to fglrx
airlied: hopefully OSOL gets with the kernel modules stuff.
prav33n: airlied: Yeah, they are working on the DRM
airlied: they kinda sucked at getting a drm working before.
prav33n: One major issue that bothers me now is that my cursor becomes like dissolving in flowing water in certain rows of the screen when I use dual screen setup
MostAwesomeDude: airlied: Is there a reason why we don't maintain the OpenSolaris modules?
airlied: MostAwesomeDude: because its isn't really an open source OS :)
airlied: MostAwesomeDude: its like a shim of open source over a backend of secretive Sun engineers
dmb: airlied, its open source...
prav33n: Looks like this row that dissolves the cursor is approximately 10 to 20 pixels in width and they repeat on the screen at specific distances
airlied: dmb: the source is open, the ability to make changes and get them accepted not so much :)
dmb: there are some closed source drivers, but for the most part its open source from what i experienced
dmb: airlied, you can say that with a lot of open source projects
airlied: dmb: the process overhead in working on OSOL is immense.. Sun are highly protective of it.
airlied: dmb: most projects are technical barriesr.
airlied: dmb: i.e you must have a clue..
MostAwesomeDude: airlied: So it's not that we couldn't, it's just that there's very little reward for it?
dmb: kind of like mozilla/firefox
airlied: MostAwesomeDude: exactly the return on investment isn't that high.
yangman: prav33n: it's a known issue, but no one's been able to diagnose the cause
prav33n: yangman: Which one?
prav33n: yangman: Disolving cursor?
yangman: prav33n: cursor issue
airlied: I think alex changed the cursor alignment in -ati recently
airlied: not sure if radeonhd got the same issue.
airlied: also the Solaris DRM kernel interface deviates a bit from the upstream one.
airlied: and they didn't consult us during the development.
prav33n: However I am seeing this only in dual screen setup
airlied: now they want us to merge this major mass of bonghit ifdefs.
yangman: prav33n: most people experience it randomly. if you havea more consistent experience, I'm sure that'd help out a lot
prav33n: yangman: I am seeing it consitently
prav33n: Everyday at the same column
yangman: over multiple restarts?
prav33n: I reboot my laptop everyday
prav33n: When I use panel output only, I don't see the problem
prav33n: When I use dual screen (VGA+DVI) (PANEL closed), I see this issue consistently
prav33n: If there is a bug open on this, I can provide more information
prav33n: I will be more than happy to help tracking down the issue
prav33n: I was suspecting hardware cursor could be an issue?
prav33n: yangman: Exactly
prav33n: yangman: That is the bug
prav33n: yangman: Thanks for the pointer
prav33n: I will update the bug with my case
prav33n: However I don't run the latest driver
prav33n: yangman: Updated the bug with whatever information I have