Nightwulf|work: hi all
reid: anyone here use jaunty with a Radeon X1250 card? (onboard laptop)
olafskaug: I'm having trouble with a radeon 2400 mobile card. I can start X without trouble, but the 2D acceleration seems to be broken or poorly configured. Followed the community documentation on how to set it up and I put the AccelMethod to EXA and set Option DRI. anything else i might have forgotten?
olafskaug: also my Driver is set to radeonhd
olafskaug: glxinfo | grep "renderer string" returns "OpenGL renderer string: Software Rasterizer"
honk: there is no 3d accel for that card unless you use git - and even then it
honk: it's utterly useless :]
olafskaug: I'm not especially after the 3d rendering. Mostly interested in fixing the laggy picture when scrolling and such
olafskaug: would it help just using the "radeon" driver?
honk: if exa is up and running, scrolling should be pretty smooth
olafskaug: could the problem with scrolling be that it just says Option "DRI" without the "on" after it?
olafskaug: I'll try using radeon driver instead
honk: post your logfile on a pastebin
Diablo-D3: hey guys
Diablo-D3: whats the newest card that has working 3D support?
airlied: do you mean newest or best?, they bring out new low end cards
Diablo-D3: or best
Diablo-D3: something that doesn't totally suck ass, but isnt ancient either
airlied: x1600 or x1900 range is probably the best choice at the moment if you can find them
Diablo-D3: Im glad fglrx works on my 4850
airlied: it shouldn't be too far away with the open driver
airlied: the code is nearly all there, just the bugfixing fun.
nenolod: is there like, a hardware compatibility list?
honk: man radeonhd
Diablo-D3: hi nenolod
nenolod: honk: that doesn't tell me if my radeon 3600 is supported by DRI or not.
Diablo-D3: its experimentally supported.
Diablo-D3: nenolod: I told you, plug it in, build radeonhd from git, and see if it works
Nightwulf|work: airlied: which version of opengl will be supported?
Diablo-D3: worse case is, it doesnt
airlied: Nightwulf|work: same as r300/r500 initially
nenolod: Diablo-D3: i don't wish to pollute my debian system with stuff
airlied: Nightwulf|work: so 1.4 or 1.5
Diablo-D3: nenolod: ....
Nightwulf|work: airlied: ah...but 2.x or even 3.x is planned to be supported?
airlied: Nightwulf|work: in the gallium driver
Diablo-D3: Nightwulf|work: generally, new opengl versions have nothing to do with video cards
airlied: which hasn't started yet
Nightwulf|work: Diablo-D3: i know...but that wasn't the question ;-)
Diablo-D3: Nightwulf|work: they're just collections of extensions (except 3.x's new API style)
Nightwulf|work: airlied: ah...i c...thx
nenolod: Diablo-D3: my goal is to accomplish 3d acceleration under ParavirtOps/Xen, since the NVIDIA driver (even with my patches) is very crashhappy
Diablo-D3: nenolod: heh, get this
Diablo-D3: fglrx doesnt work under xen on the dom0 either
nenolod: Diablo-D3: i am presently using fbdev, it is great
Diablo-D3: why do you care about 3D anyhow?
Diablo-D3: compiz is very boring
nenolod: Diablo-D3: because i have contractual obligations
Diablo-D3: then do what Im seriously considering
Diablo-D3: build another machine.
Nightwulf|work: Diablo-D3: yes, i know...that's why i asked wether it will be supported by the driver...but the hardware decides wether a GL function is executed in hardware or software
nenolod: Diablo-D3: presently broke until next month.
Diablo-D3: Nightwulf|work: no, technically DRI does
nenolod: Diablo-D3: need 3d support now.
Diablo-D3: nenolod: have credit card, will travel
nenolod: Diablo-D3: touché
Diablo-D3: Nightwulf|work: the problem is, mesa doesn't handle some things correctly itself
Diablo-D3: Nightwulf|work: like, shaders still suck dick from what I've heard
Diablo-D3: Nightwulf|work: so if mesa can't handle what an extension does or doesn't know of the extension, DRI can't expose the hardware that does it
Nightwulf|work: Diablo-D3: that's not the whole truth...if a card doesn't support e.g. shareds 3.0, even mesa or other parts of the driver can't let the card do it in hardware
Diablo-D3: Nightwulf|work: in a lot of cases, new driver development stalls because mesa sucks dick
Nightwulf|work: Diablo-D3: yes, might be...I'm not the mesa specialist :P
Diablo-D3: Nightwulf|work: if the card doesn't support a feature, obviously the driver wont
nenolod: Diablo-D3: i can pick up a radeon x800 xt for 20 bucks
nenolod: Diablo-D3: so i guess i will do that
Diablo-D3: Nightwulf|work: but simply checking opengl's version in an app and expecting the extension to be there is stupid
Diablo-D3: Nightwulf|work: _always_ check the extension
Nightwulf|work: Diablo-D3: of course
Diablo-D3: Nightwulf|work: so, your question really should be "can radeonhd accelerate extension foo yet"
Diablo-D3: nenolod: those are so old
nenolod: Diablo-D3: and yet, they outperform r500s
Diablo-D3: nenolod: not really
airlied: they shouldn't outperform r500s
Diablo-D3: they dont outperform r500s of the same class
nenolod: Diablo-D3: actually, my mac has an x800 xt. but it's probably all apple infected and won't work in a PC
Diablo-D3: you're comparing r400 to r500 dude
Nightwulf|work: Diablo-D3: well, yes and no.....if the driver supports a special version of opengl, it means of course that it does so in hardware only if the card supports it
Diablo-D3: Nightwulf|work: but there are no "special versions of opengl"
Diablo-D3: Nightwulf|work: its either standardized, or it doesnt exist
Diablo-D3: vendor extensions _are_ standardized even if they aren't ARB extensions
Diablo-D3: mesa doesn't need to be able to render something, it just has to be able to expose the extension in the API
Nightwulf|work: Diablo-D3: you tend to misunderstand people, do you? i could have said "a certain version" too...doesn't matter
Diablo-D3: see above, however
Diablo-D3: your app cant do something using DRI if mesa doesnt have it in it's api.
Diablo-D3: nenolod: just buy that x1650pro
Diablo-D3: nenolod: its $40
Diablo-D3: nenolod: you've wasted like ten times that on stupid shit
nenolod: Diablo-D3: such as...?
Diablo-D3: nenolod: kash's paycheck.
nenolod: but "me", and "my company", are two different things
Diablo-D3: yes, but its your company; you're the head of it and you have like 3 employees
Diablo-D3: you probably could unilaterally negotiate a bigger paycheck with yourself
nenolod: i could, but it's unnecessary
Diablo-D3: in fact, if you really needed a lot of cash
Diablo-D3: you could use yourself
Diablo-D3: nenolod, representing atheme, suing nenolod, representing rx
Diablo-D3: since kash kashed atheme's data
nenolod: that's... kind of retarded
Diablo-D3: well, you could settle with yourself out of court
nenolod: if you tried to sue yourself, they would send you to a looneybin
Diablo-D3: well, if you get a law degree
Diablo-D3: you can also represent both sides.
Diablo-D3: although it might be a conflict of interest
Diablo-D3: so you probably could sue yourself for representing yourself and yourself at the same time
vivo: hi all, just a note, RadeonHD IRC Logs stops at 2009-04-17
Diablo-D3: nenolod: I should tell z this
Diablo-D3: nenolod: he'd probably figure out a way to end up with more cash than he started with
Diablo-D3: nenolod: at any rate, its $40
Diablo-D3: who cares about $40
nenolod: Diablo-D3: well, i don't want to spend $40 and then have it not work
Diablo-D3: its probably going to work
Diablo-D3: everytime you buy new hardware you run the risk of it not working
MostAwesomeDude: X1650? It'll work. Just don't buy AGP.
Diablo-D3: hes on pci-e
Diablo-D3: and generally, for the record, agp has always sucked and will always break your shit
phercek: Is it expected that radeonhd does not support video modes bigger than 2560x1024? For example xrandr does not report mode 3840x1024 60 Hz as supported but it is reported as supported by EDID (and in windows the mode works).
MostAwesomeDude: Your Virtual isn't big enough. Additionally, I didn't know there were monitors that big. :3
Diablo-D3: you mean that small
phercek: it is not that big monitor but it is connected through matrox triplehead2go
Diablo-D3: its three 1280x1024 monitros
MostAwesomeDude: Well, at any rate, which video card is this?
phercek: triplehead2go splits one dvi input to up to 3 monitors; so the biggest mode supported is 5040x1050
Diablo-D3: phercek: yeah, I never understood why they botched that
Diablo-D3: the smallest height it should have supported was 1200
phercek: it is radeon hd 3850
MostAwesomeDude: Hm. Does xrandr report that your Virtual is large enough? If not, you'll have to force it bigger in xorg.conf.
phercek: is it possible to increase "the virtual" ?
Diablo-D3: yes, he just told you
phercek: ok; I see my virtual is only 2560x2560
phercek: when I increased virtual to 3840x1024, it did not help
MostAwesomeDude: Hmm. The mode still doesn't show up?
phercek: yes it does not
MostAwesomeDude: Hm. Did you try forcing the mode?
phercek: no; I'll try
phercek: xrandr responds with cannot find mode 3840x1024
MostAwesomeDude: Hm. You may have to use the --newmode to get it added. Weird that it doesn't show up in the EDID.
phercek: monitor (based on Xorg.0.log) reports it in EDID but xrandr does not list it
phercek: I set virtual exactly to 3840x1024, I tried even bigger but then I did not get any image, so I do not know what happend that time.
MostAwesomeDude: I think the scanout limit's somewhere around there.
phercek: what kind of hsync vsync should I add there? (this is not written in the additional video mode description)
MostAwesomeDude: Hm. Not sure TBH.
MostAwesomeDude: I really am not sure what to do.
ahox: Hi, i have some trouble with my radeon/radeonhd driver. Am I correct that the radeonhd does not support displayport yet?
phercek: hmmm when I added it, it got added only to the analog output and not to the digital one; this looks like a problem with the top clock limit of dvi; but does not spec allow exceed the limit when it is in dual dvi mode?
MostAwesomeDude: ahox: AFAIK no DisplayPort support yet.
phercek: if it is it; then is should not be limited for dual dvi; on windows it works
MostAwesomeDude: phercek: In theory yes, in reality maybe not. You'd have to ask agd5f or somebody else that knows more about modesetting.
ahox: MostAwesomeDude: thanks, so I guess I have to go back to the radeon driver and wait a bit more
phercek: MostAwesomeDude: ok, thanks; maybe agd5f responds since he seems to be online
MostAwesomeDude: phercek: He's always idling, just like the rest of us. He's probably still asleep though. :3
phercek: any specific time I should try to bug him?
MostAwesomeDude: phercek: Ping him and wait.
MostAwesomeDude: agd5f: ^^
phercek: ok; thanks
agd5f: phercek: no displayport support yet
phercek: looking up what displayport is
phercek: agd5f: should I understand it that way that there is no dual-dvi support yet too?
agd5f: phercek: dual-link dvi shoudl work fine
phercek: ok; then I do not understand how this is related to my problem; I mean the card does not have display port connector anyway, and it works on windows xp
phercek: agd5f: if you would want to look at logs of what I did, they are here: http://www.hck.sk/users/peter/pub/xorg/
agd5f: phercek: your config is wrong
agd5f: you need to to add: Depth 24 to the displaydubsection
phercek: agd5f: thanks, I'll try it
phercek: agd5f: adding "Depth 24" did not help; xrandr still does not show mode 3840x1024 as available
agd5f: phercek: what does xrandr show as the maximum size?
phercek: 2560x1024 60Hz is the biggest mode available
agd5f: phercek: when you run xrandr what does the first line say?
phercek: check http://www.hck.sk/users/peter/pub/xorg/xrandr.txt
phercek: agd5f: update there (http://www.hck.sk/users/peter/pub/xorg/) alst the xorg.conf and the log file
MarcOChapeau: Hi all :)
MarcOChapeau: I have an interesting issue with Xv and video playback.
MarcOChapeau: I have a radeon 2900XT running the radeonhd driver from master, libdrm from agd5f's repo (r6xx-r7xx-3d branch) and mesa r6xx-r7xx-support branch.
MarcOChapeau: When playing a video in fullscreen, the picture is slightly smaller than the screen size, and the gnome pannel remains visible.
MarcOChapeau: My TV is connected via an HDMI cable to my radeon HD. The issue dissapears if I stop using Xv, or if I connect my TV through a VGA cable (in this case Xv works fine). is that a know issue ?
yangman: agd5f: anything new on the cursor/watermark situation?
agd5f: yangman: nope
yangman: agd5f: is there a register range I can poke at?
yangman: trying to refactor the cursor code atm
agd5f: yangman: http://www.botchco.com/alex/xorg/avivo_watermark.diff
agd5f: is what I started playing with
agd5f: but doesn't seem to fix the cursor issues
agd5f: might help with display underflow issues though, haven't had time to dig to much further
yangman: the cursor dimension setting code in radeonhd is all out of wack
yangman: it actually just ends up being 64x64 all the time
agd5f: yangman: yeah
yangman: I've been able to produce some wacky corruption bands by setting different crtc sizes
agd5f: X doesn't specify a cursor size
agd5f: so we alwyas use the max size advertized
agd5f: yangman: there are actually two different cursor issues
agd5f: one is fixed by never having the cursor image end on a 128 pixel boundary, the other I haven't figured out yet
yangman: I'll have to rework the 128px boundary one into the refactored code
phercek: agd5f: is the situation with the 3840x1024 mode a bug? should I report it somewhere?
agd5f: phercek: sure. bugs.freedesktop.org
phercek: agd5f: ok; thanks for help
Obscene_CNN: well it looks like the power management stuff saves me about 100 ma in current consumption.
v4nelle: guys can i use tv-out with radeonhd?
Zajec: v4nelle: no
Zajec: v4nelle: if you mean S-Video (i'm quite sure you do)
yangman: hm. can't reproduce the multiple-of-256-only cursor corruption anymore
agd5f: yangman: only happens when bothe crtcs are enabled
yangman: agd5f: yeah. there was a case where only exact multiples of 256 would corrupt and not a range
Athas: Is there a guide to setting up RadeonHD on Debian Unstable?
yangman: either setting cursor dimensions fixed it or the other corruption is masking it
agd5f: yangman: yes, that's the 128 pxiel bug I was talking about
agd5f: the range one is the one I haven't figured out yet
frische: hi. my radeon 2600hd works with the radeon/ati driver but not with radeonhd. the error msg is 'screen not found'
frische: Xorg.log - radeonhd : http://pastebin.ca/1405180
frische: Xorg.log of a working session with radeon/ati: http://pastebin.ca/1405181
frische: i built radeonhd from git
frische: but iirc previous versions had the same problem
yangman: seems like a lot of work and extra CPU cycles to work around the cursor issue when enabling it on both CRTCs is cheap and easy
agd5f: yangman: but the issue only manifests when both crtcs are active
yangman: agd5f: well, I think I'm going to implement the old workaround into the refactored code. shouldn't be hard to only do it when both crtcs are active
agd5f: yangman: does enabling both cursors actually work?
yangman: agd5f: yeah. reliably
agd5f: yangman: does it fix the 128 pixel bug or the range corruption?
yangman: agd5f: definitely the range corruption. 128, I'd have to check
yangman: yup, both
yangman: might be worth it to get a dump from fglrx or widows catalyst to see what they do
yangman: the workaround's worked fine for radeonhd until RandR cursor was implemented
frische: Did someone took a log at those logs? Can you think about anything else that could be useful?
ajavid: what is the diff between radeon and radeonhd, why is the reason for developing two seperate branches of radeon drivers?
honk: dont crosspost ;p
ajavid: I meant to ask it here
ajavid: but in the other channel i pushed up arrow accidently
ajavid: thats why if you notice, it was followed by a 'doh'
ajavid: as in 'oops'
honk: well, you did get your answer that way ;p
Geoffrey2: my motherboard has an integrated Radeon X1250, would the radeonhd driver be my best bet? any particular reasons I would want to use the standard radeon driver instead?