Eythan: Hello, is there an EXA support for rv635 in the last radeonhd driver ?
yangman: Eythan: not yet. still waiting for 3D docs
Eythan: ok, thx
yangman: I'm sure the usual places will report on it when it does happen. hang tight
libv: cxo: check out the different timeframes for radeonhd and then radeon adding support for r5xx hardware
libv: bs politics indeed.
MostAwesomeDude: libv: I was running xf86-video-avivo before it was cool. :c
libv: it's why airlied stopped replying after you said "having 2 streams really doesnt make sense"
libv: MostAwesomeDude: i was talking to AMD at the time and under a strict embargo to not talk about any of the goings on
MostAwesomeDude: libv: Man, that's pretty cool. I think that's probably cooler than me.
libv: no it wasn't cool, this embargo was not enjoyable at all.
libv: but the initial result was pretty cool, yes
airlied: hehe.. I wrote initial r500 support in radeon 2-3 years ago, my embargo unfortuantely never lifted...
airlied: but to be honest I stopped talking earlier because I don't really give a crap..
ndim: egbert: Had I mentioned https://bugzilla.redhat.com/show_bug.cgi?id=466578 to you?
ndim: egbert: Especially https://bugzilla.redhat.com/show_bug.cgi?id=466578#c4 ?
Zajec: ndim: egbert: I get probably the same issue with M82
Zajec: similar colors but more readable than in case of that redhat bug
Zajec: it happens for both: HDMI and VGA
Zajec: I use git Date: Wed Oct 15 14:44:35 2008 +0200
womble123: hi guys, anyone around to help me with something?
ndim: womble123: I guess you will know the answer to that question once you have asked your question.
womble123: haha ok...
womble123: right spec 1st
womble123: pent D 2.8, asus mboard, hd2600xt
womble123: used to work fine then i turned it into a htpc and changed the gfx card (nvidia) which was having problems with overscan
womble123: so i bought the hd2600 again
womble123: been trying to install ubuntu 8.04
womble123: oh im using hdmi too the tv btw
womble123: after it installs and i do the 1st reboot i get no picture on the screen for about 5 min (no signal detected)
womble123: then it comes to life with the hard wallpaper and does nother else
womble123: any ideas?
womble123: i must say, it did work fine before i charged the card so maybe I have bought a duff one?
libv: womble123: so hd2600, hdmi doesn't work
libv: womble123: this is a secondary card, or the only card in the system?
womble123: only card in the system
libv: ok, let me steal the Big Screen from egbert :)
womble123: im not sure, im just testing it out now with dvi>vga
adamk: Did 8.04 come with the radeonhd driver? If so, it's probably a pretty old version.
womble123: yes i think so
libv: dvi->vga adapter is ok (different output)? or dvi->hdmi adapter (which is the same output as it's the same electrical signal as DVI)
womble123: yeah i have been trying dvi>hdmi
womble123: i was trying to install windows also, goes fine but when i install the drivers it wont boot anymore
womble123: right ubuntu loading up with dvi>hdmi
womble123: opps dvi>vga
womble123: well thats alot better!
womble123: infact no, its locked up again
womble123: after entering username etc
libv: it locked up?
womble123: mouse froze, same as if i just installed the drivers under windows
libv: womble123: TMDSA works nicely over hdmi here
libv: trying LVDS
libv: yeah, both are rocksolid
libv: tmds is the protocol behind DVI/HDMI
libv: tmdsa is the first dvi digital output
womble123: is it possible it the card then
libv: lvds can drive both lvds and tmds depending on how you set it up, and ati called this output just lvds :)
womble123: oh ok
libv: back in the days when outputs were still simple :)
womble123: well i have the exact same card in this pc
womble123: so i might swap over and see if i have problems with this one then
libv: womble123: yeah, sounds like a good next step
womble123: oh, slightly unrelated
libv: i'm going into opensuse 11.1b2 and see what happens there (was in trusty old 10.3 now)
womble123: is it possible to use 5.1 over hdmi in mplayer?
libv: womble123: with christian koenigs latest code, should be
matteo: ah, a nice 4870 and no acccel at all, all due to some missing docs
matteo: that's a pity
cxo: libv, Correct me if i'm wrong, these drivers are not written by software developers because they have no social lives, but because there is strong community interest in seeing ati hardware supported on Linux
cxo: So anything, that doesnt fall into that goal is classified as "bullshit politics"
libv: cxo: the reason for having another driver is bullshit politics.
libv: cxo: and the support inside radeon for r5xx and up is just about a year old i think, so feel to draw your own conclusions out of that
cxo: From a technical perspective, there maybe legitimate reasons for why we need 2 drivers supporting the same hardware. But without a clear user oriented distinction made between the two, multiple drivers with the same goal are not necessary in main line
libv: cxo: there is no technical reason, at least not one that was there more than a year ago.
libv: cxo: everything after that was wholly artificial and driven by...
cxo: There is no need for radeonhd xor radeon to support -hd cards, with http://www.x.org/wiki/RadeonFeature in mind
cxo: matteo, it should be supported within the next 6 to 12 months
MostAwesomeDude: cxo: But you know that both drivers will add support anyway, and near-simultaneously, too. :3
cxo: When resources are so limited in open source development, this all seems like such a carnal waste
libv: hrm, koenigs code looks really cool
libv: although i'm not sure "aplicateable" is a real word, but it's quirky so it stays .p
libv: wtf, apart from some further spelling issues, this is fantastic code
yangman: libv: merging HDMI audio soon, then? :)
MostAwesomeDude: cxo: You're right. With my coding skills, I'm wasting my time with school. :3
niendo: after installing xorg-server 1.4.2
niendo: xrandr seems to fail trying to arrange my monitors
cxo: MostAwesomeDude, if you thought school was there for you to learn coding skills, then you need a lot more school :)
niendo: using xrandr --output VGA-0 --left-of DVI-0
MostAwesomeDude: cxo: No, school is here so I can get a degree, and fill in whatever tiny gaps remain in my knowledge.
MostAwesomeDude: Also so I can get more exp. before wandering out into the big, bad world.
cxo: experience :)
yangman: niendo: those look like radeon output descriptors. radeonhd uses a different naming scheme. run `xrandr -q` and take a look at how you should be referring to them
niendo: yes...I mean: xrandr --output VGA_1 --left-of DVI-D_2
niendo: I tryed radeon instead but it does not work correct, too
yangman: niendo: is there an error? or just unexpected behaviour?
niendo: no there is no error message
niendo: the only strange think is, if I tyoe
niendo: xrandr --output VGA_1 --right-of DVI-D_2
niendo: I see on both screens the result I expected for DVI-D_2 for the first command
niendo: VGA compatible controller: ATI Technologies Inc Mobility Radeon HD 2400
niendo: I'm using the the latest git...
yangman: sounds like xrandr isn't being smart with crtc assignment
niendo: for radeonhd
yangman: try: xrandr --output VGA-1 --auto --left-of DVI-D_1 --crtc 1
niendo: nothing happened...
yangman: pastebin xradr --verbose please
yangman: it's a laptop?
yangman: and you've already turned PANEL off?
niendo: but it worked before updating the xserver
niendo: yes I needed it
yangman: xrandr --output VGA_1 --mode 1280x1024 --left-of DVI-D_1 --crtc 1
yangman: looks like I gave you the wrong string for VGA_1 last time, if you just copy-n-pasted it
niendo: now both screens became black for a while and thats all
niendo: on both screens still the same output
niendo: xrandr --output DVI-D_1 --left-of VGA_1 --crtc 1
niendo: is working...
niendo: X Error of failed request: BadMatch (invalid parameter attributes)
niendo: Major opcode of failed request: 155 (RANDR)
niendo: Minor opcode of failed request: 21 ()
niendo: Serial number of failed request: 21
niendo: Current serial number in output stream: 21
niendo: it gives this error, but I have 2 different screens
yangman: that's pretty odd
yangman: but, at least something works ;)
niendo: using the radeon driver, there were some Xinerama options in KDE enabled
niendo: is there a way to get this stuff with radeonhd?
niendo: yes, thanks for your help...
yangman: it's all done with randr now
yangman: explicitly enabling xinerama may actually break things
yangman: modern WMs should be able to deal with randr-driven multi screens
niendo: yes, but using radeon I could maximize windows only to one screen
yangman: the driver shouldn't make a difference
yangman: I'd say the WM is being dumb
yangman: it's worked perfectly here for months
yangman: with xfce wm then metacity
niendo: maybe multiscreen configuration must be set up in xorg.conf in kde..?
yangman: I don't know. very unlikely, though
niendo: kde needs a "Section ?Extensions? Option ?Composite? ?Enable? EndSection" in xorg.conf
niendo: then it works
niendo: thats great
yangman: that's... weird
yangman: Composite shouldn't have anything to do with this
niendo: yes, maybe its unrelated - now its working in this xsession, too ..
niendo: anyway, thanks for the time and hints
cxo: just looking at Juliens patch
cxo: +test "x$XSERVER_LIBPCIACCESS" = xyes)
cxo: thats really silly, there is no need for the "x" if you use quotes
cxo: either it should be...
cxo: 1) test x$XSERVER_LIBPCIACCESS = xyes) or
cxo: 2) test "$XSERVER_LIBPCIACCESS" = yes)
cxo: that sort of stuff makes things noisy and when autoconf n00bs come in to update the file they do the same thing, stop the ugliness
legume: yangman: FWIW My problem with games and HD having DGA was nothing to do with the build.
legume: I just needed to use either game specific cvars or export SDL_VIDEO_X11_DGAMOUSE=0 and they all work now.
legume: FPS is lower than Radeon, is this expected with CP?
yangman: ah, SDL-related
yangman: is it significantly lower or just slightly?
legume: yangman: 65 -> 40 openarena anholt demo
legume: yangman 85 -> 57 urbanterror tutorial as timedemo
yangman: hm. that's quite a bit.
legume: yangman: it's an AGP card if that makes any difference
airlied: its wierd since radeonhd is designed for the hardware, whereas radeon is a hack. sorry I couldn't resist..
yangman: airlied: :p
airlied: legume: you getting correct AGP mode in all casee?
airlied: I think dmesg will tell you.
yangman: legume: honestly, I've no idea. you'll have to wait to hear back from someone that does
legume: airlied: I believe so both at 4x if that's what you mean
airlied: hmm so not that, not really sure what else could affect 3D the DDX hasn't a hue amount to do with it.
airlied: mtrr's being setup?
MostAwesomeDude: Anybody here running Mesa from git, and bemoaning the lack of fog coords? I have patches you can test. :3
legume: airlied: I'll have to reboot to check BRB
legume: airlied: Maybe a problem?
legume: reg00: base=0x00000000 ( 0MB), size=1024MB: write-back, count=1
legume: reg01: base=0xe0000000 (3584MB), size= 64MB: write-combining, count=2
legume: reg02: base=0xd0000000 (3328MB), size= 256MB: write-combining, count=2
legume: [drm] Can't use AGP base @0xe0000000, won't fit
legume: airlied: Checks logs - Looks like that's the same for radeon anyway
Obscene_CNN: heh, I thought I might have come up with a way to speed up the CS and wasn't getting any real changes in speed when I tested it. So I grep'd for RHDCSWrite and RHDCSRegWrite and low and behold they are all in functions that aren't used by my RS780 :P
airlied: Obscene_CNN: rs780 is r600 you don't have any CS usage.
Obscene_CNN: I noticed. I thought it would drop back to using the mmio CS code though.
airlied: Obscene_CNN: nope you have not accel at all
airlied: mmio is still accel.
Obscene_CNN: however all the CS stuff is in XAA EXA video etc
Obscene_CNN: oh well at least my macros replacements for _RHDRegRead, _RHDRegWrite, and _RHDRegMask made some very minor improvements to other things.
Obscene_CNN: why my mouse cursor acceleration flat out halls ass now ;-)
Obscene_CNN: hulls even
Obscene_CNN: <-- can't think
cxo: Obscene_CNN, maybe it saw a cat?
ndim: double mouse events perhaps?
cxo: scurry cat
tikenja: hi all
cxo: Dont work so hard penguin http://www.glasbergen.com/images/g800.gif
tikenja: i just want to know if its normal with radeonhd driver and r5xx card i ve just 3500 fps on glxgears
tikenja: and OpenGL renderer string: Mesa DRI R300 20060815 TCL
cxo: Life sucks, doesnt it?
tikenja: why not r500???
cxo: my 9600 radeon gets that too
cxo: 64mb agp
airlied: glxgears is not a benchmark.
cxo: thats my line, dammit
tikenja: yes i know but to know more o less performance
cxo: its more of cpu performance actually
tikenja: ah ok
cxo: and bus performance
tikenja: so how can i test well 3d performance?
tikenja: a game?
tikenja: its free?
cxo: timedemo 1; demo demo001 etc..
cxo: yes the demo is
cxo: quake3 is completely free
cxo: but quake3 is ancient
tikenja: but better test with q4 no?
cxo: quake4 will be a more meaningful benchmark, but i dont think the radeonhd driver is good enough for quake4
airlied: openarena is probably better.
tikenja: but i win 1000 fps on glxgears with radeon strange no?
tikenja: or maybe ut2004
cxo: ut2k4 is old
cxo: openarena is just quake3, may only require the primitive support currently available
tikenja: but i cant have dri r500 on opengl may that inprove peformance
tikenja: more than r300
airlied: tikenja: we just wondered why radeonhd would be slower than radeon on 3D.
airlied: maybe phoronix could do an article on it :)
cxo: ati lova phoronix dot com
napoleon: any idea if an ati hd4850 falls into this driver-usage category?
Tanktalus: nooob: I'm pretty sure it does ... though I think it's at the top end, thus the least amount done. I can't remember, and the wiki isn't loading...
cxo: its as supported as your mother's tits
cxo: ok i think that didnt quite come out right, i mean it doesnt have good support