Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

Radeon IRC Logs For 2009-6-24

Search This Log:

King_InuYasha: can the ATI Radeon HD 4890 run compiz with multiple monitors?
King_InuYasha: using either the radeon, radeonhd, or fglrx drivers?
airlied: fglrx probably can, open can't
muep: yet :-)
erilliam: When is r700 expected to get KMS?
MostAwesomeDude: Right after r600 gets it, probably.
erilliam: When is r600 getting it? :)
MostAwesomeDude: No idea.
erilliam: I saw your post on slashdot about it, but weren't you talking about it happening very shortly?
MostAwesomeDude: Yeah, a guy has been workin' on it, and it shouldn't be too difficult.
MostAwesomeDude: But it's going to need the memory manager hooked up first, which will be a bit difficult.
erilliam: fair enough. Thanks for your time :)
MostAwesomeDude: But it's at the bottom of people's lists, so I doubt it'll be in the next few days.
erilliam: Oh btw, will proprietary drivers ever get KMS?
erilliam: or is there something restricting (licensing?) their ability to?
MostAwesomeDude: Well, most of the KMS API is generic, so I doubt it'll ever show up since it would mean that those drivers could be used in conjuncction with open drivers.
erilliam: (thinking about nvidia proprietary and fglrx here)
MostAwesomeDude: Those drivers *do* use KMS-like technologies, just not in a way that's compatible with our stuff.
erilliam: Cool. Thank you
rah: MostAwesomeDude: "it's at the bottom of people's lists"?
Batou: Hmm, I noticed that the Radeon Feature chart has R700 Powerplay rated at 'Mostly', do any of the drivers slow down the fan yet?
rah: MostAwesomeDude: what's above it?
Batou: R700's fan is crazy loud and I'd love to dial it down.
nanonyme: Batou: Mostly is relative. :)
nanonyme: I got the impression you need KMS for even writing a decent power management.
Batou: Aw.
Batou: It sounds like a jet is taking off inside of my case right now.
Batou: It's nice and quiet in Windows.
airlied: you can add Option "ForceLowPowerMode" "true"
Batou: I did. It didn't seem to do anything on my driver version.
Batou: Do you know which version adds support for that for R700?
Batou: (WW) RADEON(0): Option "ForceLowPowerMode" is not used
Batou: (WW) RADEON(0): Option "DynamicPM" is not used
Batou: (WW) RADEON(0): Option "ClockGating" is not used
Batou: Says my Xorg.0.log
chithead: Batou: forcelowpower is not supported in any release, you need to build from git
Batou: brdgman on phoronix mentioned it but said it just lowered the PCI-E bandwidth, does it also quiet the fan?
Batou: If it does that's worth building from git for me.
Batou: Also, does the git release work ok on xorg-server
Batou: Or would I want to build xorg from git as well?
Batou: Any idea?
glisse: Batou: work okay on 1.6.x
Batou: Any idea on the fan thing? That's the bigger issue.
Batou: Saving power's nice and all but it's the fan that's driving me crazy.
glisse: Batou: dunno, i think someboard are capable of auto slowering fan if powerconsumption goes down
glisse: Batou: we will definitly get to the powermanagement and fan speed control at one point
glisse: thing is we have a lot of things we need to finish first
Batou: Well, ok. I dunno, An R700 fan going at full blast seems like a pretty big usability problem to me though lol.
Batou: It affects even 2d applications.
Batou: tomshardware measuerd the R700 at 52.2dB noise in 3d mode at low temps... 62.2dB if it breaks 80C. I'm not sure where it is in Linux but it's definitely louder than Window's default 3d mode.
stalkerg: what about r600-gallium?
stalkerg: maybe somebody write template aka r300-gallium?
MostAwesomeDude: stalkerg: Just as soon as the kernel supports r600 CS.
stalkerg: hm...
stalkerg: if use drm from git?
rah: MostAwesomeDude: what's above r700 KMS in "people's lists"?
Batou: Any idea how far off KMS is for R600/700? More curious than anything, the fan is a bigger deal but the progress of the ATi drivers has been pretty quick and interesting lately.
Batou: lol, good timing rah.
rah: heh
logari81: how can I generally find out which missing GL feature causes this "WARNING! Falling back to software for render mode != GL_RENDER"
rah: http://pkl.net/~node/teasel/photos/pc/radiator.jpg
rah: this is my r700 :)
rah: the fan is always at the same speed and it's very quiet
glisse: it means that on the kms todo list r6xx/r7xx is not at the top
rah: glisse: what is at the top?
hifi: rah: that should be fanless?
rah: glisse: and what is below the top, between the top and r6xx/r7xx? :)
rah: hifi: "should" be? :)
Batou: Nice. I'm on the stock fan and it's amazingly loud.
hifi: the fan looks like it was glued on top of the heatsink :)
Batou: It's not a problem when PowerPlay is working.
rah: hifi: there's a big clip all down the side of the fan
rah: hifi: surely you can't miss it?
hifi: sure, but it just doesn't look something that was shipped with the card
hifi: thats not cool!
rah: it was shipped with the heatsink :)
rah: the heatsink was not shipped with the card :)
hifi: wheres a big boobed animated girl :<
rah: erm..
hifi: (joking about the marketing)
rah: ok..
stalkerg: if i buy notebook with hd4330 me need start write driver. ^_^
rah: pokes glisse
rah: :(
MostAwesomeDude: I'm fairly certain that the top of the list is bug-fixing, followed by bug fixes and including other bug fixes.
MostAwesomeDude: logari81: You're probably in Blender? At any rate, it's probably GL_SELECT.
logari81: MostAwesomeDude: I am not in blender, I am in gmsh (mesh generator for computational engineering), I use a mesa version from recent git, but what I would actually like to know is if there is a systematic way to diagnose if it is GL_SELECT or whatever
MostAwesomeDude: logari81: Well, if the GL mode isn't GL_RENDER, it'll give you that fallback message.
MostAwesomeDude: And it's almost always GL_SELECT.
Vash63: Hmm. So 'radeon' from git crashes my xorg.
Vash63: The overlay I brought it from updated mesa as well, so it could be that also.
logari81: MostAwesomeDude: ok it is clear, I had just misinterpreted the message
MostAwesomeDude: logari81: Yeah. It's kind of a hard bug to fix; classic Mesa can't really do anything about it. Gallium will (eventually) have a builtin fallback that tries to keep accelerating non-GL_RENDER ops.
Vash63: http://vash63.googlepages.com/Xorg.0.log
Vash63: Apperently the file was too big for dpaste.
Vash63: (EE) RADEON(1): No connected devices found!
Vash63: finished all detect
Vash63: before xf86InitialConfiguration
Vash63: (WW) RADEON(1): No outputs definitely connected, trying again...
Vash63: That appears to be the error part, then it just starts repeating.
Vash63: Anyway, I gotta sleep. I might put up a bug report tomorrow or something if it makes sense to.
Zambezi: adamk: I'm back in Debian Lenny now after major failure in Debian Squeeze, so I'm going to give it a shot again after basic setup and restoring are complete. Squeeze went nuts when I installed drivers for my scanner.
Zajec3: can list_for_each_entry(connector, &dev->mode_config.connector_list, head) crash driver? if so, when?
Zajec3: ibelive it happens for me in
Zajec3: radeon_encoders.c:radeon_get_connector_for_encoder(struct drm_encoder *encoder)
Zajec3: how can I make DRM_DEBUG() display?
Zajec2: debug=1
Zajec2: i was told to use debug=8... it doesn't work ;/
Zajec2: oh, and now DRM_DEBUG_MODE doesn't work ;/
glisse: Zajec2: debug=0xf should work
glisse: debug 8 is to debug modesetting only
Zajec2: ahh, finally
Zajec2: thanks
glisse: and most distrib they only appear in dmesg
glisse: not in /var/log/messages
Zajec2: yup, i'm checking dmesg
Zajec2: via ssh
suokko: /var/log/debug?
Zajec2: [drm:drm_setup_crtcs], picking CRTCs for 8192x8192 config
Zajec2: [drm:drm_setup_crtcs], desired mode 1600x900 set on crtc 3
Zajec2: is this normal? using crtc 3? normally it doesn't exist
Zajec3: i managed to fix KMS to not crash...
Zajec3: but mode set by KMS is quite weird: http://estudent.put.poznan.pl/rafal.milecki/dsc00219.jpg
Zajec3: it's console (fbcon)
Zajec3: heelp ;) can someone help we with following picture: http://estudent.put.poznan.pl/rafal.milecki/dsc00219.jpg
Zajec3: any idea what may cause that?
Zajec3: it's supposed to be console (fbcon)
suokko: One possible explenation: fbcon renders using wrong resolution/depth to framebuffer. But I don't know how it can be that kind of result
moo_: very artistic...
Zajec3: moo_: ;)
Zajec3: suokko: hm, maybe... have no idea how to check that
suokko: Make fbcon write to log resulution/color depth every time it receives any. And also make radeon output what modes it is setting
Zajec3: is there something like debug for fbcon?
Zajec3: can I enable it, if I compiled fbon into kernel (not as module)?
ajavid: hi, I am on debian stable and using x1900xt radeon driver on X
ajavid: is it possible to get 1680x1050 resolution in cli mode using the fb from radeon?
suokko: Zajec3: Just use printk when there isn't any ;)
ajavid: because regular vesafb only goes as high as 1600x1200 and Iuse 1024x768 (vga791)
ajavid: this just doesn't look good, fonts are wide and nasty looking
ajavid: 1600x1200 same problem I face
ajavid: my main question is this, how can I get 1680x1050 resolution in cli mode? is this even possible?
suokko: ajavid: i think stable driver can not do that. But KMS could do it if you get unstable packages from somewhere. (You could try radeon-kms packages for Ubuntu)
ajavid: suokko, actually
ajavid: suokko, I'm not sure if it matters, but the radeon driver in X and all of Xorg on stable debian I build myself from sources of 7.4
ajavid: I have free 3d on x1900xt on X
ajavid: suokko, could you kindly point me in the directios of relevant reading material
suokko: ajavid: You need at least 2.6.31 kernel and mesa 7.6 and special bracnh for xf86-video-ati
ajavid: ahhh
ajavid: ok, well that sucks
ajavid: no fb for me!
suokko: http://jglisse.livejournal.com/ This might be good source of info but git branch instructions are a bit outdated
ajavid: for a while at least I'm on .29, I'd have to get a new kernel, and rebuilt Xorg AGAIN!
ajavid: :(
ajavid: suokko, thanks very much bro, I apprecaite your help, I'll think on it in deep
ajavid: :)
suokko: ajavid: newer info is at http://airlied.livejournal.com/
ajavid: thank you
aberres: hey guys :)
aberres: is the radeon driver supposed to support hdmi? the monitor does not get a signal it really likes. either it completely refuses to show something, or the image is "scrambled" and has weird colors
zhasha: MostAwesomeDude: any recent developments I should know aboot?
MostAwesomeDude: zhasha: I don't think so.
spstarr: we don't say aboot in Canada
tlp: more like "a boat"
mjr: you're trying to trick us, you wily canadian you; we've all seen South Park
mjr: though while on that topic, the nuking of Finland was _slightly_ exaggarated
mjr: some of us managed to stay alive in shelters far from population centers
mcgreg: aberres: it doest support hdmi. I use a radeon 4670 with it
aberres: mcgreg: ok, good. so it is just some "modesetting" (or whatever is the correct name) which goes wrong. to bad that i do not really have time to look into it the next two and a half weeks, but as i have no external monitor most of the time this is not that much of a problem. but afterwards i really must find a way to solve this
aberres: is is a hd 2600 mobile btw
aberres: mcgreg: did you setup something special, or just enable it with xrandr=
mcgreg: nothing... it just works
mcgreg: normally like vga ... it works since 2 years
mcgreg: simply plugin.. done
mcgreg: using a hp w2207h monitor, 22"
mcgreg: my old gfx card has DVI .. so I needed a adapter, now the new radeon 4670 has hmdi directy
aberres: sounds good. maybe it is related to the fact, that i have a notebook here
aberres: i did not try to plug in the cable before booting though
mcgreg: hmmm
mjr: dvi<->hdmi adapter no big deal though
mcgreg: mjr: no, it wasnt
mcgreg: 5 euro .. worked great
mcgreg: actually xrandr uses auto detect. it doesnt matter when you plug it in
mcgreg: nromally
mjr: mcgreg, yeah, it's just a mechanical adapter, the base hdmi signaling is the same, though with interleaved audio and higher possible clock rates and that jazz
mcgreg: indeed
aberres: yeah. it is seeing the monitor just fine (ok, it does not detect it native 1900x1200 resolution, but just smaller ones, but i guess i have a to small virtual screen)
aberres: anyway: if anyone has an idea, of a tip to try, tell me. i will be AFK for the next about 24 hours though. but i will read backlog for sure
mcgreg: hmm I didnt have any problem detecting the screen shize as well. native resolution 1680x1050 worked out of the house, thought the monitor does a emulated 1920x1080 too
chithead: aberres: problems with 1920x1200 and hdmi are sometimes due to old hdmi standard and reduced blanking
aberres: the max resolution is determined by the internal tft. with a little tweaking of xorg.conf this should be gone. i have been two month not at home so i had not much time to try yet
aberres: chithead: and with lower resolutions?
chithead: lower resolutions should not be affected
aberres: so hopefully it is something else. the hardware is not that old. a hp8510p and a relatively new hyundai
DanaG: Hmm, I fired up the old desktop... and for some reason, radeon kms was causing horrid null-pointer exceptions and segfaults an "no block devices found", oh my!
DanaG: I expected poor performance, perhaps... but not null pointer dereferences.
DanaG: For example, it gave me EIP: [<(null)>] (null)
DanaG: What can I do now, to debug this?
Zambezi: adamk_: Can you check this pastebin now if it look more promising? http://pastebin.com/d5f11c04e I haven't edited xorg.conf yet.
daum: hey guys - anyone have any experience with the Radeon X1550 256MB 64-bit GDDR2 PCI Express x1 ? for dual monitor support i'm looking for
ajax: hmm. object_connector_convert looks egregiously wrong.
ajax: wonder what i'll break by fixing it.
MostAwesomeDude: ajax: Your style of thinking scares me sometimes. :3
MostAwesomeDude: daum: Dual monitor stuff should work fine.
ajax: MostAwesomeDude: it's paranoid, but they _are_ out to get me.
MostAwesomeDude: ajax: Probably. Anyway, IIUC pixmaps know where they're located relative to the internal geometry, right?
ajax: MostAwesomeDude: that's what ->screen_[xy] are for, yes
MostAwesomeDude: Trying to figure out if I need anything besides is_shattered and shards[] on the exa priv.
MostAwesomeDude: It just looks...too clean.
Zambezi: suokko: I'm back in Lenny now, but the command right-of jada jada behaved the same.
MostAwesomeDude: But I'm sure that nice clean feeling will go away once I start doing the crazy Xrender shatters.
suokko: Zambezi: What about left-of?
Zambezi: suokko: xrandr --output VGA_1 --right-of DVI-I_1/digital
suokko: I mean did you try if it worked better
Zambezi: suokko: With the last commando?
suokko: yes
Zambezi: suokko: Yes. VGA-monitor get "disabled". I can see the backgroundcolor, but I can't drag and drop frames to it.
suokko: That means you have dual head working. Then only problem is to find the screen. I did notice some clipping problems that prevented me from moving windows fully to 2nd desktop but they disppeared if I loged out and logedin
suokko: i did use gnome-dispaly-settings so all settings were persistent
Zambezi: suokko: Gnome won't come near my computer though. It's too slow for it.
Zambezi: suokko: I tried in Ratpoison just to exclude it's an Ion-problem, but same there. I can move around the mouse on the second monitor, but not open or put any window. And I'm trying to stay away from fglrx to support the comunity.
mroconnor: you know the firefox 3.5 from devnull really doesn't use a lot of mem
mroconnor: sorry wrong chan :(
DanaG: okay, so, compiz doesn't start right on R100.
Zambezi: mroconnor: And that's with Flash installed or not?
DanaG: To get Xorg to even start without just turning off the display and hard-locking, I had to downgrade to the non-KMS versions of things.
mroconnor: yes it is
Zambezi: mroconnor: Before Fx 3.5 hit my distro, there's already a new version out since ages.
mroconnor: itechnically what I have is 3.6 from HG repo
DanaG: http://pastebin.com/f1800debb
DanaG: ugh, why is glxinfo showing me as using swrast?
nanonyme: You seriously don't expect an answer to that? :/
DanaG: It's not showing anything such as "driver entry point failed"
soreau: What does X log say?
DanaG: http://pastebin.com/f1800debb
DanaG: that was last time I tried starting compiz.
DanaG: this time is:
DanaG: http://pastebin.com/f724da6d0
nanonyme: I wonder what causes those damn many EXA bugs in the former log.
soreau: #
soreau: (EE) RADEON(0): [dri] RADEONDRIGetVersion failed to open the DRM
soreau: #
soreau: [dri] Disabling DRI.
DanaG: radeon 357824 0
DanaG: drm 157984 2 radeon,i915
DanaG: [drm] Initialized radeon 1.30.0 20080528 for 0000:02:00.0 on minor 0
nanonyme: Checked privileges?
nanonyme: That is, that your user can actually use the device.
DanaG: ls: cannot access /dev/dri: No such file or directory
nanonyme: Oh.
DanaG: Before, when I was using the radeon-kms repo, it was just plain hard-locking, or giving various random memory corruption.
nanonyme: Is it the same if Intel DRM driver doesn't get loaded, btw?
nanonyme: Just wondering if the two are compatible.
DanaG: nope, no /dev/dri
DanaG: Not even an empty dir, in fact.
soreau: that node will be created if everything loaded properly. Clearly, that isn't the case
suokko: Nice memory usage for "ring": http://pastebin.com/m74b86d09
suokko: "1171m"
DanaG: hmm, now I need to figure out why, exactly, it's not working.
soreau: DanaG: Reinstall X, mesa and the driver for starters(?)
DanaG: That wouldn't explain the broken kernel part, though.
yangman: /dev/dri is created by the kernel drm modules
soreau: DanaG: Well there's your problem, you're kernel's b0rken ;)
DanaG: I am kernel's broken?
DanaG: is teasing for wrong "your"
DanaG: mm, got radeon working.
DanaG: Had to stop Xorg, unload drm and radeon, and then reload both.
DanaG: Might be good to add a warning: "reloading radeon with Xorg running likely won't have the desired effect."
DanaG: Oddly enough, it's giving me the same symptoms the R300 was giving me:
DanaG: Slow as shit normally... but fast when benchmark is running.
DanaG: hmm, what would make it act like that?
DanaG: And most specifically, wobbly is the thing most siginificantly affected.
bridgman: wobbly is slow or fast ?
bridgman: and which GPU ?
DanaG: Wobbly is slow
DanaG: 02:00.0 VGA c mpatible c ntr ller: A I echn l gies Inc Rade n RV200 QW [Rade n 500]
DanaG: argh
DanaG: damn slashexec
DanaG: 02:00.0 VGA c mpatible c ntr ller: A I echn l gies Inc Rade n RV200 QW [Rade n 500]
DanaG: Looks like it loses "o" and "T"
DanaG: Wobbly is slow, until I trigger benchmark.
bridgman: um.. that's what I have at work... I don't think it's fast enough to wobble windows ;)
DanaG: Try hitting "benchmark" -- you may be surprised.
DanaG: Uncheck "disable limiter", though.
bridgman: are you saying wobbling gets faster when a benchmark is running ? Compiz benchmark ?
DanaG: Yeah.
DanaG: Way faster.
bridgman: interesting, I'll try that tomorrow
DanaG: CPU is Athlon XP 2GHz.
bridgman: tres wierd
suokko: death by concurrency: load average: 424.50,
suokko: Even r200 is good enough for wobble windows if not much else is going on.
suokko: Too cache control makes memory manager very slow
DanaG: IN fact, even non-wobbly is slow as hell without benchmark running.
DanaG: 1600x1200 is a bit too much for 64 megs of VRAM. =P
DanaG: Too bad my CRT claims to want 1280x1024, not the correct 1280x960.
DanaG: xrandr doesn't show 960.
bridgman: so EDID says "give me 1280x1024" ?
suokko: 1600x1200 is only about 2M so 60M left after frame buffer :)
bridgman: 1600x1200 is 2M pixels, 8Mbytes, isn't it ?
bridgman: front, back, depth, we're at 24MB
suokko: yes
suokko: Unless 16bit colors ;)
suokko: then it is 4M :)
bridgman: couple of hundred billion here, couple of hundred billion there...
DanaG: yeah, Viewsonic seems to think CRTs should be driven at 5:4, even when they're actually 4:3.
DanaG: =P
chithead: the whole square pixel thing was not introduced until ibm vga
DanaG: Yeah, but this monitor is from, oh, the Athlon XP era. =P
DanaG: Hmm, what would make Xorg eat CPU just drawing the desktop, now?
DanaG: It's roughly 30%.
suokko: KMS?
suokko: agp card?
suokko: try setting it to pci mode if yes to both
DanaG: It's an AGP card; no KMS, because the KMS kernels were giving null pointer dereferences, and all sorts of stuff.
bridgman: yeah, I remember that; IBM introduced the "professional graphics adapter" first, and I thought that was great; actual square pixels on a PC
bridgman: then vga came along shortly afterwards IIRC
DanaG: Whoever created 1280x1024... should be slapped.
DanaG: I'm also not sure which MESA I'm using.
DanaG: er, Mesa.
hatseflats: DanaG: what's wrong with that resolution?
DanaG: It doesn't match the aspect ratio of this screen, for one.
hatseflats: well, that's one :)
DanaG: It also doesn't match the aspect ratio of any other screen resolutions AT ALL.
DanaG: oh yeah, /me plugs the Radeon 7500's fan back in... it was getting hot enough to make the heatsink "pushable"
hatseflats: DanaG: yes it does, QSXGA
DanaG: (that is, it was softening the thermal pad.)
hatseflats: 2560x2048 :)
DanaG: oh, and it also sucks that nobody in the whole damned world makes a 147-DPI desktop LCD.
rah: quit the chitchat and get r7xx working with KMS!
DanaG: 123456 890-=AaBbC DdEeFfG HhIiJjKkLl mNnO PpQqRrSs tUuVvWwXxYyZz
DanaG: gotta' love the /exec plugin
chithead: DanaG: ibm used to make the t221, but it is out of production now
rah: (I paid good money for my ATI card so I could bitch at people on IRC)
hatseflats: sorry, I don't read base-sixty-mongled, DanaG.
DanaG: should have been 1234567890-=AaBbCcDdEeFfGgHhIiJjKkLlMmNnOoPpQqRrSsTtUuVvWwXxYyZz
rah: DanaG: what are you execing?
rah: and don't say "echo 1234567890-=AaBbCcDdEeFfGgHhIiJjKkLlMmNnOoPpQqRrSsTtUuVvWwXxYyZz"
DanaG: yes, that's exactly it.
rah: Thu Jun 25 00:02:13 BST 2009
rah: that's about the only use I've ever found for /exec
DanaG: i.e. it loses 7, c, g, M, o, T.
rah: umm
rah: "Oooh, that thing has numbers on it!"
DanaG: Pidgin irchelper.
rah: that's an interesting name for an IRC client
DanaG: Lets you set custom response.
rah: I see
DanaG: I hope you get the reference. =P
rah: and also has /exec with added Fail
DanaG: Or rather, "IRC More", not "IRC Helper"
rah: I don't get it I'm afraid
rah: what's it from?
DanaG: Portal.
DanaG: The "curiosity sphere"
rah: that's a game isn't it?
DanaG: yeah.
rah: I see
DanaG: Guess I shouldn't assume everyone plays games -- especially Windows ones. =P
ajax: they made portal for windows?
ajax: i assumed it was a ps3 release
rah: god it's a *windows* game?
rah: !cool
bridgman: DanaG; in fairness, back in the good old days when people made 1280x1024 monitors they actually *had* a 5:4 aspect ratio ;)
bridgman: and if you wanted 4:3 you went with 1152x864 and saved about a thousand dollars on video RAM ;)
mjr: isn't that actually in the good new days?
chithead: or 1280x960 if you had more than 2mb vram
mjr: I don't think I've seen 1280x1024 used except on 4:3 CRTs or 5:4 LCD:s
mjr: but maybe I've missed something
bridgman: in the early 90s 1280x1024 was mostly 1 bit-per-pixel due to the cost of video ram
bridgman: there were some 8 bit and 24 bit cards but they were $$$
bridgman: and we used to wrap our feet in old newspapers to stay warm ;)
airlied: still remembers sun pricing for 24-bit cards.
ajax: remember 8/24 packed overlays? yeah. that was a mistake.
bridgman: yeah, there were a lot of bad paths chosen back then
bridgman: impossibly complex approaches to save relatively few $$
bridgman: it's amazing those companies are still around
bridgman: oh wait...
airlied: bridgman: some of them are unfortuantely still around :)
mjg59: ajax: Remember apps that required them?
ajax: all too vividly
airlied: misses colormap flashing.
ajax: i think maya still does, or at least reeeeally wants an rgb overlay
mjg59: http://www.amiga-hardware.com/showhardware.cgi?HARDID=863 was probably one of the weirder high-res approaches of the era
ajax: i should finish pseudocolor visual emulation
airlied: ajax: can we get a maya license?
ajax: probably.
ajax: you could probably get away with the trial, for that matter
MostAwesomeDude: Why is Maya needed?
airlied: because it probably has no hope of running on open source drivesr
airlied: MostAwesomeDude: clearly it needs r300 frag shaders :-)
MostAwesomeDude: airlied: Gotta find somebody else, I'm already busy. :3
airlied: ajax: the trial is XP only
ajax: aaw
mjg59: Only other 8+24 app I've seen was some Irix scientific visualisation code
mjg59: We couldn't get it working on MGA, even though it claimed to support that
airlied: ajax: we just need 3D spice :)
bridgman: we have a big vat of 3D spice, how much do you need ?
bridgman: although I guess we should use some on 6xx/7xx mesa first ;)
airlied: bridgman: you kicked the rs690 ppl yet? :-)
bridgman: yeah, but haven't heard back yet
airlied: keep kicking then :-)
bridgman: will do
AndrewR: bridgman, you actually remember PGA? (professional graphics adapter). I only very recently learned about this thing ;)
bridgman: I'm kinda old
bridgman: I was designing graphics hardware before PGA came out
bridgman: which is scary
rgrasell1: hey i have a problem. I have a radeon x1600, running on ubuntu 9,04. Whenever I use anything opengl intensive, my xserver freezes. Cn anyone help
rgrasell1: ?
bridgman: (remember to mention that you're running on a macbook pro ;))
rgrasell1: lol. Im running on a 1st generation macbook pro.
rgrasell1: For example: if i open glxgears, it'll run fine for a few seconds, then xserver will freeze. And i cant force close it or anything
hatseflats: awesome, mesa-git breaks on the spot :p
hatseflats: ya think mesa'd at least build straight out of git
MostAwesomeDude: hatseflats: I almost guarantee some part of your userspace is too old.
hatseflats: really? won't argue that, I wonder what is though
hatseflats: running testing-tree of gentoo, updated just a few hours ago
hatseflats: you'd think that's relatively up-to-date, by some standard anyway
MostAwesomeDude: F11 builds it fine.
bridgman: http://incolor.inebraska.com/bill_r/ibm_pga_graphics_adapter.htm
hatseflats: configs, I'll show mine if you show yours ;)
MostAwesomeDude: hatseflats: Which driver isn't building?
hatseflats: I'm getting a "xDRI2CopyRegionReq has no member named `bitmask`" in dri2.c
AndrewR: bridgman, http://forums.nekochan.net/viewtopic.php?f=19&t=16721195
suokko: isn't sure that bug in Jaunty is fixed yet.
AndrewR: bridgman, but i have strange, may be not very correct question ...
suokko: There is quite many reports different gl apps freezing in many r300-r500 cards
rgrasell1: suokko: is there a fix or orkaround?
AndrewR: bridgman, from wiki article about Kurt Akely ... Kurt Akeley, The Silicon Graphics 4D/240GTX Superworkstation, IEEE Computer Graphics and Applications, July 1989, pp. 71-83. May be you already have pdf? Or i will just wait few more months ....
hatseflats: MostAwesomeDude: ehm, in a ati build, got any idea what'd it might be?
hatseflats: 'xcept a general jfgi :)
suokko: rgrasell1: you can try: https://launchpad.net/~ubuntu-x-swat/+archive/x-updates
rgrasell1: suokko: thanks. that hsould help a lot
suokko: I'm not sure if they help but should be worth trying
MostAwesomeDude: hatseflats: Your dri2proto is out of date.
hatseflats: 2.1
hatseflats: lessee if I can crank that up to rev HEAD
DanaG: Hmm, I got KMS to work... but trying to start Xorg simply results in a hard-lock.
AndrewR: DanaG, for me Option "renderAccel" "0" and/or nopat in kernel command line helps a lot
AndrewR: rv280 KMS
DanaG: nopat? Hmm, I can give that a try.
hatseflats: MostAwesomeDude: can safely say that that's not it, I just installed a clone of the git head of gti2proto, same error
DanaG: I wish AMD had something like the HP business netbook... but with the Athlon Neo.
hatseflats: s/gti/dri/ of course
DanaG: Oh, I see............
DanaG: I was using mismatched packages. Ooops.
bridgman: I think Neo is just in the dv2 right now; what other stuff is missing ?
DanaG: Note to self: when using xinput2 repo AND radeon-kms repo... make sure all the ATI stuff is from the latter.
bridgman: then again I think there are a bazillion different dv2s
DanaG: Look at the Mini 2140 for a comparison -- especially the high-res one.
bridgman: ok, hold on
airlied: xinput2 is all in master now
DanaG: 10", 6-cell battery, available in 1366x768.
DanaG: NOw that I've used 147 DPI... I refuse to buy lower-dpi ever again. =P
hatseflats: oh my
hatseflats: such high demands! p
hatseflats: s/p$//
bridgman: ah yes, everything else finest kind
hatseflats: darn this bloody keyboard
DanaG: Oh yeah, how do I tell if the main karmic repo will have the xinput2 in its main stuff?
DanaG: available versions of core are: and (omitting long suffixes) -- latter is the PPA.
hatseflats: ppa?
DanaG: Looks like karmic doesn't have xinput2. :(
DanaG: Personal Package Archive.
DanaG: I wonder... if a Radeon 7500 overheats... will it shut off... or just die permanently?
DanaG: The fan on this thing is damn annoying.
AndrewR: http://www.classiccmp.org/pipermail/cctalk/2002-July/003479.html (article from 1987 ... it was only 22 years ago .. so, you are not too old. Just computers moving forward (?) very fast)
AndrewR: bridgman,
DanaG: Hold on while I try the "correct" radeon xorg driver.
DanaG: failed to add fb
hatseflats: DanaG: had a friend who had a 7500 series overheat, died permanently, can't say if that's always the case tho
DanaG: http://pastebin.com/ffb2b73f
DanaG: yay for GPM!
DanaG: Copy and paste in console... sweet.
DanaG: [drm:radeon_cp_init_kms] *ERROR* invalid ioctl with kms radeon_cp_init_kms
DanaG: I see that twice.
DanaG: (dmesg, that one is.)
airlied: running a non-kms ddx on kms
DanaG: Hmm... radeon-gem-cs3 is the name of the DDX packagethis time.
DanaG: argh, something randomly keeps spewing "...done." to my console.
DanaG: ah, wrong PPA.
DanaG: nope, it's the right one.
DanaG: Okay, I'm boggled.
DanaG: Any other advice?
airlied: don't run with scissors?
MostAwesomeDude: Do run with scissors :3
MostAwesomeDude: Seriously, it sounds like you're trying to load the wrong DDX.
MostAwesomeDude: Maybe you installed a non-KMS DDX in /usr/local at some point?
DanaG: Is there anything I can grep for in "strings" of the radeon_drv.so?
DanaG: I see "initializing kms color map"
DanaG: hmm, maybe try running gdm, instead of just startx....
DanaG: ...nope.
MostAwesomeDude: DanaG: Did you check in your Xorg log to make sure you're loading the right radeon_drv.so?
DanaG: same one: /usr/lib/xorg/modules/drivers/radeon_drv.so
DanaG: hmm, should I manually specify an accelmethod?
DanaG: nope, no change.
DanaG: I think xorg is confused... it keeps saying... acceleration disabed, enabled, disabled, enabled.
DanaG: Which is it? =}
airlied: sounds like an old kms DDX maybe
DanaG: ah.
DanaG: Guess I should compile a new one myself.
DanaG: hmm, or the xinput2 ppa is breaking it all.
DanaG: I'll try reverting the xorg-core (and all that *cough*jazz*cough*) to the ones from kms repo.
_Groo_: hi/2 all
_Groo_: airlied: alex are you there?
DanaG: ah yeah, my xorg core is 1.7.
_Groo_: anyone using xf86ati from kms-support branch?
airlied: I'm dave :)
_Groo_: airlied: dave, alex :D
_Groo_: airlied: anyway, xf86 from your kms-support tree crashes X hard.. thats in the usual rs485 evil card
_Groo_: airlied: latest linus git tree, drm master, mesa master
airlied: in kms mode or not?
_Groo_: xorg
_Groo_: kms mode
airlied: yeah rs485 is broken until someone fixes it
_Groo_: airlied: my screen turns into striped colors , x tries the usual 3 times, and crashes
_Groo_: airlied: aaaaaaaarghhhhhhhh why no one cares for poor grooooo!!!!!!!!
_Groo_: airlied: no fbo, no X... might as well go back to phosphor monitors
airlied: I'd be fixing rs485 now but I've got a hangover
airlied: and its in the office
_Groo_: airlied: ¬¬
_Groo_: airlied: hangovers are good for coding
DanaG: IS using a CRT.
DanaG: A good one, too. 1600x1200 85Hz.
_Groo_: DanaG: nice :D
airlied: but they are useless for getting me to the office
_Groo_: airlied: didnt you say you had a rs485 inside a box at home?
DanaG: goes to try changed xorg.
_Groo_: airlied: anyway, when you fix rs485.. modesetting? fbos? please? snif..
airlied: it'll be kms, I've no idea why fbos are busted
_Groo_: airlied: the 28 worldwide users will apreciate it :)
_Groo_: or where 26.. hmmm
_Groo_: airlied: can i do any tests to help you debug fbo? besides the mesa/prog tests
_Groo_: airlied: im not even bother about the garbled screen bug...
_Groo_: airlied: i kinda got used to the idea the 26/8 users will never get suspend to ram again
airlied: the prog tests failing is pretty bad
airlied: since they don't fail on my non-rs480 hw
_Groo_: airlied: they run but they dont show what they are suposed to.. so.. they fail
_Groo_: in the mesa tests, rbos work, some crash and fbos fail to show what they are suposed to, if you want, tell me you email and tomorrow ill send you a complete regression test
_Groo_: airlied: hmm i believe i have you on my gmal...
_Groo_: gmail
_Groo_: airlied: im gonna test and send to you tomorrow
_Groo_: airlied: but the modeseting and xf86 would suffice for now.. :/
_Groo_: well anyway,... gotta go.. hope you cure your hangover soon.. and remember, a happy rs485 user is one less terrorist
DanaG: hey, it worked.
bridgman: how much damage are you gonna do with an rs485 ?
_Groo_: bridgman: do you really wanna call my rs485 powers?
_Groo_: bridgman: and airlied doesnt know that ;)
_Groo_: bridgman: and in this world economy AMD cant afford to exasperate all the 26/8 users!
DanaG: odd... vt-switching left me with my mouse on top of the console image.... I had to alt-f2 and xrandr rotate to get back Xorg.
bridgman: Groo; I agree, I take it back
bridgman: but still...
bridgman: ;)
airlied: bridgman: I have considered blowing up a few rs480s
bridgman: yes, I understand
DanaG: argh, goddamn metacity won't let compiz start.
airlied: I do have an rs480 here, but no cpu or RAM for it
airlied: I should do something evil to it
MostAwesomeDude: Please don't let "evil" -> "sending it to MAD"...
airlied: evil is probably getting a cpu/ram :)
bridgman: yeah, hook up with agd5f when he gets back
bridgman: we should have at least one 485 in capable hands
mjg59: airlied: Sigh. Am I going to need to hook you up with 480 *again*?
airlied: bridgman: oh I have one in the office
bridgman: Groo, ignore that, I'm going to stop typing I think ;)
airlied: mjg59: oh I forgot you had the craptop
airlied: mjg59: do you still have it?
mjg59: Yeah, somewhere
airlied: bridgman: it just doesn't have a case or HDD, so I have to sub it into one of my other systems
bridgman: (covering ears) la la la la la
bridgman: got it
airlied: ponders how to get craptop from mjg59
airlied: my travelling synergys are quite low these days
mjg59: KS?
mjg59: Some distance in the future, though...
airlied: probably, need to get some prices for that
DanaG: hmm, was using swrast.
mjg59: airlied: Likely to make XDS this year?
airlied: mjg59: unlikely, KS might be my only trip and only then because Tokyo
MostAwesomeDude: Tokyo!
mjg59: Yeah, not the most convenient destination
airlied: mjg59: better for me than most :)
mjg59: Pretty much the first time you've been able to say that...
DanaG: (EE) AIGLX error: Calling driver entry point failed
DanaG: (EE) AIGLX: reverting to software rendering
DanaG: (II) AIGLX: Screen 0 is not DRI capable
airlied: jeez who wrote this dma code, (note to self: I did)
DanaG: okay, that broken DRI is my last mystery.
airlied: no DRI2 in X server? or something like that
DanaG: http://pastebin.com/f7f55a21a
DanaG: xorg log.
airlied: wrong mesa then by the looks of it
airlied: or mesa not built against proper libdrm
DanaG: I see two options for mesa: 7.4.1-1ubuntu3, and 7.6.0~git20090621.something-0ubuntu0sarvatt2
DanaG: something being the commit id, perhaps?
airlied: probably the second one
DanaG: "something" is "df70d304"
DanaG: Last thing to do is to figure out if that other (dual-dvi) radeon 7500 I found at school... has at least the same amout of VRAM as mine.
DanaG: If so, I want to swap it -- mine has a godawful annoying fan; that one has passive cooling.
DanaG: hmm, I'll try that other mesa.
DanaG: Still swrast.
DanaG: dagswrast it! <--- lame pun. =P
DanaG: For some reason, it also gives me this extra modeline:
DanaG: (II) RADEON(0): Modeline "0x0"x0.0 0.00 0 0 0 0 0 0 0 0 (0.0 kHz)
cbmuser: glisse: the problem with firefox slowing down X completely when having kms enabled has gone for since today's dist-upgrade in ubuntu 9.10
DanaG: Now I just have to find out which mesa to compile.
DanaG: ...because the PPA one does not work.
airlied: DanaG: master
DanaG: Hmm, will that work on top of the KMS stuff I now have?
airlied: if you have libdrm built right
DanaG: I honestly have no idea if I do.
DanaG: http://pastebin.com/f582b2036
DanaG: that's dpkg -l *drm* *radeon*
airlied: if you have a libdrm_radeon installed it should be okay
DanaG: hmm, now I just have to figure out how the heck to switch branches to one that already exists.
DanaG: ah:
DanaG: git reset --hard master
DanaG: git checkout master
DanaG: that reset master to remove locally-modified files, so it wouldn't refuse to checkout.
DanaG: now I pull. =P
airlied: git clean -df
DanaG: ah, thanks.
DanaG: it was giving error: Untracked working tree file 'src/mesa/drivers/dri/radeon/radeon_bo_drm.h' would be overwritten by merge.
DanaG: What we really need is a single "switch to this branch, update it, and discard everything on my end" command -- with an "are you sure?????" confirmation. =P
DanaG: Just need the ./configure parameters now.
DanaG: anaG: Just need the ./configure parameters now. │nald ][ ▒│
DanaG: argh
DanaG: ) DanaG: argh │nald ][ ▒│
DanaG: ./configure --prefix=/usr/local --with-dri-drivers=radeon,r300
DanaG: ah, that
DanaG: 's what I meant to paste.
soreau: Looks like your trying to work with broken fingers there =)
DanaG: no, I'm trying to copy and paste... and sometimes you get different results between shift-ins and right-click,paste.
DanaG: gmake[5]: INSTALL@: Command not found
DanaG: lolwut? ={
airlied: okay torcs should be fixed in master
DanaG: (EE) AIGLX error: Calling driver entry point failed
DanaG: oops, hard to tell which window in finch is active.
DanaG: Cool, got help in #ubuntu+1. Thanks.
bkero: Has anybody had issue with getting rotation to work on an RV380?
DanaG: I'm curious what changed between R100 and RV200.
bkero: http://staff.osuosl.org/~bkero/radeon.jpg covers it. 6.12.1, gentoo, xorg-server-1.5.3, triple heading with the other two on an nvidia card. xrandr -o left scrambles everything, but it still responds to input and doesn't freeze
rgrasell1: What is EXA and UXA?
virtuald: rgrasell1: the current x acceleration architecture and the intel one (uxa)
DanaG: www.csc.calpoly.edu/~dgoyette/tdfx
DanaG: that's 3DFX thingy.
DanaG: Removing the card did not bring back NPOT support.
DanaG: odd.
DanaG: Not sure what's up there.
mjg59: tdfx is running 2D on the actual voodoo card, despite them not being intended for 2D
mjg59: I'm actually impressed that it has Xv support
DanaG: oh yeah, it's a voodoo3, actually.
DanaG: Has 2D and 3D.
DanaG: so, any ideas how to fix the "no NPOT support" issue?
DanaG: It worked fine without KMS... but now it's reporting lack of NPOT support.
mjg59: DanaG: Oh, I'm sorry, I'm confusing the tdfx and voodoo drivers
DanaG: now, how would you get 3D on the thing? It'd have to go to a low enough resolution right off-the-bat.
DanaG: Bigger issue is the NPOT thing.
DanaG: argh, where's my NPOT?
DanaG: oh, and: DRM version 1.0 too old to support HyperZ, disabling.
DanaG: http://users.csc.calpoly.edu/~dgoyette/Screenshot.png
DanaG: okay, yeah, I'm lost.
DanaG: + if ( sPriv->drmMinor < 13 ) + fprintf( stderr, "DRM version 1.%d too old to support HyperZ, " + "disabling.\n",sPriv->drmMinor ); + else
DanaG: how do I find out what version DRM actually is?