soreau: Here's the gdb info from when blackscreen-happens-but-compiz-doesn't-crash http://sprunge.us/iXZh when en/disabling plugins in ccsm using 0.9, I get black screen but compiz never crashes
mcgregor: happy easter to you all :)
LordMetroid: I hav switched to fglrx, the 2D isn't that great for instance autoscrolling in firefox can at times get stuck but at least it has 3D
[Enrico]: trolling even on easter? :) [just kidding]
Pallokala: is there any reason to upgrade libdrm from 2.4.17 when using git mesa and git ati-driver only?
[Enrico]: Pallokala: for what i uderstand libdrm is used to handle communication between userspace and drm so basically depends on what version of drm you are using afaik
twnqx: ohhh just the right guy :]
KotH: twnqx: what for?
KotH: i'm innocent, i havent done anything...yet :)
twnqx: a simple vhdl question :P
KotH: only if it's simple :)
KotH: has a question himself
KotH: i'm using a HD3850 in zaphood mode...
KotH: it works nearly perfect
twnqx: 1000" & "100011" & rmwZp & aluInT & aluRla, <-- is that a concatetion of the bits?
KotH: there is just one little issue: once i move the mouse to the second screen, it's stuck there
KotH: as i have no idea how the radeon driver is structured, could someone point me to the relevant .c files?
KotH: twnqx: yes
twnqx: sadly, no idea about your question... but does sound more like an xrandr/WM issue
KotH: nope, it's neither xrandr nor wm
KotH: i dont use xrander (not the zaphood mode up there), and the wm doesnt restrict where the mouse pointer moves, but the xserver
KotH: thinks it's a simple issue of a check not done correctly and would like to fix it...
KotH: but i'm lost in code ^^;
airlied: KotH: are you using a laptop or a mouse?
airlied: there was a bug in the X server with touchpads
KotH: normal ps/2 mouse
KotH: i'm kind of sure it is an radeon related bug, as it used to work with mga vid until i stopped using it in january, and radeon showed this behaviour already then (but there were more serious bugs in zaphood mode, so i didnt report this one)
LordMetroid: [Enrico]: I tried to stick with the open source drivers but I want to play Quake Wars
idletask: I'm using the latest kernel from drm-radeon-testing, git's libdrm and mesa, X server 1.7.6 and git radeon ddx... But the kernel reports that I use an "old and broken userspace, please consider updating mesa"??
idletask: How come?
jcristau: ignore it
[Enrico]: LordMetroid: that's ok, but radeon has 3d..... not perfect but there is
LordMetroid: Is 3D being worked on?
[Enrico]: LordMetroid: it is of course. actually you can run most common OSS games, like nexuiz and similar
KotH: what does "panning area" mean in the context of xorg?
[Enrico]: LordMetroid: just a note: in case you have ubuntu 9.10 then you can't do 3d with r600 and newer hardware. you need 10.04 or the xorg-edgers ppa
idletask: Ugh, rendering is definitely slower with the latest updates :(
RiotingPacifist: (EE) open /dev/fb0: No such file or directory // (EE) RADEON(0): Acceleration initialization failed
RiotingPacifist: i'm running 2.6.33, and xserver-xorg-video-radeon 1:6.12.192+git20100324 , the radeon driver loads with modesetting enables, any ideas as to why i get the error?
twnqx: RiotingPacifist you didn't configure your kernel for it?
RiotingPacifist: twnqx: i'm running a generic kernel but modsettings is enabled "[ 3.170998] [drm] radeon kernel modesetting enabled."
glisse: RiotingPacifist: you are missing fbcon i guess
RiotingPacifist: is that a kernel module?
RiotingPacifist: I've got an fbcon module, but loading it doesn't create /dev/fb0, is this a kernel/distro problem or is fbcon to do with radeon drivers?
BioTube: as long as fbcon's loaded while KMS is on it should work
RiotingPacifist: Does fbcon need to be loaded during boot? i've loaded the module and restarted xorg but glxgears still says software rendering
glisse: RiotingPacifist: LIBGL_DEBUG=verbose glxinfo | head should tell you what's wrong
RiotingPacifist: glisse: it's not even trying, goes straight to libGL: OpenDriver: trying /usr/lib/dri/tls/swrast_dri.so
maligor: mm.. is it normal to have strange lines show blink sometimes on the screen when using dynpm? (2.6.34rc3, old userspace which I'm in the process of rebuilding building)
mjg59: maligor: Right now, yes
dileX: mjg59: that appears randomly here. mostly after starting 3D apps or games (e.g. openarena)
maligor: dileX, I was the original question asker, and yeah, 3d apps seem to make it show up more
dileX: .33-git1 got rid of the gpu lock stuff and hoped this is now better
dileX: grr, .34-rc3-git1
dileX: maligor: you have that flickering also in VT? only coldstart makes it go away?
dileX: blinking - horizontal stripes - looks like screen is not syncing properly
maligor: well, if I had to guess, it's some sort of mode unsync in the driver from pm modes
dileX: would be good to have an app to reset gpu
maligor: oh well, I'll just reboot without it
spstarr: [ 8.142670] [drm] State 0 Default (default)
spstarr: [ 8.142671] [drm] 16 PCIE Lanes
spstarr: [ 8.142672] [drm] 1 Clock Mode(s)
spstarr: [ 8.142673] [drm] 0 engine/memory: 600000/700000
spstarr: [ 8.142678] [drm] radeon: power management initialized
spstarr: only one clock mode?
spstarr: glisse's > 256MB VRAM is not merged into .34 i see still
spstarr: hmm back to intel GPU... still not ready yet
twnqx: spstarr: sed s/== 4/>= 4/g in radeon_atombios.c
twnqx: (at least that was for me the solution)
spstarr: why isn't this official then yet?
twnqx: aka too new pm table version
twnqx: dunno :P
twnqx: also, who knows at which point the current code is outdated for a newer table version again
spstarr: i'll stay in intel gpu land for now
twnqx: i constantly have flickering sicne running the chip at 100mhz
RiotingPacifist: i've got a mobility 5000 series card, i've got radeon & fbcon loaded, but i think im running in software rendering
RiotingPacifist: glxinfo gives "libGL: OpenDriver: trying /usr/lib/dri/swrast_dri.so // display: :0 screen: 0 // direct rendering: Yes"
soreau: RiotingPacifist: That is because there is no 3D support for HD5xxx yet
DanaG: Is there an ETA on that support?
DanaG: Or at least, if it's in "get the docs out" stage, an ETA on the docs? =þ
BioTube: DanaG: IIRC, there's some accel code going through IP review
evil_core: XV scaling is broken
evil_core: looks really strange when scaled ovetr 2X
evil_core: BioTube: IP == Intelectual Property?
evil_core: then it means it will take forever :P
BioTube: evil_core: well, it's been about a month IIRC
evil_core: BioTube: I know, but many mowies has started playing normal after week, yesterday I found new problematic
evil_core: BioTube: but heres no fix yet?
BioTube: I haven't been following evergreen-related changes
evil_core: its r500
BioTube: I don't know what could cause that either
evil_core: so its not related problem?
evil_core: strange means, like movie is plit in lines
BioTube: not that I know of
BioTube: all I can suggest is bisection
evil_core: impressive how it looks
BioTube: you could always roll back a few days
evil_core: its probably not few days, but 1-2 months
evil_core: abd dunno iuf its kernel or xf86-video-ati
evil_core: and it doesnt affect 1080p movies, only some realy lowres, prabably with interlacing which plays smoothly with -vo gl
BioTube: scaling's probably ddx
NightNord: Greeting everyone. I have a grave news (at least for me) =)
NightNord: I've found an very old (1997) 3d game, I was playing on Riva TNT2
NightNord: It has no shaders. It has no particles. Most impressive feature that this engine uses is sprites
NightNord: And it lags. Still, my cpu may render it much faster
NightNord: How it could be?
evil_core: NightNord: what game?
NightNord: Parkan, Imperial Chronicles
NightNord: It's russian game, you probably don't know it =)
NightNord: It's played in wine, btw
evil_core: looks a bit like ooelite
evil_core: or how it was called
NightNord: Right you are. That is a game of ~same genre and age
NightNord: Space mode is not lagging so much, but FPS mode is awfull
evil_core: so maybe use native one
evil_core: are you rsure its not dmiux/crappy dsound.dll problem?
NightNord: evil_core: see maintainer =)
evil_core: if you did those steps, then it can be related to software rendering
evil_core: if not, then it maybe fallback to 2D
NightNord: lib32 mesa is on
NightNord: (from git)
BioTube: if it's hitting fallbacks often, it might be bottlenecking on the bandwidth
NightNord: Compiled today
evil_core: I dont have this game
evil_core: if its easy downloadable i could try playing with it
NightNord: Hm... I dunno about copyrights...
evil_core: if it support glide, I would play with it
evil_core: yay, you are from Russia, I am from Poland ;)
NightNord: Ah, ok =)
evil_core: many games on wine runs in D3D HEL mode, and are damny slow then
evil_core: dunno why and where HEL is realized(on wine side?)
evil_core: i.e. Dethkartz, you must run it with sw or glide mode
Thunderbird: likely the HEL you mean is just the name we advertise in directdraw / direct3d7
evil_core: Hardware Emulation Layer
evil_core: dunno, its also under wine
evil_core: and is selectable under soem games
Thunderbird: the name has no real meaning in wine
Thunderbird: older direct3d versions are more tricky to emulate
evil_core: its like full sw fallback
Thunderbird: we don't have a 3d software fallback at all
evil_core: it has
evil_core: if I choose HEL in some games, they got less than 1fps
Thunderbird: I have no idea what the game is doing then
evil_core: and are smooth without "HEL" in name
NightNord: Direct3D HAL it's named often
NightNord: Still, that's just a common name for 'running via direct3d'
NightNord: Not so much games using opengl =)
Pallokala: are these optimum kernel bootoptions for 2.6.34-rc3: "radeon.modeset=1 radeon.dynclks=1 radeon.dynpm=1 radeon.audio=0 radeon.new_pll=1"
Pallokala: optimum being: use all new powerthingys
evil_core: NightNord: HAL isnt HEL!
Wizzup: What does new_pll do, Pallokala ?
evil_core: but dunno in this case anyway ;)
Pallokala: parm: new_pll:Select new PLL code (int)
Pallokala: dunno what it really does ;)
Wizzup: man radeon I hope? :)
NightNord: Hm... audio=1 dynpm=-1 dynclks=-1 modeset=1 new_pll=-1
NightNord: Only modeset matches =)
Wizzup: I don't know where to find the manual on those settings
Pallokala: me neither, I enabled those on 2.6.34-rc2 (currently using) and didn
Pallokala: ...didn't really see any difference
Wizzup: dynpm makes a diff here
Wizzup: got about 40% more batt. life
NightNord: dympm - dynamic power managment?
NightNord: I has -1 and it's enabled (following dmesg)
Pallokala: I'm running desktop.
Wizzup: well it will also keep the card more cool when unused... I guess
NightNord: Driconf seems to be not working
NightNord: I can't disable hardware accel
soreau: anyone know why torcs racing seems to be so demanding?
soreau: I don't think I've ever had it running more than 1FPS with the radeon driver
NightNord: Ha. It seems that it was wine's problem. Sorry for bothering you, guys
NightNord: For some reason, it has much better perfomance with trace+wgl is on and redirect to file (or it will lag on output)
Thunderbird: next to general trace lines, there are also some additional code executed if you run using +wgl
Thunderbird: perhaps those additional glx calls (for obtaining some more debug info) are related
DanaG: I wish the X server upstream would adopt that "backclear instead of doing pointless readback from video RAM" patch.
DanaG: That window resize lags on most all computers I've tried it on.
adamk: I've only ever experienced it with fglrx.
adamk: And, as I understand it, it's almost fixed in fglrx.