flexoid: I installed Ubuntu and was very surprised
flexoid: glxgears show more than 5000 frames per 5 sec on my HD2600
virtuald: it's not a benchmark
flexoid: ok
flexoid: but on my archlinux i have only ~1000
flexoid: with KMS
flexoid: and I visually see the difference in speed
flexoid: what they have done in Ubuntu?
virtuald: i think they only packaged it
virtuald: hmm, ubuntu may be using UMS still
gtklocker: Well, try the [radeon] repo on archlinux to see the difference
flexoid: virtuald, kms i'm sure
virtuald: ok
flexoid: Can you advise me a good benchmark?
flexoid: gtklocker, what kind of difference i will see?
gtklocker: with -git drivers
gtklocker: you'll see BIG difference
gtklocker: Perry3D and I are doing a good work
gtklocker: we choose the kernels and packages version rightly
gtklocker: have a look
gtklocker: ;)
flexoid: gtklocker, can you give me a fool repo path?
flexoid: full* i'm sorry
gtklocker: http://bbs.archlinux.org/viewtopic.php?id=79509
flexoid: oh
flexoid: some time ago i use it... results are not so good as in ubuntu now...
flexoid: but i will try
twnqx: hm
twnqx: displayport connection detection is dead again :X
twnqx: hm, even manual override doesn't get me a screen :(
airlied: agd5f: btw I'm seeing nmis again on rv530
dgbaley27: I have two displays on DVI-0 and VGA-0. I want to disable VGA-0 at boot because otherwise my console is limited on DVI-0 to the size of VGA-0. I've been trying kernel command line options like video= and fbcon=.
evil_core: cg_shadows "2" makes impritant speed degradation in q3a
evil_core: depth buffer is not accelerated on r500?
evil_core: no depth, stencil
evil_core: r500 has hw accelerated stencil buffer?
otaylor: evil_core: yes
dgbaley27: I'm seeing in Xorg.0.log: Option "{AGPMode, BusType, DynamicPM, ClockGating}" is not used
dgbaley27: Does this mean they don't work with my card, or other options are overriding them, ...
adamk: dgbaley27: I know a number of Xorg options are not valid any more when using KMS.
adamk: I would assume those are all such options :-)
BioTube: they are
dgbaley27: Ok, I don't even know what's really necessary. But I did notice some tearing and I wanted to try to get rid of that.
[Enrico]: all those options should be passed to the radeon module instead when using KMS
dgbaley27: [Enrico]: in modinfo, I don't see an equivalent option. However, I do see radeon.agpmode does this option do anything when it's connected via an internal PCIE?
[Enrico]: dgbaley27: the driver should autodetect it correctly, so i suggest you to let the driver autodetect. anyway radeon.agpmode=0 will enable pci mode
evil_core: dgbaley27: pageflip is for avoiding tearing, supported up to r400
[Enrico]: dgbaley27: ehm i'm sorry -1 is PCI
dgbaley27: Ok, well i'm on r600 so it looks like i'm waiting
dgbaley27: Is pageflip part of one of the entries in the feature matrix?
[Enrico]: dgbaley27: i have no tear effects and i'm on r600 too. can you explain where you got tearing?
dgbaley27: I was watching a divx video fullscreen with VLC whilst dual monitors were enabled
dgbaley27: i have 256M ram enabled for the gpu
[Enrico]: dgbaley27: mhm i never tried vlc with radeon, but with mplayer it works very good (i use xv as video output)
dgbaley27: Do I need to do anything with xorg to have xv working?
[Enrico]: dgbaley27: the advantage of mplayer (indeed use a gui like smplayer or gnome-mplayer) is that it reimplements some xorg parts, when vlc relies only on xorg
[Enrico]: dgbaley27: no xv should be working out of the box you can check with xdpyinfo | grep XVideo
dgbaley27: Yeah I have it. I see that VLC lets me use XVideo too. I don't really know what it was using by default
[Enrico]: dgbaley27: by default xvideo is used afaik
amarsh04: I've been using vlc frequently on the machine with a radeon 3200hd on the motherboard - dvds may need de-interlacing but I haven't noticed tearing
evil_core: it can be also bug
evil_core: check hires h264 like 720p or 1080p
dgbaley27: Let me ask my real problem now: I have a monitor on VGA-0 and DVI-0. I'd like to disable VGA-0 at boot, otherwise the console on DVI-0 doesn't take up the whole monitor. Do you guys know how to do this, from the kernel command line perhaps?
evil_core: I got strange errors, movie going forward and backward in loop in some strange divx files
evil_core: you can do it at any time
dgbaley27: evil_core: most stuff looks good. It's probably the cpu, but i can even play fullscreen video off youtube now without any skipping
dgbaley27: evil_core: you mean disable the port at any time?
evil_core: and change size
dgbaley27: How?
evil_core: now at UMs, but it was somewhere in sys
evil_core: in fb0 dir
evil_core: I was playing with that yesterday for fun
dgbaley27: So are there some fbcon= options I can pass through grub? I figure if VGA-0 is disable when the KMS first kicks in, the whole 1680x1050 of my primary monitor will be used.
dgbaley27: right now, only 1280x768 of it are used, because of the TV on VGA-0. But I only rarely use the TV, so I just want it off
MrSunshine: hmm, what is the channel for gallium/mesa dev stuff? :)
Ke: this probably, if you are insterested in radeon specific parts
MrSunshine: well, its about how gallium uses drm or not
MrSunshine: noticed a "swdri" driver in the gallium tree
MrSunshine: does that free one from use of a drm kernel driver if one does not want to use one? :)
zhasha: MrSunshine: #dri-devel
MrSunshine: oh :)
zhasha: alternatively there's the mesa3d-dev mailing list
zhasha: err, mesa-dev now. they moved it
glisse: MrSunshine: it won't use hw acceleration thought
MrSunshine: glisse, but the actual driver will be inited and used?
glisse: it will use cpu to perform rendering
zhasha: MrSunshine: am I reading correctly that you want to use r300g but avoid drm?
aszlig: hullo
aszlig: what could that be possibly related to (https://redmoonstudios.org/~aszlig/fdbc/distorted.png <- starcraft 2 beta in wine, but happens on all 3d games in wine), mesa, ddx, dri, kernel?
airlied: dgbaley27: video=TV-1:d should disable the TV connector, if its called TV-1, it might be Svideo-1 or DIN-1
dgbaley27: airlied: VGA-0 is what i'm trying to disable, and no that isn't working. When I had an intel gpu, video=LVDS-0:d worked.
dgbaley27: but video=VGA-0:d is not working
dgbaley27: airlied: out of curiosity, do you know what actually processes that video= line? Is it the kernel? radeon? I assume it's a driver thing because it worked with intel and not ati
zhasha: aszlig: that looks like what happened when r300g was using linear buffers and I screwed up the clear function
aszlig: its actually on a r600
aszlig: so could this be a issue inside the kernel code?
airlied: dgbaley27: it might be VGA-1, drm core process it
airlied: dgbaley27: have a look in /sys/class/drm for the names
aszlig: or better: could this be a kernel issue at all, or can it be excluded from the list of canidates that may cause this?
dgbaley27: airlied: ahhhh that sucks. I was using the output from xrandr
dgbaley27: I'll be right back
dgbaley27: Thanks so much
dgbaley27: I don't have to bump my head any more when I want to use VGA
dgbaley27: Which is especially nice right now because i was hiking yesterday and a limb broke from a tree and cracked my head with pinpoint accuracy
airlied: ouch
NTU: you cracked your skull?
dgbaley27: not literally
dgbaley27: lots of blood though
NTU: sorry to hear.
dgbaley27: It's all good, I was walking closely with my grandfather and dog, so I'm actually really glad it was me.
twnqx: airlied: did something in d-r-t break displayport to your knowledge?
twnqx: it doesn't recognize when i plug my projector via dp->hdmi adapter
twnqx: and manual override, e.g. add mode and turn it on doesn't activate the output either
dgbaley27: airlied: are there other options for that video= flag? Because now I can't manually turn on VGA with xrandr
airlied: twnqx: http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=commit;h=aa9613916a461027fdade8661177660db0975806
airlied: revert that maybe
airlied: dgbaley27: wierd it should only affect fbcon
twnqx: gonna try
evil_core: http://lists.x.org/archives/xorg-driver-ati/2010-May/015597.html
evil_core: does it work for r500 too?
evil_core: does it mean pageflip?
evil_core: this glisse patch
dgbaley27: airlied:
dgbaley27: http://pastebin.com/HVwjBuwk
evil_core: This morning Jerome put out a single, 500-line patch that implements this useful functionality. This patch goes against the latest xf86-video-ati DDX driver Git code-base and it also requires a change (for querying the hardware CRTC ID) to the Radeon DRM code found within the Linux kernel, which can be found in a separate patch that was recently published by Jerome.
evil_core: where is this second patch?
dgbaley27: thanks for all the help, got family here now
evil_core: could anybody check what happens with warsow in KMS?
evil_core: it loads data so long, that server time outs
evil_core: it deosnt happen in UMS
twnqx: airlied: nope
twnqx: gonna try a pure 2.6.34rc5
glisse: evil_core: i posted second patch on dri-devel
glisse: and yes it works for r5xx
glisse: works for all radeon
evil_core: glisse: yeah I see
evil_core: but what does second patch?
glisse: add stuff in the kernel
gtklocker: is there any hope for gpu accelerated video?
glisse: can't work without kernel change
evil_core: glisse: so Its not really work w/o that patch for me?
evil_core: ouch., it was answetr for me probably ;)
glisse: there is a bug in my ddx patch that enable part of the functionality on non enabled kernel
glisse: i need to fix that
evil_core: http://lists.freedesktop.org/archives/mesa-dev/2010-May/000528.html
evil_core: I got d-r-t
evil_core: cannto find that patch
evil_core: no glisse there
evil_core: only barnes
evil_core: glisse: is it that patch, and I need it for d-r-t?
evil_core: ouch, I looked into mesa-devel ;)
evil_core: [PATCH] drm/radeon: async event synchronization for drmWaitVblank
evil_core: its all I need?
evil_core: glisse_ with that kernel patch, q3a becamed very laggy
evil_core: glisse: vsync doesnt work, I got 1200fps in glxgears
glisse: there is 2 patch on dri-devel
glisse: zzzZZZZzzz
evil_core: so there are two patches on dri-devel? (I got one from dri-devel)
evil_core: http://lists.freedesktop.org/archives/dri-devel/2010-May/000408.html
evil_core: that one, and it caused lags
evil_core: http://lists.freedesktop.org/archives/dri-devel/2010-May/000478.html
evil_core: or maybe that one I need also :/
twnqx: airlied: i take back "not working"
twnqx: and change it to "not completely working"
twnqx: e.g. my tft is regonized with the dp->hdmi adapter, then hdmi->dvi cable
twnqx: my porjector, however, is not recognized
twnqx: when connected via adapter and hdmi cable
evil_core: glisse: its broken, when glxgears started in gallium mode, then it was working until I moved window
evil_core: now every glxgears started is black
evil_core: and scrolling in urxvt is laggy
NTU: evil_core, i think hes sleeping
evil_core: but no errors in dmesg, xorg log or anything
evil_core: NTU: so you are netxt choosen one :D
NTU: i cant really help much. i've only been looking very closely at mesa code for a few weeks
agd5f: airlied: nmi's are from changing the pcie lanes. not sure if there is a way to avoid it
NTU: where does R600 get its GLX extensions specifically the ones that turn on GLX 1.4?
MostAwesomeDude: The GLX extension list is in common setup code shared between all the DRI drivers.
NTU: does GLX 1.4 require page flipping?
MostAwesomeDude: Not as far as I know.
MostAwesomeDude: In glxinfo, you should see the client (Mesa) and server (X) versions of GLX. The "supported" version of GLX that will actually be used during the session, is the intersection of those sets.
evil_core: anybody know what causes blackscreen with pageflip/gl app ghangs that needs xswerver restart?
FatBob: Hello
NTU: hey there!
NTU: how can we help you?
FatBob: I currently have an r800, and I want to know
FatBob: How can I help you!?
NTU: hehe
FatBob: If I can at all
FatBob: ...?
NTU: what do you mean how you can help me / us? like answer questions about the performance of r800?
FatBob: I suppose, if that's what's needed to help the project progress.
NTU: oh you want to help develop?
FatBob: Probably not, but then again, I don't know what that would entail.
FatBob: I could probably do some sort of testing.
NTU: ah
FatBob: Unless you want a mediocre coder with no driver programming experience.
NTU: im one of those people
NTU: hehe
FatBob: Lol
NTU: well if you want to help out that doesnt involve programming, you can use KMS on your evergreen hardware and see if you encouter bugs
FatBob: How do I go about doing that? Do I need to recompile
NTU: most like;y
NTU: *likely
NTU: it depends how late your distro is
NTU: actually i dont think any distros support r800 kernel modesetting :(
NTU: maybe fedora rawhide..
FatBob: Well, this is a fresh machine, I was running Arch, then Ubuntu, attempted a Gentoo install, etc
FatBob: I really don't mind change.
FatBob: I can always use the experience.
NTU: agd5f: does r800 have ucode that you need?
NTU: or anybody.. for that matter
NTU: wait yes it does
NTU: nevermind that. FatBob: let me edit my KMS guide on my wikispace page and then you can experiment :)
FatBob: Cool
NTU: updated! :) http://neo-technical.wikispaces.com/radeonhd-3d
FatBob: Awesome.
NTU: just be sure you do not start X after compiling new kernel. sometimes.. it doesnt work and you need to use live cd to recover or edit master boot record. then after you compile xf86-video-ati, mesa, and libdrm, it should all work
NTU: well not master boot record... menu.lst or grub.cfg to be more precise.
FatBob: I was about to ask...
FatBob: Have you used grub2 at all?
FatBob: nevermind
NTU: sorry i ws getting some yougurt and bread
NTU: ]i used grub 2 before
FatBob: it's all fine, I'm just happy I can contribute
FatBob: I half expected I wouldn't be able to help without getting further in LFS
NTU: linux from scratch? i love that
NTU: its a bit more of a pain than slackware but.. it pays off in the end
FatBob: Well, I'm not interested in actual deployment, just using it as a learning experience.
NTU: gentoo and lfs are the best for learning :)
FatBob: I'll take that advice.
FatBob: I've been considering gentoo just for the cflags.
FatBob: what's the point of having a processor with SSE4.2 if you can't use it?
NTU: oh btw for r800 kms you'll need to pull from drm-radeon-testing or just use airlied's tree. git clonel git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6.git && cd drm-2.6 && git checkout -b radeon-testing origin/drm-radeon-testing
NTU: *clone
NTU: it says to use linus's tree but thats for r600 / r700. ill edit the guide again
FatBob: Thanks, I was a bit worried about the lack of a r800 mention.
NTU: and updated again :)
Agrajag-: g'day, something is causing my 2nd monitor brightness to change depending on what's displayed on the screen. it seems that if there's lots of graphics on the screen it goes very bright, if there's not it goes darker. i have no idea what might be causing this, it's very annoying because it often goes very bright. any ideas?