Discussion: the ati/radeon open source drivers & radeon repository

gtklocker · 2010-03-07 14:51:34

Wow... +100 fps from the last update.

3570 frames in 5.0 seconds = 713.522 FPS

ZoYeR · 2010-03-07 18:38:30

gtklocker wrote:

Is powersaving enabled in kernel26-git ?

I'm wondering that too, could anyone please tell us? .

mjr · 2010-03-07 19:29:22

Hi folks!
At first, i'd like to thank Perry3d for the repo. I didn't have the free time to build all this bleeding edge code from aur, so i was stuck with catalyst drivers or radeon/radeonhd in the official repo.
Now i'm very excited of what can be done using latest radeon drivers and kernel.
On the other hand i'm currently running arch on my laptop, so i NEED power management working. Is it finally enabled? I installed today's packages and couldn't get it working.

@endlessroad1991: I have the same card (on a Dell Studio 1555), so... did you managed to dynamicpm working?

Last edited by mjr (2010-03-07 20:36:09)

ungraven · 2010-03-07 22:38:28

Hi,

I recently came across this on the mesa-3D list:

http://sourceforge.net/mailarchive/mess … 0gmail.com

It may interest anyone trying to get Heroes of Newerth (and possibly other things) working with r600/700.

endlessroad1991 · 2010-03-08 05:37:08

@mjr : I have the same laptop as yours
I've found 2 ways of power saving, both of them are effective to some content,
but neither of them can catch up with official ati driver for Windows.

Firstly, upgrade all packages in Perry3D's repo.

Without KMS (It won't work if KMS is enabled!!):
In your xorg.conf file, add 2 lines to "Device" Section:
Option "DynamicPM" "on"
Option "ClockGating" "on"
--------
If the two options are enabled successfully, you will see following lines in /var/log/Xorg.0.log:
(**) RADEON(0): Option "ClockGating" "on"
(**) RADEON(0): Option "DynamicPM" "on"
......
Static power management enable success
(II) RADEON(0): Dynamic Clock Gating Enabled
(II) RADEON(0): Dynamic Power Management Enabled
--------
If you desire low power cost, you can add an extra line to "Device" Section of xorg.conf:
Option "ForceLowPowerMode" "on"

With KMS (It won't work if KMS is disabled!!):
Add
radeon.dynpm=1
to grub commandline,
or add
options radeon dynpm=1
to /etc/modprobe.conf
--------
If it's enabled successfully, you will see following lines in output of "dmesg | grep drm":
[drm] radeon: dynamic power management enabled
[drm] radeon: power management initialized

On my laptop, both of the 2 ways decrease fan noise significantly, which is really a great news to me.
But I haven't tested which of them can save more power and then lead to a longer battery life.
Maybe you should give both of them a shot and decide which to use.

Last edited by endlessroad1991 (2010-03-08 07:56:23)

handy · 2010-03-08 07:47:56

Thanks for your repo Perry3D, it makes life so much easier.

I used the repo to install the -git kernel & other packages. After which I'm still the 400 fps down in glxgears that I lost yesterday after upgrading the xf86-vide-ati-git, so upgrading all of the other packages, which were a good month old didn't improve that.

The error messages are now gone as shown below:

handy ~ $ glxgears
6342 frames in 5.0 seconds = 1268.252 FPS
6330 frames in 5.0 seconds = 1265.934 FPS
6336 frames in 5.0 seconds = 1267.168 FPS
6331 frames in 5.0 seconds = 1266.144 FPS
6330 frames in 5.0 seconds = 1265.925 FPS
6337 frames in 5.0 seconds = 1267.371 FPS
XIO: fatal IO error 104 (Connection reset by peer) on X server ":0.0"
after 121722 requests (121719 known processed) with 0 events remaining.
handy ~ $ glxinfo |grep -i opengl
OpenGL vendor string: Advanced Micro Devices, Inc.
OpenGL renderer string: Mesa DRI R600 (RV630 9583) 20090101 TCL DRI2
OpenGL version string: 2.0 Mesa 7.9-devel
OpenGL shading language version string: 1.10
OpenGL extensions:

The phoronix test suite showed a minor drop in frame rates across the board:

http://global.phoronix-test-suite.com/i … 6676-10968

Interestingly, the changes made today have allowed me to be able to play Glest. It has remained unplayable on the OS drivers up until now. Even though I previously had better than 1600 fps in glxgears & tested a bit faster in the PST.

So that's good news. There are obviously a lot of improvements going on in the code, that don't easily show themselves, not even in the PST.

Last edited by handy (2010-03-08 08:02:27)

mjr · 2010-03-08 13:14:06

@endlessroad1991:
Thanks for your answer, i figured out both methods and tried yesterday for a while.
I found the dynpm in xorg.conf (without kms) better than the other. Anyway i'm using kms because it's very nice...
I'm waiting now pm full support, in 2.6.34 maybe...

By the way, and off-topic of course:
Could you make the laptop suspend without the messed up sensors issue? I can get it suspended and then wake it up, but the fan never spins and sensors reads for thermal zones 1-3 are 0°.

Last edited by mjr (2010-03-08 13:15:06)

ijanos · 2010-03-08 13:23:26

Hey Studio 1555 owners, how cool can you make your ati card? Mine stays around 65°C

Neuro · 2010-03-08 16:16:32

Ok, one question:
Kwin with compositing spawns some errors:

WARNING: Application calling GLX 1.3 function "glXCreatePixmap" when GLX 1.3 is not supported! This is an application bug!
WARNING: Application calling GLX 1.3 function "glXQueryDrawable" when GLX 1.3 is not supported! This is an application bug!
WARNING: Application calling GLX 1.3 function "glXDestroyPixmap" when GLX 1.3 is not supported! This is an application bug!

I'm using yesterdays GIT, radeon in UMS. my glxinfo says:

IRQ's not enabled, falling back to busy waits: 2 0
name of display: :0.0
display: :0 screen: 0
direct rendering: Yes
server glx vendor string: SGI
server glx version string: 1.2
client glx vendor string: Mesa Project and SGI
client glx version string: 1.4
GLX version: 1.2
OpenGL vendor string: Advanced Micro Devices, Inc.
OpenGL renderer string: Mesa DRI R600 (RV620 95C4) 20090101 TCL
OpenGL version string: 2.0 Mesa 7.9-devel
OpenGL shading language version string: 1.10
OpenGL extensions:

Why is the server glx version 1.2 and not 1.4? What am I missing?

Coragem · 2010-03-08 20:04:23

Guys first id like to thank you all for the time invested in this thread i read almost 20 pages of it and managed to get the OSD working on the second try.

Right now im using stock kernel from core, and driver and mesa GIT packages from that nightly repo on page 15 for Arch 64.

KMS is On late start, i want to try early start later. UMS does not work on my ATI 4830.

Iḿ using a crossfire setup so even tho OSD driver is way better than i last tested. if ATI fixes the driver for new Xorg ill switch back to catalyst. the performance hit is too high.

I have a few questions if you guys can help me.

grep -i "Direct rendering" /var/log/Xorg.0.log
(II) RADEON(0): Direct rendering enabled

but Glxinfo.

name of display: :0.0
display: :0 screen: 0
direct rendering: No (If you want to find out why, try setting LIBGL_DEBUG=verbose)

How can i set this libgl_debug=verbose and how i check its status? why one says direct rendering yes and the other no?

Neuro im getting the same 1.2 version on glxinfo as you saying guess its normal?

Last edited by Coragem (2010-03-08 20:05:09)

Perry3D · 2010-03-08 20:22:16

@endlessroad1991: I appended your howto to the first post. I hope that's ok?

@Coragem: If direct rendering is disabled your system is really slow. There should be more errors on /var/log/Xorg.0.log or dmesg.

@Neuro: I think server is the xorg-server and the client is mesa. But that is just a speculation.

Coragem · 2010-03-08 21:01:18

Perry,

Thats the point, i dont see any other errors. and says its using DRI2, X is running normal speed, Google earth Opengl works same as Chromium-BSU, Stellarium.

grep -i "Direct rendering" /var/log/Xorg.0.log shows:
(II) RADEON(0): Direct rendering enabled

So i belive its working but why glxinfo says its off?

Iḿ getting around 900-1000 fps in glxgears with a 4830 and Phenom II X4 kinda low for what you guys saying around 3000. (i know its not good benchmarch but...)

How can i use this LIBGL_DEBUG=verbose to check???

And Perry thank you for the other replies on the thread you helped me a lot.

Any ideas? Thank you.

Last edited by Coragem (2010-03-08 21:06:46)

mjr · 2010-03-08 21:37:24

@ijanos:
I could keep it 64° with catalyst, 73° with KMS-enabled radeon (dynpm=1) and about 71° with UMS radeon and ClockGating/DynamicPM on.
So you have a Studio 1555, i guess... then could you make it suspend and resume usable?

ijanos · 2010-03-08 21:48:22

mjr wrote:

@ijanos:
I could keep it 64° with catalyst, 73° with KMS-enabled radeon (dynpm=1) and about 71° with UMS radeon and ClockGating/DynamicPM on.
So you have a Studio 1555, i guess... then could you make it suspend and resume usable?

I have to confess that I am using windows because of this Though I am using my arch install from my previous machine in virtualbox (and running the x server on windows, its suprisingly fun).

mjr · 2010-03-08 21:58:18

ijanos wrote:

mjr wrote:
@ijanos:
I could keep it 64° with catalyst, 73° with KMS-enabled radeon (dynpm=1) and about 71° with UMS radeon and ClockGating/DynamicPM on.
So you have a Studio 1555, i guess... then could you make it suspend and resume usable?
I have to confess that I am using windows because of this Though I am using my arch install from my previous machine in virtualbox (and running the x server on windows, its suprisingly fun).

Ha! I'm not using windows but i'm always considering it... This Studio is my first laptop, and every step i take in making arch work better i get the (scary) feeling that laptops are meant to be running windows. I'm afraid that these little quirks somehow are gonna cutoff the computer's lifetime...
But every time i boot windows... yikes... if freezes, and annoys me with autoupdates and stuff. It really makes me sick, so i get back to archie

Last edited by mjr (2010-03-08 21:58:50)

ijanos · 2010-03-08 22:07:33

Dont feel that way. My prevoius acer travelmate was running better with linux. I have the same feeling about windows and my computers lifetime, like "oh god, its using the HDD again...". But I have to admit, this win7 is pretty stable for me, freezed only in 6 months, it boots in 30 seconds (okay i tweaked it a little ) and running arch in vbox smoothly, i'm using linux as my main developement platform, windows is just the host until ati supports get better.

ZoYeR · 2010-03-09 00:11:33

I've read on Phoronix that Linux 2.6.34-rc1 has been released. Maybe power management has improved there.
I'm trying this out tomorrow, if I'm right, kernel26-git pkgbuild should provide me with that version .

I don't really care about power consume, I'm running an ATI HD4850 on a desktop PC, but without pm the fan spins at full speed and that noise really gets on my nerves... it's impossible to sleep with the computer on .

mjr · 2010-03-09 03:49:11

Actually i've got the updated pkgs and my linux version is 2.6.33-05679-g4da0b66-dirty.
@Perry3D: is this pkg 2.6.34-rc1? i guess not, so i'm hopefully waiting for it...

Last edited by mjr (2010-03-09 03:49:27)

Cdh · 2010-03-09 04:54:23

@Perry3D
Do you have packages for i686 too?

venky80 · 2010-03-09 08:59:06

if there is one thing in which linux sucks big time ..it is power management. With win 7 my laptop runs so much cooler and suspends/hibernates seamlessly....

endlessroad1991 · 2010-03-09 10:12:39

@Pery3D:
I'm glad to make them help others, so thank you for appending them to the first post~~
@mjr:
I haven't tried yet... In my everyday life, I don't use suspend...

My card stays around 68C with xorg.conf way, and 70C with KMS way.
My laptop fell onto the ground yesterday and didn't boot since then, I've sent it to Dell...

ChemBro · 2010-03-09 13:09:45

I guess, power management works with 2.6.34-git. My pc is not so loud anymore (it is quiet). Have to test it more, but it seems to work.

Before 2.6.34-git, my pc was loud, even if didn't do anything on my desktop. Didn't test it with games, though.

Perry3D · 2010-03-09 13:49:59

@mjr: I think i updated the kernel 2 days ago. So the drm-patches should already be included. Yesterday Linus taged the kernel with rc-1. If i have some time i will update the package this evening.

@Cdh: No, i just offer the packages that i build for my system. Is there a simply way to build i686 packages on a x86_64 system?

@Coragem: I have nearly the same setup (Phenom II X4 and a HD4850). But direct rendering is working with DRI2. Did you try to disable KMS? With KMS i get almost the same result (~1000fps). I think there was a performance drop when vline support was introduced. But now we have tear free opengl/gl video .

Coragem · 2010-03-09 17:39:24

Perry, UMS was my first try, but i get this message:

(WW) RADEON(0): Direct rendering is not supported when VGA arb is
> >> necessary for the device
> >> (EE) RADEON(0): [dri] DRIScreenInit failed. Disabling DRI.

I read in some Debian buglist this was not going to be fixed for UMS and only would work in KMS.

So i turned KMS on and it worked.
Do you have any idea what could fix it so i could run using UMS?

Thank you.

Perry3D · 2010-03-09 18:44:28

Stay with KMS: there is no reason to use UMS. Ok, at the moment it is slower than UMS but that will hopefully change in the future.

/edit:kernel in the repo is updated:

perry64 ~ % uname -r
2.6.34-rc1-dirty

Last edited by Perry3D (2010-03-09 20:42:17)

Arch Linux

#926 2010-03-07 14:51:34

Re: Discussion: the ati/radeon open source drivers & radeon repository

#927 2010-03-07 18:38:30

Re: Discussion: the ati/radeon open source drivers & radeon repository

#928 2010-03-07 19:29:22

Re: Discussion: the ati/radeon open source drivers & radeon repository

#929 2010-03-07 22:38:28

Re: Discussion: the ati/radeon open source drivers & radeon repository

#930 2010-03-08 05:37:08

Re: Discussion: the ati/radeon open source drivers & radeon repository

#931 2010-03-08 07:47:56

Re: Discussion: the ati/radeon open source drivers & radeon repository

#932 2010-03-08 13:14:06

Re: Discussion: the ati/radeon open source drivers & radeon repository

#933 2010-03-08 13:23:26

Re: Discussion: the ati/radeon open source drivers & radeon repository

#934 2010-03-08 16:16:32

Re: Discussion: the ati/radeon open source drivers & radeon repository

#935 2010-03-08 20:04:23

Re: Discussion: the ati/radeon open source drivers & radeon repository

#936 2010-03-08 20:22:16

Re: Discussion: the ati/radeon open source drivers & radeon repository

#937 2010-03-08 21:01:18

Re: Discussion: the ati/radeon open source drivers & radeon repository

#938 2010-03-08 21:37:24

Re: Discussion: the ati/radeon open source drivers & radeon repository

#939 2010-03-08 21:48:22

Re: Discussion: the ati/radeon open source drivers & radeon repository

#940 2010-03-08 21:58:18

Re: Discussion: the ati/radeon open source drivers & radeon repository

#941 2010-03-08 22:07:33

Re: Discussion: the ati/radeon open source drivers & radeon repository

#942 2010-03-09 00:11:33

Re: Discussion: the ati/radeon open source drivers & radeon repository

#943 2010-03-09 03:49:11

Re: Discussion: the ati/radeon open source drivers & radeon repository

#944 2010-03-09 04:54:23

Re: Discussion: the ati/radeon open source drivers & radeon repository

#945 2010-03-09 08:59:06

Re: Discussion: the ati/radeon open source drivers & radeon repository

#946 2010-03-09 10:12:39

Re: Discussion: the ati/radeon open source drivers & radeon repository

#947 2010-03-09 13:09:45

Re: Discussion: the ati/radeon open source drivers & radeon repository

#948 2010-03-09 13:49:59

Re: Discussion: the ati/radeon open source drivers & radeon repository

#949 2010-03-09 17:39:24

Re: Discussion: the ati/radeon open source drivers & radeon repository

#950 2010-03-09 18:44:28

Re: Discussion: the ati/radeon open source drivers & radeon repository

Board footer