Are you really getting such great improvement? I thought that the current state of the patches was basically providing the infrastructure, but that the actual decrease in clock frequency was minimal?
Anyway, the patches don't apply here on a vanilla kernel. Are they meant for a git checkout or am I doing something wrong?
I do have a decrease of 4-5 degree (C) which on my laptop is a difference between a noisy fan and just a blowing fan. I'm surely hoping for more once everything is in place (I tried long time ago fglrx and I think it did better than this) but this is a good start for my HD3200.
GPU: Mobility Radeon 3470 (that's an RV620)
Originally Posted by pingufunkybeat
Notebook: Sony Vaio SR21M
With FGLRX I never had heat-related problems (32-33 degrees, don't remember exactly), but previously with the radeon open source DDX and DRM module it quickly went up as high as 46 degrees just a few minutes after boot, then 48 and 50 degrees. I wanted to test the development of the driver, so I always damned hell and all its demons in removing FGLRX, then really removing all of it, compiling new linux kernel, applying patches, then DDX, mesa... I was thinking I just was irreparably mad at testing the development of the driver on a notebook.
Then the gods, erh... the developers, wrote the code for the ForceLowPowerMode and DynamicPMOrSomething options for the DDX, and I was all hip and cool trying that in UMS. KMS was just heaven at the time.
Then these patches hath descended from the heavens, and now I can write on my notebook without burning it and my wrists to ashes. All while using KMS, and all that comes thereafter. Writing in a Firefox window, with a LAMP stack running in the background, I'm currently experiencing 37 celsius degrees.
Thank you all.
I've moved to Arch Linux's AUR package mesa-full 20100117 (with gallium enabled) on kernel 2.6.33rc4 + Ati PM patches and xf86-video-ati-git.
I must say the patches do a really good job. I get 52C in idle while Catalyst managed 50C. That's really really good
Unfortunately something's still wrong. I'm getting the following warning in glxgears, though they themselves work:
do_wait: drmWaitVBlank returned -1, IRQs don't seem to be working correctly.
I've installed radeon_ucode firmware stuff so that the IRQ patches will be working, but this warning doesn't go away.
Also, I don't know if it's a sideeffect of not having Vblank (that's what the IRQs are for, yup?) in OpenGL, but sometimes in Kwin (KDe 4.4 RC1) I'm getting yellow translucent bands instead of window drop-shadows. These didn't appear when using kernel 184.108.40.206 with radeon UMS. Anyone has a clue?
The KWin issues are probably a Mesa regression. Mesa broke KWin effects badly in the past, but it was fixed (the commit reverted). Now it's back, though not nearly as bad as it used to be.
I've just built kernel 2.6.33-rc4 vanilla + PM patches (from 0002 to 0010, since 0001 seems to be upstream yet), but I get 68 to 70 Celsius (I get the same without patches), while I get around 50 Celsius with catalyst. Am I missing something?
I have hd4570 on dell laptop.
Whoa! The 0010 patch is new to me, thank you for the info. I'll be trying it soon.
Originally Posted by baghera
Did you build drm and radeon as modules or built in?
If they're modules, append radeon.modeset=1 radeon.dynpm=1 to the kernel commandline, or modprobe the module with options modeset=1 dynpm=1
You can set modprobe rules in the /etc/modprobe.d/ directory, with a line like:
options radeon modeset=1 dynpm=1
Thanks for the 0010 tip.
Thanks for dynpm tip, I was loading module with dynclks=1 but it has no effect, now I'm running at 63 Celsius. Better than nothing, but I guess I have to wait for more aggressive PM.
It's just gpu that runs at 70 Celsius, both cpu are around 45 degrees.
Originally Posted by Loris
I'm sure everything is well-configured because if I switch to catalyst temperatures are ok, gpu is around 50 and also cpu are ok.
I'm not surprised of those temps since I also get 64-65 degrees using ForceLowPowerMode in xorg.conf without KMS.