Are there any other devices sharing the interrupt of your graphics card?
Which driver version are you using?

My system:
HP nc6000
Intel Pentium M 1.6 GHz
Intel ICH4 chipset
AMD/ATI Radeon 9600 Mobility

Graphics card driver version is 8.36.5, I guess that's the most recent
one available.
To save some power I do a "aticonfig --set-powerstate=1" AFTER X
startup, in my case in ~/.fluxbox/apps with a [startup] . If you do this
before X starts it will be overridden on X startup.
Switching to a virtual console (Strg+Atl+F1) and back into X11 after
about a second (Strg+Alt+F7) makes the fglrx driver stop producing
interrupts.

Another important thing: Do you use any 3D desktop functions like Xgl,
AIGLX, Beryl, Compiz, ... ???
In 3D the VBLANK interrupt is needed and fglrx will produce as many
interrupts per second as your screen refresh rate is set to!

Peter



On Mon, 2007-05-21 at 15:51 +0200, Flyser wrote:
> > Another issue I found while trying to get the CPU use C3: AMD/ATIs fglrx
> > driver has a strange bug: While X is running the driver causes as many
> > wakeups per second as the current screen refresh rate and the CPU will
> > never enter C3. When switching to a virtual console and back into X, the
> > driver won't cause that many interrupts anymore and the CPU suddenly
> > uses C3 state...
> > So everytime I start my computer I switch to a virtual console and back
> > into X to work around that bug.
> > Should I report this issue to AMD? Where should I report it?
> 
> While using fglrx I still get ~60 wakeups per second, switching to a VT
> does not help. Can you provide some Information about your system? e.g.
> Graphic-Board, Distribution, Framebufferconsole or not?, Splash, ...
> 
> Anyway thanks for this great tool, it helped me to nearly double the
> battery-lifetime on my Notebook!
> _______________________________________________
> Power mailing list
> [email protected]
> http://www.bughost.org/mailman/listinfo/power

_______________________________________________
Power mailing list
[email protected]
http://www.bughost.org/mailman/listinfo/power

Reply via email to