Re: [ACPI-sppt] Who controls cpu fan (Asus M2N) ?

On Fri, Jul 04, 2003 at 02:09:04AM +0200, Rainer M. wrote:
> > > I'm using acpi with my Asus M2400N and kernel 2.40 and acpi patch
> > > 20021122 successfully and a customized dsdt table.
> > >
> > > In my Linux system, I'm not running some kind of daemon which controls
> > > the cpu fan with respect to the current cpu temperature.
> > >
> > > There's a bios option, which is called "Intel Speed Stepping", where I
> > > can choos from the following options:
> > > - Maximum Performance
> > > - Battery optimized
> > > - automatic
> > > - disabled
> > >
> > > The effect of this is, that the bios determines the cpu frequency and cpu
> > > fan speed *at boot time* for the entire session:
> > >
> > > If "battery optimized" is chosen, cpu freq T5 (50%) and fan is off.
> > > If "max. performance" is chosen, cpu freq is 100% and fan toggles between
> > > full power and off according to current heat situation.
> > > (when "automatic" is chosen, it just depends on whether or not the ac
> > > adapter is plugged in at boot time).
> > >
> > > Obviously, I'm rather interested in the "battery optimized" variant, for
> > > I can not stand the noise of the cpu fan running at full power. Once
> > > Linux is booted, I use some speed stepping program, which adjusts cpu
> > > clock according to the current cpu load, so that I'm not confined to the
> > > 50% cpu clock the bios preselected at boot time.
> > >
> > > But:
> > > For some strange reason, the cpu fan would _never_ switch on during such
> > > a session, not even when I'm compiling something!
> > >
> > > When starting in the maximum performance mode, the fan gets switched on
> > > an off very often, and during compilation, it runs permanently.
> > >
> > > So, I'm seriously concerned as to whether battery mode is really healthy
> > > for my processor.
> 
> I now found out the following: 
> 1. I started the computer in "battery optimized" mode where the cpu fan won't 
> ever start to run. 
> 2. I did echo 0 > /proc/acpi/processor/CPU1/throttling to speedstep the cpu to 
> 100%.
> 3. I started to compile something for half an hour
> 
> And what happened? Instead of my processor catching fire or the cpu fan simply 
> starting to run the cpu slows down in frequency to 300MHz in the end. But 
> /proc/acpi/processor/CPU1/throttling still showed T0-mode (I found out using 
> a bogomips-tool that the frequency gets scaled down). 
> If then I stop compiling and wait for some minutes, the cpu's frequency 
> increases again. 
> 
> If my interpretation is correct, then this behaviour can't be due to acps 
> because if acps scales down the frequency it would be visible at 
> /proc/acpi/processor/CPU1/throttling, wouldn't it?
> But can the dsdt table be responsible for this or is it a cpu-built-in 
> feature?

If you have a P4, then yes.  The CPU is able to throttle itself
if it detect overheat condition.

Look at the IA-32 Intel Architecture Software Developer's Manual Volume 3:
System Programming Guide, Order Number 254572.
Look then at 13.15.2.1 (and appendix B) for how to enable/disable this
feature.


-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100006ave/direct;at.asp_061203_01/01