On 2003. 02. 11., Russell McManus has wisely spoken:

>> > Yet another stupid design decision in PC hardware. Fans should auto-switch
>> > on when the temperature is too high. Implemented in hardware. I guess we can't
>> > circumvent this? (Other than keeping the fan always running)
>> 
> I also see similar behavior with my Shuttle SS51G mini-pc.  Windows
> will spin down the fan, but it runs at full tilt all the time in
> NetBSD.

Well, my HP XE3 seems to be able to do this by hardware - it runs a
NetBSD 1.5.3, definitely without ACPI, and if I compile the kernel or
use Silc, the fan spins up, and spins down is the cpu is idle again.

SCRLY:

Hajdú Zoltán