Cavanaugh, Craig wrote:
> Supposedly APM is disabled when running SMP. The kernel option to issue
> idle instructions lives under APM.
No.
The kernel could tell APM: "we're not really doing anything". APM
could then decide to reduce the CPU clock rate for instance, but in
any case the CPUs are still halted (using little power) anyway.
> Does this mean the idle instructions are not being issued under a
> SMP kernel, or are they being issued regardless of the state of APM?
No, The APM calls are not made, but the CPUs are halted. That already
helps 90% of the way.
If the Transmeta CPU would not be translating the x86 code in
software, they would have to handle the "longrun" stuff (reduce clock,
reduce core voltage) in the APM driver.
The "reduce clock when mostly idle" trick doesn't really work.
If the CPU has something to do for 1ms every second, at half the clock
it will take 2ms, but half the power. So assuming 0.1W while halted,
and 40W while running full blast you'll use 0.999 * 0.1W + 0.001 * 40W
= .1399W . At half the clock it'd use: 0.998 * .1W + .002 * 20W =
0.1398W.
The transmeta trick of reducing the core voltage by a bit while
running at slower clock DOES work.
Roger.
--
** [EMAIL PROTECTED] ** http://www.BitWizard.nl/ ** +31-15-2137555 **
*-- BitWizard writes Linux device drivers for any device you may have! --*
* Common sense is the collection of *
****** prejudices acquired by age eighteen. -- Albert Einstein ********
--
=- To unsubscribe, email [EMAIL PROTECTED] with the -=
=- body of "unsubscribe linux-abit". -=
