Having spent more than 10 years with embedded programming, I'm well familiar with sleeping processors, thank you, so don't blow a gasket

But where did I claim that the processor would idle at 125 or 250MHz?

What I did mention was the ability of getting a lower clock speed. It isn't as simple as that the processor has a power consumption linear to the clock frequency and that half the clock frequency means twice the the time, ending up with the same power consumption.

A very important fact here is that a higher clock frequency often requires a higher core voltage while a lower clock frequency may allow a reduction of core voltage.

So if you have:
12 0.7500 V 125 MHz
20 0.8500 V 250 MHz
then having the processor perform work (no, not sleep) at 125MHz can make the processor perform X amount of work at twice the time but consuming less power than if performing the same amount of work at 250MHz.

See for example the 7th post (Elanzer @ 04-25-2010 07:14 PM) of this link about some tested core voltage possibilities at different clock speeds:
http://forums.internettablettalk.com...&pp=40&page=84

There really are good reasons why a lot of mobile equipment are using dynamic clock frequencies instead of doing it old style and always stay at max clock frequency and only save power by sleeping between the bursts. The main thing is what clock frequency you can get at lowest possible core voltage - and if longer computation times at lower clock frequencies adds extra power consumption in external hardware that needs to be enabled for a longer time.