OK, I'm quite familiar with the fact that MHZ is a bad speed rating for a CPU in the sense that it does not always correlate with performance (i.e., 50% more MHZ does not always equal 50% better speed). Any computer scientist/engineer worth his salt has heard a thousand times that if you really want to know how a system is going to perform, test it out on *your* target workload.

This is like suing the Big Three over the fact that horsepower is a terrible measure of a car's performance/power/speed because other factors (tires, the four fatass guys in the back, whatever) can make the "50% more horsepower" not add up to "50% more power".