I would like to point out a few things I have learned over the years. First, the only place in any part of the computer industry a MB is 1,000k is hard drive makers. No other part of the computer world sees it that way.
Some of you who have been around will remember this, back in days when computers were first being made to actually be for home use, whole programs fit on a 5 1/4 floppy that held 360k of data max. When very small hard drives, by todays standards, were costing much more then massive hard drives cost today, the difference of even a 1/4 MB was a major difference.
So at some point, one of the drive makes decided to use the 1,000k to inflate the space of there drives to give them selfs an edge on the market. It was a simple marketing ploy. And of course, all the others followed to even the playing field.
The 1,024k = MB standard was defined way back in the 50's when the very first punch card computers were built. Before there was really such a thing as a hard drive as we know it today. In fact I remember my dad having a true 10 MB hard drive. And the famous last words, "I will never fill that much space up". We never let him forget that.
It is the same concept as when AMD actually started to make there CPU's that could keep up with Intel's. At that time, Intel was losing too much of the market too fast. So Intel's quick answer was to lessen the amount of data processed within a clock cycle. And because the number of clock cycles per second is what the speed rating of a CPU is based off of, Intel showed much higher click speeds then AMD, which is why the use the + rating on there CPU's.
Which explained why a 2 gig'ish AMD 3000+ would benchmark around the same as an Intel 2.8-2.9 gig CPU. Regardless of the number of clock cycles per second, the actually amount of data processed per second was close to the same.
Enjoy~