Geez man, jump down the poor guy's throat why don't ya??
It's the same principle as say DDR memory for example. . .
DDR is Double data rate. So DDR 400 memory actually is only really running at 200 MHz but the data rate is doubled. So imagine a 200 MHz wave form on an osciloscope. If you can fit one read/write operation in on the rising part of the wave, you're operating the memory at normal 200 MHz speed. But what if you could fit 1 more operation on the falling side of the wave? Now you've just doubled the number of operations you can perform in one 200MHz cycle! So effectively it's like you're running 400 MHz kinda. So hence the name DDR. The memory frequency is really only 200MHz but you are doing 2 operations each cycle instead of 1. The FSB works on the same principal. So when you say 133 is the FSB, remember that you have to take the multiplier into account to find out the effective speed. The actual wave form of the FSB is just like looking at a 133MHz wave form. But if you can manage to cram 2 operations on the rising side and 2 operations on the falling side, it looks like you are going 4x as fast so hence 533MHz. The limitation comes when you reach the physical limits of how many operations you can perform in 1 cycle. . .The lower your frequency, the higher your multiplier can be. The higher your frequency (faster) the less time you have to get in those operations so at some point you are going to hit a limit and have to lower your multiplier. Maybe you can get in 4 operations during a 133MHz wave but increase that wave frequency to say 300MHz and you have a much smaller wave form in which to cram all those operations. Get it?