Wildebeast, here's how it works, roughly.
In any given batch of chips that the fabs turn out, there is a "process center" is the average speed being produced. However, some will run faster than others, while some will run slower, *at the specified conditions*. that means within the thermal tolerances, normal voltage, etc.
The ones that run the fastest while still in spec are sold at 1.4ghz for example. The average ones are 1.2ghz, and the slow ones are 1ghz.
Overclocking has 2 aces up its sleeve though.
1) Often times the CPU maker marks a fast chip as a slow one, if its running out of the slower versions. This is why Duron 600s can go to 1ghz -- if AMD marked their true rated speed, they'd more likely be Duron 900s.
2) If we can CHANGE the conditions the CPU works under, we can increase the speed. That means make sure it stays cooler than the normal limit, and also increase the voltage. This is how most TBirds are getting to 1.4ghz+.
with all that in mind, you can see that a 1ghz Athlon might actually be a faster chip that's been marked down.. so when you up the voltage and slap on a good cooler, it overclocks a large amount.
if you buy a 1.3ghz Athlon, you know you're already getting the upper end of the process variation, so all you do is up the voltage and cool the chip better. This gets you some gain, but not as much as before.
Generally a rule of thumb is that you often can reach the higher speed CPU by OCing a lower speed one, but that same low speed CPU would not be able to overclock as far (in terms of mhz total) as one that was already marked as one of the fastest, would.
err, didn't mean to write an essay about it.
