There are many variables that will affect the speed that a chip will run. This includes: max speed of the silicon, architecture of core, vcore, and temps.
The max speed of the silicon is determined by how the silicon is made. AMD uses SOI (silicon-on-insulator) that adds an additional layer of silicon that helps to reduce gate-leakage and will thus reduce energy consumption of the chip, but the downside to this is that SOI transistors don't switch as fast as non-SOI transistors, which Intel uses and can hit much higher frequencies.
The architecture of the chip can be designed around speed. This was clearly evident in the Netburst architecture where the chip was designed with very very long pipelines for purely the sake of high frequencies without any regard for energy consumption. The current K8 architecture is scaling to the lower 3GHz range whereas the current Core architecture is scaling to the mid to upper 3GHz range.
Vcore and temps have a similar relationship. This concept is pretty logical since the higher the vcore you put in, the higher the temps will be. Of all of the chips from the same wafer, those chips will have slight variances in terms of how much vcore it needs to run at a given speed stably. Some chips need less vcore while others will need a bit more. This is how chip manufacturers bin their chips. They don't have a production run of one speed grade and then a separate run of another speed grade. Because the chips are all nearly identical to each other, with the execption of vcore needed to attain a predetermined clockspeed and small imperfections from the etching, they all should
hit the same max speed. The problem that occurs that the higher frequencies for any chip is heat removal. What compounds this problem is that heat is generated expontially as more vcore is added. See figure 1.1. It will be easier to get to the highest possible speed of a chip with a chip that starts off with a higher speed and a lower vcore. Getting a higher clocked chip will shift you to the next higher curve. The higher on the curve you are to start with, the better your chances of hitting the highest possible speed because you will need less vcore, which equates to less heat generated. Now, it doesn't mean that it is impossible to attain the highest possible speed with a lower clocked chip, but it just means that it will take a lot more vcore and you will have to figure out a way to super cool the chip.
I hope this made some sense.....