When speaking of CPU performance, one can usually calculate processing power using a simple formula:

IPC (instruction per cycle) x MHz (clock frequency)

Until recently - at least in relation to CPU generations ranging from the 286 to the 486 - the IPC of the x86 family has remained relatively constant, which made clock frequency the determining factor in CPU performance.

That ended quite abruptly with the coming

of the 5th generation of processors. At that point, the internal structures of Intel and AMD processors has diverged and changed considerably, which made the relative IPCs of their products a much more important factor.

In essence, CPU performance can be enhanced in one of three ways: by enhancing the number of instructions that can completed per cycle (i.e., increasing the IPC), by augmenting the clock frequency, or by improving both factors.

Historically, changes in the performance of succeeding generations of CPUs could be represented by the graph below, wherein clock frequency and IPC changed in tandem.

Recently however, with the introduction of the Intel Pentium 4, this has not proven to be the case, as is demonstrated with the following graph.

The architecture of the Intel Pentium 4 has changed things considerably. With it, Intel implemented their Netburst technology, which was designed in such a way that it delivers IPC performance 10% to 20% inferior to that of the P6 generation.

New generations of the x86 family have traditionally included augmentations in both IPC and clock frequency - a trend that the P4's designers have broken with quite dramatically. As a consequence, those who use clock frequency as the primary indicator of CPU performance are prone to overestimate the performance of the Pentium 4, which, despite its incredible clock speeds, suffers from an IPC that is lower than that of the Pentium 3. As a result of that fact, the P4's clock frequency must be much higher than the P3's in order to obtain better overall performance.

Relative IPC performance is also a big reason why the AMD Athlon 1.2GHz and 1.4GHz are often able to outperform Pentium 4 systems running at 1.8GHz and 2GHz - the Athlon's IPC value is simply much higher than that of its chief competitor. This all results in an era in which consumers can no longer rely on a simple comparison of CPU frequencies to evaluate relative system performance.

A concrete example of this new phenomenon can be had by comparing the performance of desktop systems to that of high grade server platforms. The Compaq Alpha, for example, uses a 1GHz processor, while the IBM RS/6000's runs at 450MHz, the HP PA8600 "Superdome" which runs at "only" 550MHz, and Intel's own top-of-the-line Itanium CPU, which tops-out at 733MHz to 800MHz. Despite the fact that all of these processors run at lower clock frequencies than comparable high-end desktop systems, they benefit from very high IPC's, and are able to demonstrate incredible performance figures.

So, it comes directly in response to this new state of affairs that AMD has taken to debunking the infamous "megahertz myth". Their reasonings should be pretty clear at this point, and their goal is nothing less than to establish a new standard of performance measurement capable of attracting customers who may else wise be drawn by the misleading promise of higher Hz.