This is why the G4 (and the PPC family in general) is faster then the Pentium (I,II,II,IVs and most CISC chips) and megahertz comparisons between diffrent chip types are inaccurate.

I used to think that everyone that knew anything about computers didn't compare megahertz between diffrent chip families. It turns out that Intel did a better job of lieing to the public then I thought. I soon got sick of explaining this to people so I made this page, an easy course in why megahetrz does not equal power.

Before I go on, this is also a good resource.

And also this, this might help.

Even though almost all chip manufactures advertise chips in megahertz, megahertz is not the way you measure processing power. It is really measured in FLOPS or floating operations per second.

To get FLOPS we try this equation-

1.25 x 1 billion cycles/s) * (8 floating point ops/cycle) * (2 processors) = 20GFLops

This is the math for a 1.25 gigahertz G4, the FLOPS show the physical power of the chip. But thats not all. Your computer might run slower or faster depending on how well the code vectorizes.

The stuff behind all this is basically, a chip divides up a process into parts, the number of parts depend on your chip. These parts are called, the pipeline. Scroll back up to the animation on the top of the screen, what can you see? The top bar is a Pentium IV, the Pentium family uses a 20 stage pipleline. Of course it also needs a higher megahertz because it has to go though more steps. Even though the data (the white box) goes down the Pentium faster, the second bar, finished faster, even with a slower megahertz. The bottom bar is a G4, that uses a 6, 7, or 8 stage pipeline, so it doesn't need as much megahertz to process data as fast as the Pentium. Thats not all though, lets say both computers made a mistake. The pipeline must be "cleared", so the chip wipes the pipeline. Of course a longer pipleline takes much longer to "clear", a pentium can take over twice as long to clear as the G4.

RISC and CISC, is basically reduced instuction set computing and complete instuction set computing. The Mac has always used RISC, and for a reason. RISC is more efficent and faster then CISC. Tradional x86 (mainly Intel) processors have used CISC but now Intel has finally learned that CISC is old technologly, and Intel has started to make RISC processors.

So basically thats it, any questions?