Question #1: How does a CPU work
If you get realy basic, you end up with a half adder. ( way way way oversimplified, so don't shoot me )
Say you have a "register" ( that is where the CPU stores the binary #'s that it is going to do the "operation" on. ) assume you want to add 2 numbers together. Lets say 10 and 8. Lets further assume this is a simple 8 bit address for simplicity.
First the CPU will get the decimal #'s in binary form. 10 = 0 1 0 1 0 0 0 0 and 8 = 0 0 0 1 0 0 0 0
Then the CPU will "add" the binary 1's and 0's together. ( forgive below, text is always hard to draw in, and I am only going to show the first 5 bits, as the rest do not matter, as they are all 0's )
0 1 0 1 0
0 0 0 1 0 +
-------------
If the CPU shows 2 0's being add ed together, that equals a 0 in the sum ( totals ) line. If it sees a 0 and a 1, then that = 1. If it sees a 1 and a 1, then it will make that a 0, and add a 1 to the next line.
So the answer we get for the above is:
0 1 0 1 0
0 0 0 1 0+
-------------
0 1 0 0 1 = 18 ( 2 + 16 )
Does that kind of help? That operation would be called a "cycle" or such.
Since you seem to be asking this as a non-computer person, I will show you how the "bits" are counted so the above makes more sence. AKA how to turn a # into a binary #.
The 0's and 1's are in a horizontal row above. That means that the 0 ( or 1 ) farthest to the left, counts as the numeral 1. the next is a 2. after that a 4, then an 8, 16 yada yada. see below
0 0 0 0 0 0 0 0 0 <---- 8 bits. every 0 that turns into a 1 gets counted as it's decimal equivalent.
1 2 4 8 16 32 64 128 <------ Sorry, but the forum automaticly deletes extra spaces.
So if the first and last 0's in the above 8 bits are 1's then we would get 1+128=129. If they are all ones, you would get 255, but including "0" we get the magicial # of 256.
( Yes I know this is a HUGE simplication of what a CPU does, but I was trying to get it to the most basic so we can help Smags87.)
Quetion 2: AMD vs. Intel Design Differences
Frankly, I do not know a whole lot about CPU archetecture, but I do know a little. As was said above, AMD tends to do more work per cycle ( MHZ comparison ) than an Intel CPU does. Intel has what is called a "longer pipeline". in Intel's Hyper Pipelined Technology, there are 20 steps ( AFAIK ). The longer the pipeline is, the higher (theoreticly) the clock speed can go. ( without a die shrink ). As long as the processor anticipates what information it needs at the next step,( Branch Tree Prediction) then the longer pipeline will have no effect on the "work" that a CPU does in a given time frame. The problem lies in when the data is not ready for the next step. Then the processor has to wait, or start that cycle over. and 20 steps is a big jump compaired to the P3's 10 steps, that would all have to be done over, so the CPU tends to do less work per clock. ( PM Would be able to tell you much more than I ever could. Hope he stops by. )
<edit>
Names have been changed to protect the innocent.

</edit>