I like that article. A little confusing without actually having a system in front of me, but Im sure it will be easy once we have our hands on it.
I think they made it sound a lot more complicated than it is needed for process which is essentialy:
change bus speed to 125 or 166 Mhz - find what multiplier gives you sensible clock speed.
For example:
set multiplier to 1,25 you have effective bus speed of 125 then set multi to 30 = you have 3750 Mhz cpu, increase multi to x31 you have 3875, x32 - 4000, etc...
