This optimization has been obsolete since the introduction of 3DNow! (PFRSQRT) and SSE (RSQRTSS, RSQRTPS).

These SIMD instructions compute a comparable approximation, but are much faster.

oO

So how long would it take me to understand this? (i know nothing of programming)

A reincarnation?

you tell me

It's mostly logarithms, not math. The standardized floating point algorithm is based heavily upon that concept; I think you'll understand it fine.

It's also not very advanced from a programming perspective. The other guy is right though; this optimization is a sight old, and nothing all that special, despite its cleverness.