Stuart Allman wrote:

> I need [a square root] for figuring out the magnitude of a
> frequency after an FFT.

In this case you could be better off without taking the root.  There
are some simple algorithms for finding the magnitude that you might
like to investigate.  The only reference I have to hand is:

A.E. Filip, "Linear Approximation to sqrt(a**2 + b**2) having
Equiripple Error Characteristics"; IEEE Trans AU-21, Dec. 1973,
p. 554.

Good luck.

David
--
david.tait@man.ac.uk