Re: [oc] FFT core

[John William Dalton <jdalton@asiaonline.net.au>]

> Great. Are you going to implement radix-2 or
> any other scheme?

I'm planning to use the DeSpain algorithm.

I'm not sure that it falls into a straightforward
'radix' classification.  Essentially, an 'n' point
FFT is recursively decomposed into a pair
of smaller FFTs.  Eventually, you end up with
a bunch of transfoms whose radices are the
prime factors of 'n'.  With the constaint that
'n' is a power of two, you get a bunch of
radix-2 operations.  The main advantage of the
DeSpain algorithm is that many multiplications
are reduced to trivial rotations by pi/2. 

--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml