Newsgroups: sci.image.processing,comp.dsp
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!udel!gatech!swrinde!pipex!uunet!psinntp!calspan!calspan.com!dennison
From: dennison@calspan.com (Ray Dennison)
Subject: Re: Arbitrary N FFT code
Message-ID: <dennison.790954083@calspan.com>
Sender: usenet@calspan.com
Nntp-Posting-Host: gomez
Organization: Calspan Advanced Technology Center
References: <3fopac$n63@news.iastate.edu> <NEWTNews.27118.790629056.Postmaster@stella_1.itg.ti.com>
Date: Tue, 24 Jan 1995 13:28:03 GMT
Lines: 19
Xref: glinda.oz.cs.cmu.edu sci.image.processing:12137 comp.dsp:16136

In <NEWTNews.27118.790629056.Postmaster@stella_1.itg.ti.com> mges@msg.itg.ti.com writes:


>In article <3fopac$n63@news.iastate.edu>, <dickw@iastate.edu> writes:

>> Does anybody have a reference to some source code for an
>> arbitrary length FFT?  I.e., N not necessarily a power of two
>> nor a product of primes.  Thanks in advance.
>> 
>> Dick Wallingford

>N is necessarily a product of promes, if it is an integer. :^)

Unless its a prime itself. Back to the original question, remember what an FFT
is - a FAST Fourier Transform. The FFT expolits the symmetry in a DFT - 
DISCRETE Fourier Transform - to reduce the number of multiplies and additions.
The symmentry exists BECAUSE of the prime factorization. If you wish to use
arbitrary lengths then you must use the DFT. Just about any basic reference 
on FFTs should discuss the DFT. The usual approach in textbooks is to discuss the DFT first and then show the FFT as a special case.    
