Xref: utzoo sci.math:12137 sci.math.num-analysis:1051 sci.math.stat:1565 comp.lang.c:31344
Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!tut.cis.ohio-state.edu!unmvax!pprg.unm.edu!krukar
From: krukar@pprg.unm.edu (Richard Krukar [CHTM])
Newsgroups: sci.math,sci.math.num-analysis,sci.math.stat,comp.lang.c
Subject: Re: Fast Fourier Transforms sources in C.
Keywords: FFT, Computers, C
Message-ID: <28581@pprg.unm.edu>
Date: 24 Aug 90 16:01:48 GMT
References: <34902@uhnix1.uh.edu>
Reply-To: krukar@pprg.unm.edu (Richard Krukar [CHTM])
Followup-To: sci.math
Organization: U. of New Mexico, Albuquerque
Lines: 22

> Is any one aware of any public domain sources for Fast Fourier Transforms or
> Any kind of software related to it. I am interested in a C language 
> implementation.

	I have implemented split-radix (2/4) forward and reverse real and
imaginary fft routines that I am willing to release in the public domain.
There is considerable loop unrolling ( I wrote a program to do this ) which
makes the source pretty big ( the source is bigger than the object ).  So
there might be some problems with distribution.  These routines were written
specifically for image processing and can not handle more than 4096 complex
points or 8192 real points, but the hooks are there to fix this.

	In transforming a 512x512 real image, these routines showed a
significant speed improvement (40%-50%)over the IEEE standard routines.  I
feel this improvement is very compiler dependent.

	Anyway, given sufficient interest, I will hook in the larger than
4096 capability as well as some stuff to speed the sucker up a little more.
Any preferences on release methods as well as the level of loop unrolling
should be mailed to me.

	Richard Krukar (krukar@pprg.unm.edu)