🚀 go-pugleaf

Newsgroups: sci.image.processing
Subject: Re: Favorite FFTs (was 'Re: 2d dft phase info')
Summary: 
Expires: 
References: <1992Sep7.110102.25536@bnr.co.uk> <1992Sep8.232448.14932@mpifr-bonn.mpg.de> <1992Sep9.083621.7140@mcclb0.med.nyu.edu>
Sender: 
Followup-To: 
Distribution: 
Organization: Los Alamos National Laboratory
Keywords: 

I wish that I had known of the existence of your favorite FFT's a few weeks
ago when I was looking for public-domain FFT codes.  It would be VERY useful
if anyone with favorite FFT's could post where to obtain them by ftp.  
Here are the ones that I know about:

ncar.ucar.edu: /dsl/lib/fftpack 
     1-d real/complex and complex/complex FFT's, mixed-radix, 
     sequential-storage only.
ncar.ucar.edu: /dsl/lib/ecmfft
     complex/complex  multiple simultaneous fft, radix 2, 3, & 5.  Array storage
     may be used.  This code has the disadvantage of needing a work array
     as large as the data array.  Aside from that fairly fast.  I have run
     it on Crays, Convexes, VAXes, SGI's, HPs700's, Suns, and IBM RISC/6000s.
ncar.ucar.edu: /chammp/shallow/lib/fft99f.f:
     a real/complex version of the ecmfft code described above.
pprg.unm.edu: /pub/khoros/contrib/cfft.shar.Z
     a 2-d real/complex or complex/complexradix-2 fft code, designed to be used
     as a filter from UNIX or integrated into KHOROS.

What I'm looking for is a 2-D real/half-complex mixed-radix code that doesn't
require additional storage, analogous to the VECLIB SRC2FT routine found on
Convexes.  I'm doing FFT filtering of 2304x2304 images, and the requirement
of additional storage makes the memory requirements too large to be practical.

--Joel Berendzen