🚀 go-pugleaf

warrenp@titan.colorado.edu (Pete Warren) wrote:
>Greetings all,
>
>I am trying to perform large scale FFT's on complex data in a column (or
>row) wise formats.  Does anyone know of a good, proven routine for either
>Fortran (preferred) or C ?  There are routines for complex 1-D data and
>for real column wise data but not for both.
>
>What I'm actually trying to do is build a code interface node for Labview
>that performs this function efficiently.
>
>I would appreciate any help or pointers.
>
>Thank you,
>
>- Pete Warren
>warrenp@colorado.edu


I have to admit, that I haven't a clue as to what you mean by column wise, but let 
me point you to a source that may be valuable anyway.

Way back in '79, the IEEE Press published a set of Fortran routines for signal 
processing.  I think that they are still available.  ISBN 0-87942-127-4, distributed 
by John Wiley and Sons.  

Anyway, there were both one dimensional and two dimensional real and complex 
routines designed to work to data sets so large that they were on the disk instead 
of main memory.

Inexpensive.  



Opinions expressed herein are my own and may not represent those of 3M.

In article <4jev78$bp7@dawn.mmm.com>, rbmccammon@mmm.com says...
>
>
>warrenp@titan.colorado.edu (Pete Warren) wrote:
>>Greetings all,
>>
>>I am trying to perform large scale FFT's on complex data in a column (or

>Way back in '79, the IEEE Press published a set of Fortran routines for 
signal 
>processing.  I think that they are still available.  ISBN 0-87942-127-4, 
distributed 

I vaugely remember a (mentioned above?) IEEE Press book called Digital 
Signal Processing that i used for a FFT routine called FFT842.  The FFT 
performed radix 8 manipulation on the data until there was only a radix 4, 
radix 2, or no operation to be performed on the data.  It was a modified 
Cooley-Turkey (sp?) that sped up the operation.  It was a complex FFT but 
used two reals for the real and imaginary (fake :-)) part.

The source is in FORTRAN.  I have a copy of the source.

Gerry

email address thomas_picard@ccmail.orl.mmc.com

The opinions expressed here are mine alone.

Pete Warren wrote:
> 
> Greetings all,
> 
> I am trying to perform large scale FFT's on complex data in a column (or
> row) wise formats.  Does anyone know of a good, proven routine for either
> Fortran (preferred) or C ?  There are routines for complex 1-D data and
> for real column wise data but not for both.
> 
> What I'm actually trying to do is build a code interface node for Labview
> that performs this function efficiently.
> 
> I would appreciate any help or pointers.
> 
> Thank you,
> 
> - Pete Warren
> warrenp@colorado.edu
The following fft subroutine is from web resourse( I am not sure where)
, maybe it helps.
      SUBROUTINE FOUREA( N, ISI)
C
C THE COOLEY-TUKEY FAST FOURIER TRANSFORM IN ANSI FORTRAN
C
C DATA IS A ONE-DIMENSIONAL COMPLEX ARRAY WHOSE LENGTH, N, IS A
C POWER OF TWO.  ISI IS +1 FOR AN INVERSE TRANSFORM AND -1 FOR A
C FORWARD TRANSFORM.  TRANSFORM VALUES ARE RETURNED IN THE INPUT
C ARRAY, REPLACING THE INPUT.
C TRANSFORM(J)=SUM(DATA(I)*W**((I-1)*(J-1))), WHERE I AND J RUN
C FROM 1 TO N AND W = EXP (ISI*2*PI*SQRT(-1)/N).  PROGRAM ALSO
C COMPUTES INVERSE TRANSFORM, FOR WHICH THE DEFINING EXPRESSION
C IS INVTR(J)=(1/N)*SUM(DATA(I)*W**((I-1)*(J-1))).
C RUNNING TIME IS PROPORTIONAL TO N*LOG2(N), RATHER THAN TO THE
C CLASSICAL N**2.
C AFTER PROGRAM BY BRENNER, JUNE 1967. THIS IS A VERY SHORT VERSION
C OF THE FFT AND IS INTENDED MAINLY FOR DEMONSTRATION. PROGRAMS
C ARE AVAILABLE IN THIS COLLECTION WHICH RUN FASTER AND ARE NOT
C RESTRICTED TO POWERS OF 2 OR TO ONE-DIMENSIONAL ARRAYS.
C SEE -- IEEE TRANS AUDIO (JUNE 1967), SPECIAL ISSUE ON FFT.
C
-Xiangyang
.