paulr@syma.sussex.ac.uk (Paul T Russell) (01/23/91)
I'm having problems getting the FFTR2BF FFT routine from Dr Bub
to work. This is the version with automatic scaling after each
butterfly (I have had no problems with the simpler routines).
If anyone has used this routine successfully could they please
answer some questions:
- Does it really want a cos/-sin table rather than the -cos/-sin
table used by the other routines ?
- How long is the twiddle factor table ? ie. if the FFT is N points,
is the twiddle factor table N points or N/2 points ? My guess is
that it's N/2 points and represents two quadrants, but this
is not documented in either the source file or the help file.
- Does it really expect the 'passes' parameter to be log2('points') ?
Stepping through the code on the simulator it looks like 'passes'
and 'points' are out either by a factor of 2 ('points') or by
an offset of 1 ('passes').
- Is there a newer version of this routine ? The file I have says
'Latest Revision - June 13, 1988'