On 8/23/09, ?iga Lenar?i? <ziga.lenarcic at gmail.com> wrote:
> myfft uses bordeaux-fft. I wrote a thin layer to convert maxima list
> to a complex lisp array and back. The difference grows for bigger
> sample sizer (for which the reason must be inefficient list
> conversion of O(N^2) in current fft).
> Any interest?
Well, I suspect you're right about the list/array conversion
being the bottleneck. If so, then just replacing that with
faster code is going to get us most of the speed improvement
with a tiny fraction of the effort. I can merge in a patch
for the list/array conversion if you write it.
Robert Dodier