Are you aware of Dan J. Bernstein's library?: http://cr.yp.to/djbfft.html
According to his benchmarks (from 1999?) it is faster than FFTW. I don't know whether that's still the case, a quick search did turn up http://projects.scipy.org/pipermail/scipy-dev/2002-August/001107.html which suggests that this was still the case in 2002. Support for djbfft seems to have been integrated into scipy. _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe