This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: GSL fft benchmarks
- From: Brian Gough <bjg at network-theory dot co dot uk>
- To: "Steven G. Johnson" <stevenj at alum dot mit dot edu>
- Cc: gsl-discuss at sources dot redhat dot com
- Date: Mon, 15 Sep 2003 14:02:19 +0100
- Subject: Re: GSL fft benchmarks
- References: <Pine.LNX.4.21.0309120135290.22472-100000@ab-initio.mit.edu>
Steven G. Johnson writes:
> However, you may want to observe that your radix-2 routines are
> also less accurate. The reason for this, as described in our
> commentary is almost certainly the fact that you generate your
> trig. factors with a recurrence that is less accurate than the FFT
> itself. Realizing that you want to keep it from needing extra
> storage for the trig. factors, there are a couple of possible
> solutions:
>
Thanks for the references -- I'll see if I can take them into account.
--
Brian Gough
Network Theory Ltd
15 Royal Park
Bristol BS8 3AL
United Kingdom
Tel: +44 (0)117 3179309
Fax: +44 (0)117 9048108
Web: http://www.network-theory.co.uk/