[Kst] branches/work/kst/portto4/kst/src/libkstmath

[Barth Netterfield]

FFTW is a bit more complicated to use - it typically needs a 'plan' to
optimize the fft, etc - though it will end up (possibly) being faster
if you do use the plan.  The one we are using was hand optimized for
the current chip (P4 IIRC) many years ago.  It can also use
non-power-of-2 ffts.  But the speed of them is highly unpredictable.
If N is prime, then the FFT will revert to a slow-ft and be N^2.  If
all of the prime factors of N are small (2,3,5,7; maybe 11, 13) then
it is almost as fast as a power of 2, but checking for this adds even
more complexity.

Switching to FFTW adds another build requirement which would be a huge
pain on, eg, ARM...

There shouldn't be a memory advantage or disadvantage, as far as I know.

When are you using a 7Mpt power spectrum?   Wouldn't you use
interleaved averaging if you had that many points - trading resolution
for lower noise?

I am not totally opposed to switching to FFTW, but I am not terribly
excited about it either.

cbn

On Fri, Oct 19, 2012 at 1:33 PM, Brisset, Nicolas
<Nicolas.Brisset at eurocopter.com> wrote:
>> add comment about slower fft when build with mingw
> Speaking of which, I'd be curious to know how much RAM fftw uses, and how fast it is.
> It seems to be the de-facto standard these days, and our implementation uses up an awful lot of RAM (when using 7 million point vectors, OK...). I think we had the discussion before, but I don't remember exactly what the outcome was.
>
> Nicolas
>
> EUROCOPTER DEUTSCHLAND GmbH
> Sitz der Gesellschaft / Registered Office: Donauwörth
> Registergericht / Registration Court: Amtsgericht Augsburg HRB 16508
> Vorsitzender des Aufsichtsrates / Chairman of the Supervisory Board: Dr. Lutz Bertling
> Geschäftsführung / Board of Management:
> Dr. Wolfgang Schoder, Vorsitzender / CEO; Friedrich-Wilhelm Hormel; Ralf Barnscheidt
>
>
> CONFIDENTIALITY NOTICE
>
> This communication and the information it contains is intended for the addressee ( s ) named above and for no other persons or organizations.
> It is confidential and may be legally privileged and protected by law. The unauthorized use, copying or disclosure of this communication
> or any part of it is prohibited and may be unlawful.
> If you have received this communication in error, kindly notify us by return e-mail and discard and/or delete the communication.
> Thank you very much.
> It is possible for e-mails to be intercepted or affected by viruses. Whilst we maintain virus checks on our e-mails, we accept no liability for viruses
> or other material which might be introduced with this message.
> _______________________________________________
> Kst mailing list
> Kst at kde.org
> https://mail.kde.org/mailman/listinfo/kst



-- 
C. Barth Netterfield
University of Toronto
416-845-0946