ts-7000
[Top] [All Lists]

Re: [ts-7000] Re: FFT very slow on TS-7260

To:
Subject: Re: [ts-7000] Re: FFT very slow on TS-7260
From: Curtis Monroe <>
Date: Sun, 2 Dec 2007 14:49:17 -0500
High speed FFT in ARM ASM (ch08_dsp.zip):

http://books.elsevier.com/us//mk/us/subindex.asp?maintarget=companions/defaultindividual.asp&isbn=1558608745&country=United+States&srccode=&ref=&subcode=&head=&pdf=&basiccode=&txtSearch=&SearchField=&operator=&order=&community=mk

This is from the book "ARM System Developer's Guide". Its probably the most 
optimized code for FFT (and other math functions).

Its been a while since I read the book, hope it helps.

-Curtis.




On November 30, 2007 02:44 pm, kurmannthomas wrote:
> Thanks to all for their replies...
> Does anyone know about an open source Fixed-Point FFT? Ive found one 
> or two, but they only go upto 1024 points, which in my case would be 
> 10s of real data @ 100 Samples per second...not that much..I think 
> the change to fixed point would be easiest for me since im a 
> hardware guy and not specificly a linux prog. :(
> 
> Thanks alot
> Best regards
> Tom
> 
> 
> 
> --- In  "Daniel Serpell" 
> <> wrote:
> >
> > Hi!
> > 
> > On Nov 30, 2007 6:36 AM, kurmannthomas <> wrote:
> > >
> > [...]
> > > Im going to check with another floating point FFT next week to 
> see if
> > > its an FFTW3 problem, or really the TS-7260. My goal is to be 
> able to
> > > calculate (up to) 9 FFTs with (up to) 16384 points in less than 
> 1 or
> > > 2 seconds...a DSP would have no problem with this...
> > 
> > I made some benchmarks on this board, using a mix of 50% 
> multiplications,
> > 50% additions, and I'm getting:
> > 
> > Double precision, emulated float:  0.23 mega-ops/second (~870 
> cycles per op)
> > Double precision, soft-float:  2.0 mega-ops/second (~104 cycles 
> per op)
> > Single precision, soft-float: 3.7 mega-ops/second (~54 cycles per 
> op)
> > 
> > A split-radix FFT implementation uses 4N*log2(N)-6N+8 muls and 
> adds,
> > so an 16384 point FFT will take a total of 1.6*10^6 operations, 
> 50% muls
> > and 50% adds.
> > 
> > So, using single-precision, soft-float, you should get about two 
> FFTs per
> > second, not counting filling the arrays, converting from int, etc.
> > 
> > I think you need a fixed-point implementation of the FFT.
> > 
> >     Daniel.
> >
> 
> 
> 


 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/ts-7000/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/ts-7000/join
    (Yahoo! ID required)

<*> To change settings via email:
     
    

<*> To unsubscribe from this group, send an email to:
    

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/
 

<Prev in Thread] Current Thread [Next in Thread>
Admin

Disclaimer: Neither Andrew Taylor nor the University of NSW School of Computer and Engineering take any responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the birding-aus mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU