[Elecraft] Faster FFT...off topic but relevant

[Jessie Oberreuter]

      I should note that it didn't catch my interest because the domains
/I/ generally throw FFTs against contain large numbers of signals with a 
lot of dynamic range between them, and I really do need most of the data. 
OTOH, there are almost certainly a HUGE number of domains for which 
faster/cheaper isolation of the top k signals would be a big win.
      The most common "break throughs" I see in industry occur when a fresh 
pair of eyes looks at a problem and says, "wait, you're using the wrong 
approach -- there's a much better tool for this case".  Nine out of ten, 
it's also either been around for 50+ years and has just been forgotten, or 
it's just far enough out of the domain that most of the folks on the job 
were never exposed to it.  Thus, /knowing/ that something like this exists 
and the kinds of probglems it is good for is far more important than 
trying to brute force apply it to every problem.  -kb7psg


On Tue, 31 Jan 2012, Jessie Oberreuter wrote:

>
>      I only glanced at the academic paper when it went by a few weeks ago,
> but IIRC, it offers improved performance for low [dominant] information
> content samples.  It's been 15+ years since I did the math, but I thought
> the original function series did this well.  The FFT optimizations,
> however, don't -- rather like the way it's often faster to compute
> multiple results and throw away the ones you don't need, than it is to
> make constant culling decisions along the way.  Thus, I looked at the
> paper, said to myself, "this may be a good tool when I'm trying to save
> battery power or when I know that I only need to identify a small number
> strong signals, but it isn't going to change the world" and moved on :).
> -kb7psg
>
>
>
> On Tue, 31 Jan 2012, John Ragle wrote:
>
>> Does anyone know any details about the "new" ultra-fast Katabi FFT, its
>> coding, etc. as reported in the most recent New Scientist?
>>
>> John Ragle -- W1ZI
>>
>> --
>> Sent from my lovely old Dell XPS 420
>>
>> ______________________________________________________________
>> Elecraft mailing list
>> Home: http://mailman.qth.net/mailman/listinfo/elecraft
>> Help: http://mailman.qth.net/mmfaq.htm
>> Post: mailto:Elecraft at mailman.qth.net
>>
>> This list hosted by: http://www.qsl.net
>> Please help support this email list: http://www.qsl.net/donate.html
>>
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:Elecraft at mailman.qth.net
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>