Re: FFT
- From: Phil Carmody <thefatphil_demunged@xxxxxxxxxxx>
- Date: 18 May 2006 09:07:18 +0300
"heiack" <heiack@xxxxxx> writes:
Hey, I'am a student of Computer Science, and we have a problem.
We want to compute a lot of FFT's more than 10,000 in a very samll
time, without the use of hardware acceleration.
So I'll ve to design a new algorithm.
What is wrong with the dozens or hundreds of FFT algorithms
that have already been designed?
But fortunatly we are doing the following.
First we compute a FFT about a array of 256 complex values, then we
multiply these array with a constant coefficient a do the same
again.(serverall thousends times).
The 'constant coefficients' would be the FFT of a convolution?
Are you just performing a convolution?
If you look at the computed FFT arrays they lock nearly the same.
That's called entropy. Almost all vectors look the same - random.
Or it's called a bug in your implementation. Have you compared
your results to the results from any of the dozens of other
FFT implementations out there?
If so, and the results are the same - what's you're problem?
If so, and the results are different - fix your bugs, duh!
If not, why the heck not?
Can me someone say me how I can use this in a algorithum
So you need to do a task that millions of people have been doing for
decades?
And you've decided that all the research over those decades is of
no use to you?
So you've designed your own algorithm?
And now you want to know how to use it?
and we have a problem.
Well, you're correct about one thing.
Phil
--
The man who is always worrying about whether or not his soul would be
damned generally has a soul that isn't worth a damn.
-- Oliver Wendell Holmes, Sr. (1809-1894), American physician and writer
.
- References:
- FFT
- From: heiack
- FFT
- Prev by Date: Re: Triangle altitudes intersect
- Next by Date: Re: chess problem
- Previous by thread: FFT
- Next by thread: Re: FFT
- Index(es):
Relevant Pages
|
