P
Phil Hobbs
Guest
On 2023-03-02 10:36, Joe Gwinn wrote:
Only if the kernel occupies the whole interval for all values of the
parameters. Wavelet transforms, for instance, are more or less well
localized in both time and frequency.
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510
http://electrooptical.net
http://hobbs-eo.com
On Thu, 02 Mar 2023 11:17:08 +0200, upsidedown@downunder.com wrote:
On Wed, 01 Mar 2023 15:26:08 -0500, Joe Gwinn <joegwinn@comcast.net
wrote:
But the more fundamental limitation is Amdahl\'s Law, which applies to
all parallel computations:
.<https://en.wikipedia.org/wiki/Amdahl%27s_law
This is why it\'s hard to parallelize for instance a FFT.
If you have a lot of cores, why not use discrete FT instead of FFT,
especially if you do not need all points or the input size is
something different from some power of 2 ?
Actually, the problem is with all integral transforms, regardless of
the existence of a fast algorithm, because by definition an integral
transform uses the entire input to generate each point in the output.
Only if the kernel occupies the whole interval for all values of the
parameters. Wavelet transforms, for instance, are more or less well
localized in both time and frequency.
But there are problems where there are things that can be computed in
parallel independently. These are called \"embarrassingly parallel\"
problems.
Joe Gwinn
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510
http://electrooptical.net
http://hobbs-eo.com