Possibility to split up FFT data into smaller chunks

[Vyper]

Does anyone here know if there is possible in a clean manner to convert a 65536 & 131072 point FFT data to only use 16384 points as a maximum at a time, or perhaps splitting them up in a 16K maximum FFT points at a time?

Kind Regards Vyper

[Vyper]

With points i mean FFT plans, how large plans do the S@H client use..

65536 or 131072 or perhaps even more??

Someone with the knowledge, Josef.. Ben.. Heeelp 

[Josef W. Segur]

With points i mean FFT plans, how large plans do the S@H client use..

65536 or 131072 or perhaps even more??

Someone with the knowledge, Josef.. Ben.. Heeelp 

131072 is the maximum in use, but that's controlled by values in the workunit header and a properly designed program should be able to do larger ones too.

All practical FFT algorithms do divide the work into smaller units and combine them for the final result. But the way it is split and recombined is not trivial. I can't go any further since I haven't studied the algorithms in detail.

One possible area you might search is the projects trying to find Mersenne primes. They use very long FFTs in a method for multiplying huge numbers, and it's likely someone in those communities has investigated using a GPU.
                                                                                Joe

[Vyper]

Hans found an interesting article here at FFTW's homepage about splitting FFT chunks..

That approach may be useable for him (us), getting back with response l8r..

http://www.fftw.org/pruned.html

Btw, thanks Josef for your concern.. Much appreciated with theoretical input to get past stuck problems..


Kind Regards Vyper

[Nebby]

According to the link below, nVidia plans on increasing it's FFT size restrictions in a later release of CUDA.

http://forums.nvidia.com/lofiversion/index.php?t28384.html