FFTW plan

FFTW plan

by Anton Siebert -
Number of replies: 1

Hi, while working on the gravitational wave analysis, I came up with a question regarding the rigidity of the FFTW plan: Specifically, since the GWprediction.nc and all 32 detectionXX.nc files share the same nt dimensions, is it possible to reuse a single plan for different files that contain different data? Or is the plan strictly tied to the specific memory addresses/data values of the first array it was initialized with?

Additionally, for the purpose of the assignment requirements, are we expected to reuse a single plan for efficiency, or is it acceptable to create and delete a new plan for every signal we analyze?

Thanks!


In reply to Anton Siebert

Re: FFTW plan

by Ramses van Zon -
In principle, it is possible to use a plan with another array, using the 'guru' interface of fftw3, but there are restrictions on the previous and new arrays, and I would nor recommend it.
 
But you should indeed reuse plans as much as possible. Have you considered loading the data of each case into the same existing array?