Performance and memcpy in C++ API
Created originally on Bitbucket by avmo (Ashwin Vishnu)
I recently read your paper about fluidFFT and I was trying to use the C++ API, however, I perceived that most of the classes use memcpy (i.e. fft2d_with fftw2d.cpp), is there some way to avoid this? even for the 2d case, it can degrade the performance if a lot of FFTs are done
- Oscar Ladino