MATLAB: Are FFTW library routines used for FFT in MATLAB 6.5 (R13) and later versions optimized for multiply-accumulate instructions and vector processors with AltiVec technology

fastfourierifftinverseMATLABtransform

I would like to know if FFTW library routines used for FFT in MATLAB 6.5 (R13) and later versions are optimized for multiply-accumulate instructions and vector processors with AltiVec technology.

Best Answer

FFTW is not optimized for the multiply-accumulate instructions or specific processors. The version of the FFTW library used in MATLAB 6.5 (R13) and later versions has been patched by MathWorks, and you cannot replace it with your own compiled versions.