Straight FFTW3.3.10 is faster

[alaw950-lgtm]

The generic FFTW 3.3.10 compiled with --enable-float --enable-avx2 --enable-avx512 is faster than this thing compiled as suggested with the --enable-float thrown in.

[BiplabRaut]

Hi,
Can you please share the machine, compiler and OS that you are running on ?

Also, we highly recommend you to evaluate : https://github.com/amd/aocl-fftz