I have about 2000 complex fft (size 2048) to calculate and the input/output buffers are allocated continuously. Therefore, I naturally think about using fftw_plan_many_dft i
