Any plans to publish SharpFFTW on Nuget?

[PhilPJL]

Nice work btw :)

[wo80]

Nice work btw :)

Thanks!

This project hasn't received a lot of attention, so I haven't thought about publishing a Nuget package. The package probably wouldn't need much maintenance, so if you need it, I can publish one.

[PhilPJL]

No it's ok, if needed I can re-use your code.
From your benchmarks it looks like FFTW is 40x faster than Math.Net which if true would be amazing. Unfortunately my initial attempts to use FFTW in our code only results in possibly 2x faster.
These are the results I get after modifying your test harness to work with a non-ideal test waveform with noise and calling initialize before each FFT.

I think FFTW may be worth the effort but it will need some work to use optimally in a multi-threaded app compared to Math.Net.

[wo80]

In the benchmark I tried to measure only the FFT (the Math.NET test for example includes a data copy, which should be O(n) at most, but still might affect the test result). The tests as they are implemented do the memory allocation inside the Initialize method, so if you include the method in the benchmark, in worst case you are measuring the GC kicking in. Seeing complex FFTW on the last place is a bit surprising.