Yes this is awesome work you have done there. Do you plan to implement the real-data FFT, DCT and DST in pure Julia also? Then one could really think about moving FFTW into a package. Hopefully its author is ok with that ;-)
Am Montag, 21. April 2014 22:25:27 UTC+2 schrieb Steven G. Johnson: > > > > On Thursday, April 10, 2014 3:16:12 PM UTC-4, Tobias Knopp wrote: >> >> should be fairly easy to use fftpack which has single and double >> precision and is ok from the license. >> > > I already have significantly better performance than FFTPACK in my > pure-Julia FFT, as well as (in some ways [*]) significantly more > functionality: support for arbitrary prime factors, support for arbitrary > numeric types (e.g. already supports bigfloat, and would also work with > future quad-precision types etc), and support for strided data and > arbitrary multidimensional FFTs. And it requires about 1/3 as many lines > of code compared to the corresponding portions of FFTPACK. See > https://github.com/JuliaLang/julia/pull/6193 > > [*] currently missing real-data FFTs and DCTs/DSTs. >