3 ms·

> Once you have done that (and gained two or three order of magnitude in computation time), and you're still not fast enough, then okay, You are talking about

by af3 13y ago

> Once you have done that (and gained two or three order of magnitude in computation time), and you're still not fast enough, then okay, You are talking about the language (or better implementation) that is compiled to bytecode. The speed increase that you gain by optimization is of the order of magnitude smaller than by just using language that compiles to machine code. Once you understood how it's done in python, then you need to port it to fortran/c, what is the point? Fortran is not much different from python in syntax.

darkarmani 13y ago
> Once you understood how it's done in python, then you need to port it to fortran/c, what is the point? At that point, why would you port it to fortran/c instead of using FFTW? The whole point of the exercise was to learn about and optimize the algorithm. If you can learn the easiest using python, why wouldn't you do your exploration in python, since going from python to your language of choice should be the easiest part of the exercise if you choose to do that.