Re:Show HN: A minimal FFT code
by jacobolus on Wednesday December 26, 2018 @01:02PM
(#18763500)
A lisp implementation is cute and can be a nice teaching aid but is slower than desirable for practical applications.
What I would really like to see is a reasonably short (small code size) but reasonably optimized (for an arbitrary machine that might run a browser) Web Assembly implementation of a power-of-2 FFT, possibly compiled down from another language, e.g. C or Rust. Throw in an implementation of Bluestein’s non-power-of-2 FFT for bonus points.
The existing Javascript implementations I have seen are much slower than they could be if a serious expert took a crack at it, while the existing compiled-to-Javascript or -wasm implementations weren’t originally designed with the web as a target, and tend to be bloated monsters.
What I would really like to see is a reasonably short (small code size) but reasonably optimized (for an arbitrary machine that might run a browser) Web Assembly implementation of a power-of-2 FFT, possibly compiled down from another language, e.g. C or Rust. Throw in an implementation of Bluestein’s non-power-of-2 FFT for bonus points.
The existing Javascript implementations I have seen are much slower than they could be if a serious expert took a crack at it, while the existing compiled-to-Javascript or -wasm implementations weren’t originally designed with the web as a target, and tend to be bloated monsters.