Closed pr1metine closed 1 year ago
I have not looked very carefully (yet), but on first glance this looks very good! Nice work!
Thanks for the feedback and your kind words 🙏. I‘ll give you guys a couple of days, then I‘ll incorporate your feedback
Anything else, @ejmahler ?
Thanks again for your kind words 🙏. LMK if I can improve anything else.
I have not found anything more to comment on, all looks good to me.
Thanks! I'll make a release tomorrow.
Thank you too 🙏 🙏. I'm glad to be a part of this
Hey, it's been a while. I need WASM SIMD for work. Can we push a new release containing this feature?
Should fix #89. The code for WASM SIMD is heavily inspired by the NEON implementation. Run unit tests using
wasm-pack test
:When running FFTs on f32 data, we can expect twice the performance when compared to scalar algorithms. This means, that WASM SIMD performance is on par with native AArch64 w/o Neon.
mul_complex_f64
:https://github.com/pr1metine/RustFFT/blob/33ff96aa347e82cc4e05409885093cfa256fd9e2/src/wasm_simd/wasm_simd_utils.rs#L171-L176
Additionally in some cases, I could not find equivalent WASM operations for NEON. Relaxed SIMD should help once it is implemented.