I hope it's faster.

Open tingfeng-key opened this issue 6 years ago • 2 comments

I hope it's faster for next_frame

Jul 06 '19 15:07 tingfeng-key

First step would be to add test vectors + benchmark vectors (#1 and #2). Then we could find bottlenecks and test improvements to the specific algorithms. Best would be to port some of the very optimized algorithms from minimp3. Most of the current code are the straight-forward algorithms I converted from PDMP3, etc.

Then, we could investigate using SIMD/threads to improve performance (packed_simd and rayon?). However, I don't believe these work with wasm32 currently, which is my main target.

Jul 08 '19 16:07 Herschel

Yes, minimp3 is fast. I'm using it now, but I don't know C and C++. I can't help you with this improvement. I'm sorry.

Jul 14 '19 09:07 tingfeng-key