Hi all, Another day, another little optimisation. I've added lookup tables to subs.c/fht() to streamline the initial swapping operation for n==256 and 1024. This gives an overall speedup of 1-2% on my machine (PIII 500/NT4SP5/MSVC). The tables (particularly the n==1024 table) are quite large, so it probably causes cache havoc, but the function is called often enough for it to pay off... new version attached. -- Mat.
- Re: [MP3 ENCODER] Small speedup for subs.c/fht() Mathew Hendry
- Re: [MP3 ENCODER] Small speedup for subs.c/fht() Takehiro Tominaga
subs_fht.c