Check my latest MDCT routine.
MDCT routine in my old snapshot has lot of bugs, but in latest one,
most of them are debugged. I think latest one is almost perfect...
My mdct routine is containing 3 routine which was called mdct_sub,
window_subband, and filtering_subband.
And it contains these optimization...
- loop integration
- removed sampling data buffering and inversion
- removed zi storing
- pointer/array handling
- reduced calculation by using these points.
* mm[0][i] is always same.
* short block window function has cyclicity.
Note: this routine removed buffering, and it may makes some timing
problem. You should give it "480 byte lagged sampling data pointer".
And not only mdct.c but also tables.c is changed.
On my linuxbox with Celeron464MHz, it is faster than original(lame3.33)
about 10%.
---
Takehiro TOMINAGA // May the source be with you!
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
