>>> XRPOW_FTOI(x, rx);
>>should be
>>> XRPOW_FTOI(x-.5, rx);
>>?
No, it is right.
But Acy missed to fix adj43asm initialization.
details are in my old mail on this ML.
>>quantize_xrpow_ISO from Acy's last message, but also went back to the
>>old fft.c routines. Takehiro's code had the minor force_ms problem,
>>but also produced changes in the bitstream. Not unusual if you change
>>something like the fft, but it breaks my validation tests. In such
uum, re-check in the old revision file is not a good idea for CVS...
because we can get the older revision with CVS.
new FFT is "public beta" and I checked it in because I want everyone
to check the new code.
But it makes some trouble for some people want to get "more stable one".
I am planing to make an another branch to avoid it.
does anybody have ideas ?
---
Takehiro TOMINAGA
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
