>
> I tested this new scalefac_scale method as same as you did....
> Uuum it is "not" bit-for-bit identical, but is only LSB order,
> so it seems because of mpg123's floating point rounding problem.
>
> please test and comment this.
> ---
> Takehiro TOMINAGA // may the source be with you!
> --
Looks good to me!
But a note about testing:
I dont think floating point rounding could explain the differences,
since decoding the scalefacs in mpg123 is probably all integer ops?
I tested it, but got bit-for-bit identical results. But as I found
when testing best_huffman_divide(), you have to do two things:
use the --nores option, and move the ResvAdjust() call.
(see the commented out part of quantize.c, line 115, latest version)
If you dont move ResvAdjust outside the loop, then any bits that are saved
in gr=0 (because of the more efficient code) will become available
for gr=1, and then the output will become different.
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
