[FFmpeg-devel] aac: Use the MDCT scaling rather than scale factor offset hacks

Alex Converse alex.converse
Fri Oct 9 18:41:48 CEST 2009


On Fri, Oct 9, 2009 at 10:03 AM, Robert Swain <robert.swain at gmail.com> wrote:
> 2009/10/8 Alex Converse <alex.converse at gmail.com>:
>> On Wed, Oct 7, 2009 at 4:10 PM, Alex Converse <alex.converse at gmail.com> wrote:
>>> This patch also folds the minus sign in there.
>>
>> Update for r20192.
>
> Looks OK to me. Does this improve performance noticably? I wonder if
> we can reduce the size of ff_aac_pow2sf_tab[] if the encoder doesn't
> use all of it.
>

The only noticeable performance increases is in the unloved main
profile. Performance remains the same for LC (including IS and PNS
streams).

Currently the encoder uses the outlying values in that table but I
intend on fixing that next.



More information about the ffmpeg-devel mailing list