On Fri, Sep 6, 2013 at 12:09 PM, Hendrik Leppkes <[email protected]> wrote:
> On Fri, Sep 6, 2013 at 12:02 PM, Vittorio Giovara <
> [email protected]> wrote:
>
>> Il giorno venerdì 6 settembre 2013, Luca Barbato ha scritto:
>>
>> > On 03/09/13 16:46, Vittorio Giovara wrote:
>> > > ---
>> > >  libavcodec/eamad.c     |    2 +-
>> > >  libavcodec/mpeg12dec.c |    9 ++-------
>> > >  2 files changed, 3 insertions(+), 8 deletions(-)
>> > >
>> >
>> > Benchmark it, if it is faster this way then it is over-inlining, if is
>> > not just add inline in the function you are renaming.
>> >
>>
>> Umh I'm not renaming anything I'm just making a function skip an
>> unconditional jump and the inline doesn't help there (much) afaik. But sure
>> I'll benchmark it like I did for the other function in mpeg12enc.
>>
>>
> When the function is inlined, its "n" parameter can be made a constant in
> the loop in decode_mb, which can make the generated ASM a bit
> simpler/faster.
> A benchmark might be a good idea.
>

Too little improvement to notice any difference, patch dropped for now.
Vittorio
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to