[FFmpeg-devel] [PATCH] VP8: faster deblock strength calculation

[Jason Garrett-Glaser]

On Fri, Feb 4, 2011 at 4:54 AM, Ronald S. Bultje <rsbultje at gmail.com> wrote:
> PS,
>
> On Fri, Feb 4, 2011 at 1:49 AM, Jason Garrett-Glaser <jason at x264.com> wrote:
>> - ? ?mbedge_lim = 2*(filter_level+2) + inner_limit;
>> - ? ? bedge_lim = 2* filter_level ? ?+ inner_limit;
> [..]
>> + ? ? bedge_lim = 2*filter_level + inner_limit;
>> + ? ?mbedge_lim = bedge_lim + 4;
> [..]
>> @@ -1577,8 +1578,8 @@ static av_always_inline void filter_mb_simple(VP8Context *s, uint8_t *dst, VP8Fi
>> ? ? if (!filter_level)
>> ? ? ? ? return;
>>
>> - ? ?mbedge_lim = 2*(filter_level+2) + inner_limit;
>> - ? ? bedge_lim = 2* filter_level ? ?+ inner_limit;
>> + ? ? bedge_lim = 2*filter_level + inner_limit;
>> + ? ?mbedge_lim = bedge_lim + 4;
>
> I would not be entirely surprised if gcc could come up with something
> similar, I've seen it made such decisions in the past. Of course, it
> never hurts to help gcc along a little, so you should still apply...
>
> Ronald
>

Applied.  At least here, GCC was dumb there.

Jason