[Theora-dev] Re: Theora mcomp tuning...

[Adam D. Moss]

Adam D. Moss wrote:
> The win from this patch is usually only 1-2% improvement
> in video bandwidth (compared to a trunk build with Wim's
> fix already added, for fair comparison) but I've seen as
> much as 20% improvement

The 20% bump was probably bogus -- my test method had a
bug. :)

So while this patch still seems statistically to be the right
thing to do, the quality/bandwidth savings are hardly noticable
(though feel free to try for yourself... if anyone is still
using the non-MMX encoder ;)).

--Adam