[MPlayer-dev-eng] [PATCH] vf_ass: avoid a division

[Reimar Döffinger]

On Fri, Sep 17, 2010 at 09:51:49PM +0200, Nicolas George wrote:
> Le jour de la Vertu, an CCXVIII, Reimar Döffinger a écrit :
> > Go ahead (though note that may suggestion is to check the src before
> > instead of after the multiplication).
> 
> Then I will take a little bit more time to benchmark both cases.
> 
> > Using 258 has the advantage that the shifts in the inner loop
> > are bytewise, for which a compiler can generate fast code
> > even for architectures that do not have fast arbitrary shifts.
> 
> If such architectures exist and are supported, I'll keep that in mind.

No need to bother too much with it. Most of them are not really
relevant, though I suspect that P4 might be one of them,
a bswap should be faster than a shift on it and would work just as
well for the >> 24 case (though I have some doubts any compilers
are clever enough to use this regardless).