----- Kurosu <kur...@free.fr> a écrit :
> Le 17/06/2010 11:43, Kurosu a écrit :
> > 188757 29.5856 wormux BlitNtoNPixelAlpha
> > 109184 17.1134 wormux BlitRGBtoRGBPixelAlphaMMX
>
> Those are all used in case of per-pixel alpha blending. Now, there are 2
> things this makes me think:
> - do we really need that much alpha blending? Aren't some of those
> surfaces either 255 or 0?
Study still pending... Maybe using colorkey is what is needed.
> - the former function is for ARGB + BGRA blending and is more costly;
> where are the BGRA textures coming from? Can't they be treated as ARGB?
This is due to us forcing ARGB (through masks) for new surfaces. By instead
using {R,G,B} masks from the actual display surface, there's no longer calls to
BlitNtoNPixelAlpha (they are all going to BlitRGBtoRGBPixelAlphaMMX), which I
expect to be a win, considering BlitRGBtoRGBPixelAlphaMMX is SIMD.
_______________________________________________
Wormux-dev mailing list
Wormux-dev@gna.org
https://mail.gna.org/listinfo/wormux-dev
