Re: [AD] blender operations SRC and DST

[allefant <info@xxxxxxxxxx>]

On Tue, Jun 7, 2011 at 8:18 PM, Jon Rafkind <workmin@xxxxxxxxxxx> wrote:

I changed the patch further:
 * altered internal/aintern_blend.h to deal with the new factors
 * al_blend_memory just calls _al_blend_inline now instead of having its own implementation of blending

Here is ex_blend_bench before and after the patch (summary is 2-5% slowdown). I have a 1.8ghz amd processor.
no change:
 Plain blit: 213.454 FPS
 Scaled blit: 213.883 FPS
 Rotated blit: 223.992 FPS

my change:
 Plain blit: 201.183 FPS
 Scaled blit: 208.2 FPS
 Rotated blit: 214.747 FPS

2-5% sounds harmless enough. But maybe wait for Peter, he spent a lot of effort into optimizing this and might thing differently :) Also, did you figure out when the "unoptimized reference implementation" (that comment needs to be removed in your patch) is used? Maybe can remove it completely now.