| commit | 908276b3969cf8f8eec28026363897134c0e54e0 | [log] [tgz] |
|---|---|---|
| author | senorblanco@chromium.org <senorblanco@chromium.org@2bbb7eff-a529-9590-31e7-b0007b416f81> | Thu Nov 15 20:27:35 2012 +0000 |
| committer | senorblanco@chromium.org <senorblanco@chromium.org@2bbb7eff-a529-9590-31e7-b0007b416f81> | Thu Nov 15 20:27:35 2012 +0000 |
| tree | 55fb3b7689972532535358a0f5cb53e5bd3809d7 | |
| parent | 0561a3ccd24e46259d825adf5507c7ee9531b9c5 [diff] |
Reorder the separable blur passes into XXX/YYY, with an image transpose on the last pass of each group. This results in continuguous memory reads in all passes, giving a 22% speedup on theverge.skp over the previous separable implementation, and a 30%-50% improvement over the existing implementation (depending on platform). Review URL: https://codereview.appspot.com/6851053 git-svn-id: http://skia.googlecode.com/svn/trunk@6445 2bbb7eff-a529-9590-31e7-b0007b416f81