AuroraMiddleware/skia2
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Further improve AA choice

Browse Source 
Compared with current master:
    1.26% faster in Seal_of_New_Hampshire.svg_1
    1.90% faster in native_image_to_raster_surface
    2.70% faster in Seal_of_Texas.svg_1
    2.76% faster in WestVirginia-StateSeal.svg_1
    2.80% faster in chart_aa
    3.36% faster in Iowa-StateSeal.svg_1
    3.44% faster in Seal_of_Illinois.svg_1
    4.37% faster in Seal_of_Maryland_%28reverse%29.svg_1
    4.64% faster in Minnesota-StateSeal.svg_1
    5.06% faster in Seal_of_Tennessee.svg_1
   16.76% faster in clip_strategy_path_100

Compared with the Skia without https://skia-review.googlesource.com/c/skia/+/146220
(no significant slowdowns for any SVGs)
    4.69% slower in path_fill_big_rotated_rect_aa_45
    2.50% slower in points
    1.27% faster in native_image_to_raster_surface
    1.39% faster in patch_square_meshlines_0.100000x0.100000
    2.09% faster in path_hairline_big_noAA_line
    2.62% faster in Seal_of_Maryland_%28reverse%29.svg_1
    3.63% faster in rotated_rects_aa_alternating_transparent_and_opaque_darken
    3.93% faster in zeroradroundrect
    4.60% faster in rotated_rects_aa_same_transparent_darken
    4.64% faster in ovals_1
    4.73% faster in rotated_rects_aa_changing_opaque_darken
    4.74% faster in rotated_rects_aa_changing_transparent_darken
    5.01% faster in rotated_rects_aa_same_opaque_darken
    5.07% faster in nested_aaclip_AA
    6.06% faster in ovals_3
    6.25% faster in dash_0_noclip
    7.86% faster in rotated_rects_aa_same_transparent_src
    7.86% faster in rotated_rects_aa_changing_transparent_src
    9.31% faster in rrects_3
    9.35% faster in rotated_rects_aa_alternating_transparent_and_opaque_src
   10.95% faster in rotated_rects_aa_same_transparent_srcover
   10.95% faster in rotated_rects_aa_changing_transparent_srcover
   11.46% faster in rotated_rects_aa_alternating_transparent_and_opaque_srcover
   11.90% faster in rotated_rects_aa_same_opaque_srcover
   11.90% faster in rotated_rects_aa_changing_opaque_srcover
   12.09% faster in rotated_rects_aa_same_opaque_src
   12.13% faster in rotated_rects_aa_changing_opaque_src
   20.89% faster in path_fill_small_triangle

Bug: skia:8272
Change-Id: Ic1434ea71660b0d3a7ddf1f63d098cfe7acfd861
Reviewed-on: https://skia-review.googlesource.com/149337
Auto-Submit: Yuqian Li <liyuqian@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>
This commit is contained in:
Yuqian Li 2018-08-25 00:13:41 -07:00 committed by Skia Commit-Bot
parent 3853bafbde
commit 04ccd72c1a
1 changed files with 6 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
src/core/SkScan_AntiPath.cpp
View File

@ -694,6 +694,12 @@ static bool ShouldUseAAA(const SkPath& path, SkScalar avgLength, SkScalar comple
// AA might be slower than supersampling.
return path.countPoints() < SkTMax(bounds.width(), bounds.height()) / 2 - 10;
#else
if (path.countPoints() >= path.getBounds().height()) {
// SAA is faster than AAA in this case even if there are no intersections because AAA will
// have too many scan lines. See skbug.com/8272
return false;
}
// We will use AAA if the number of verbs < kSampleSize and therefore complexity < 0
return complexity < kComplexityThreshold;
#endif