Adding this call simulates seeing a *different* bitmap every time we scale, which changes the code path (places much heavier load on constructing a new filter). This better emulates the kind of environment most embedders see us in, where they try to cache a scaled bitmap rather than frequently rescaling it.
This change will cause significant regressions in microbenchmark scores.
BUG=skia:2236
R=humper@google.com, reed@google.com, scroggo@google.com, tomhudson@google.com
Author: tomhudson@chromium.org
Review URL: https://codereview.chromium.org/183763047
git-svn-id: http://skia.googlecode.com/svn/trunk@13685 2bbb7eff-a529-9590-31e7-b0007b416f81
diff --git a/bench/BitmapScaleBench.cpp b/bench/BitmapScaleBench.cpp
index 926111d..f6c2da1 100644
--- a/bench/BitmapScaleBench.cpp
+++ b/bench/BitmapScaleBench.cpp
@@ -97,6 +97,7 @@
SkPaint paint;
paint.setFilterLevel(SkPaint::kHigh_FilterLevel);
+ fInputBitmap.notifyPixelsChanged();
canvas.drawBitmapMatrix( fInputBitmap, fMatrix, &paint );
}
private: