3-15% speedup to HardLight / Overlay xfermodes.

src/core/Sk4pxXfermode.h

 /*
  * Copyright 2015 Google Inc.
  *
  * Use of this source code is governed by a BSD-style license that can be
  * found in the LICENSE file.
  */
 
 #ifndef Sk4pxXfermode_DEFINED
 #define Sk4pxXfermode_DEFINED
 
@@ skipping 50 matching lines @@
     // There's no chance of underflow, and if we subtract p before adding src+dst, no overflow.
     return (s - p) + (d - p.zeroAlphas());
 }
 
 XFERMODE(HardLight) {
     auto alphas = SrcOver::Xfer(s,d);
 
     auto sa = s.alphas(),
          da = d.alphas();
 
-    auto isLite = (sa-s) < s;
+    auto isLite = ((sa-s) < s).widenLoHi();
 
     auto dark = s*d << 1,
          lite = sa*da - ((da-d)*(sa-s) << 1),
          both = s*da.inv() + d*sa.inv();
 
-    // TODO: do isLite in 16-bit so we only have to div255() once.
-    auto colors = isLite.thenElse((lite + both).div255(),
-                                  (dark + both).div255());
+    auto colors = (both + isLite.thenElse(lite, dark)).div255();
     return alphas.zeroColors() + colors.zeroAlphas();
 }
 XFERMODE(Overlay) { return HardLight::Xfer(d,s); }
 
 XFERMODE(Darken) {
     auto sda = s.approxMulDiv255(d.alphas()),
          dsa = d.approxMulDiv255(s.alphas());
     auto srcover = s + (d - dsa),
          dstover = d + (s - sda);
     auto alphas = srcover,
@@ skipping 183 matching lines @@
 #endif
         default: break;
     }
 #endif
     return nullptr;
 }
 
 } // namespace
 
 #endif//Sk4pxXfermode_DEFINED