Several fixes for fp 16 rendering:

dm/DM.cpp

Index: dm/DM.cpp
diff --git a/dm/DM.cpp b/dm/DM.cpp
index 388546d66fcf23c309634bc0f43d5fbb5d7db101..ff36b6152a7c088d41c20a1956cd6e73c2a3406b 100644
--- a/dm/DM.cpp
+++ b/dm/DM.cpp
@@ -958,8 +958,15 @@ static bool dump_png(SkBitmap bitmap, const char* path, const char* md5) {
         if (!px) {
             return false;
         }
+        const Sk4f vecZero(0);
+        const Sk4f vecOne(1);
         for (int i = 0; i < w*h; i++) {
-            Sk4f fs = SkHalfToFloat_01(px[i]);    // Convert up to linear floats.

[mtklein, 2016/04/14 19:20:13]

FYI folks, I do intend to replace SkHalfToFloat_01() with
SkHalfToFloat_finite(), which should let us get back to something short here.

+            // Convert up to linear floats.
+            Sk4f fs(SkHalfToFloat(static_cast<SkHalf>(px[i] >> (0 * 16))),
+                    SkHalfToFloat(static_cast<SkHalf>(px[i] >> (1 * 16))),
+                    SkHalfToFloat(static_cast<SkHalf>(px[i] >> (2 * 16))),
+                    SkHalfToFloat(static_cast<SkHalf>(px[i] >> (3 * 16))));
+            fs = Sk4f::Min(Sk4f::Max(fs, vecZero), vecOne);  // Clamp

[mtklein, 2016/04/14 19:20:13]

I like to write this as

   fs = Sk4f::Max(0.0f, Sk4f::Min(fs, 1.0f));

It puts the value visually between its min and max.

(You can use scalars and literals with Sk4f.  I think even Sk4f::Max(0,
Sk4f::Min(fs, 1)) is fine here.)

             float invA = 1.0f / fs[3];
             fs = fs * Sk4f(invA, invA, invA, 1);  // Unpremultiply.
             rgba[i] = Sk4f_toS32(fs);             // Pack down to sRGB bytes.