Plus xfermode using Sk4px.

src/opts/SkNx_sse.h

Index: src/opts/SkNx_sse.h
diff --git a/src/opts/SkNx_sse.h b/src/opts/SkNx_sse.h
index b3339f9957a2ca5041f1a730c9f7f219a13b9fc2..9423d0455122cf37d50653ca7db3b254bdf2a3ac 100644
--- a/src/opts/SkNx_sse.h
+++ b/src/opts/SkNx_sse.h
@@ -296,6 +296,8 @@ public:
 
     void store(uint8_t vals[16]) const { _mm_storeu_si128((__m128i*)vals, fVec); }
 
+    SkNi saturatedAdd(const SkNi& o) const { return _mm_adds_epu8(fVec, o.fVec); }
+
     SkNi operator + (const SkNi& o) const { return _mm_add_epi8(fVec, o.fVec); }
     SkNi operator - (const SkNi& o) const { return _mm_sub_epi8(fVec, o.fVec); }