Support Float32 output from SkColorSpaceXform

src/opts/SkNx_sse.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 SkNx_sse_DEFINED
 #define SkNx_sse_DEFINED
 
@@ skipping 460 matching lines @@
 static inline void Sk4h_store4(void* dst, const Sk4h& r, const Sk4h& g, const Sk4h& b,
                                const Sk4h& a) {
     __m128i rg = _mm_unpacklo_epi16(r.fVec, g.fVec);
     __m128i ba = _mm_unpacklo_epi16(b.fVec, a.fVec);
     __m128i lo = _mm_unpacklo_epi32(rg, ba);
     __m128i hi = _mm_unpackhi_epi32(rg, ba);
     _mm_storeu_si128(((__m128i*) dst) + 0, lo);
     _mm_storeu_si128(((__m128i*) dst) + 1, hi);
 }
 
+static inline void Sk4f_load4(const void* ptr, Sk4f* r, Sk4f* g, Sk4f* b, Sk4f* a) {
+    __m128 v0 = _mm_loadu_ps(((float*)ptr) +  0),
+           v1 = _mm_loadu_ps(((float*)ptr) +  4),
+           v2 = _mm_loadu_ps(((float*)ptr) +  8),
+           v3 = _mm_loadu_ps(((float*)ptr) + 12);
+    _MM_TRANSPOSE4_PS(v0, v1, v2, v3);
+    *r = v0;
+    *g = v1;
+    *b = v2;
+    *a = v3;
+}
+
+static inline void Sk4f_store4(void* dst, const Sk4f& r, const Sk4f& g, const Sk4f& b,
+                               const Sk4f& a) {
+    __m128 v0 = r.fVec,
+           v1 = g.fVec,
+           v2 = b.fVec,
+           v3 = a.fVec;
+    _MM_TRANSPOSE4_PS(v0, v1, v2, v3);
+    _mm_storeu_ps(((float*) dst) +  0, v0);
+    _mm_storeu_ps(((float*) dst) +  4, v1);
+    _mm_storeu_ps(((float*) dst) +  8, v2);
+    _mm_storeu_ps(((float*) dst) + 12, v3);
+}
+
 #endif//SkNx_sse_DEFINED