Support Float32 output from SkColorSpaceXform

tests/SkNxTest.cpp

 /*
  * Copyright 2015 Google Inc.
  *
  * Use of this source code is governed by a BSD-style license that can be
  * found in the LICENSE file.
  */
 
 #include "Sk4px.h"
 #include "SkNx.h"
 #include "SkRandom.h"
@@ skipping 289 matching lines @@
     }
 
     // A naive implementation with _mm_packs_epi32 would succeed up to 0x7fff but fail here:
     for (int i = 0x8000; (1) && i <= 0xffff; i++) {
         uint16_t expected = (uint16_t)i;
         uint16_t actual = SkNx_cast<uint16_t>(Sk4i(i))[0];
 
         REPORTER_ASSERT(r, expected == actual);
     }
 }
+
+DEF_TEST(SkNx_4fLoad4Store4, r) {
+    float src[] = {
+         0.0f,  1.0f,  2.0f,  3.0f,
+         4.0f,  5.0f,  6.0f,  7.0f,
+         8.0f,  9.0f, 10.0f, 11.0f,
+        12.0f, 13.0f, 14.0f, 15.0f
+    };
+
+    Sk4f a, b, c, d;
+    Sk4f_load4(src, &a, &b, &c, &d);
+    REPORTER_ASSERT(r,  0.0f == a[0]);
+    REPORTER_ASSERT(r,  4.0f == a[1]);
+    REPORTER_ASSERT(r,  8.0f == a[2]);
+    REPORTER_ASSERT(r, 12.0f == a[3]);
+    REPORTER_ASSERT(r,  1.0f == b[0]);
+    REPORTER_ASSERT(r,  5.0f == b[1]);
+    REPORTER_ASSERT(r,  9.0f == b[2]);
+    REPORTER_ASSERT(r, 13.0f == b[3]);
+    REPORTER_ASSERT(r,  2.0f == c[0]);
+    REPORTER_ASSERT(r,  6.0f == c[1]);
+    REPORTER_ASSERT(r, 10.0f == c[2]);
+    REPORTER_ASSERT(r, 14.0f == c[3]);
+    REPORTER_ASSERT(r,  3.0f == d[0]);
+    REPORTER_ASSERT(r,  7.0f == d[1]);
+    REPORTER_ASSERT(r, 11.0f == d[2]);
+    REPORTER_ASSERT(r, 15.0f == d[3]);
+
+    float dst[16];
+    Sk4f_store4(dst, a, b, c, d);
+    REPORTER_ASSERT(r, 0 == memcmp(dst, src, 16 * sizeof(float)));
+}