SkNx Load/store: take any pointer.

src/opts/SkNx_avx.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_avx_DEFINED
 #define SkNx_avx_DEFINED
 
 // This file may assume <= AVX, but must check SK_CPU_SSE_LEVEL for anything more recent.
 
 // All the SSE specializations are still good ideas.  We'll just add Sk8f.
 #include "SkNx_sse.h"
 
 // SkNx_sse.h defines SKNX_IS_FAST.
 
 namespace {  // See SkNx.h
 
 template <>
 class SkNx<8, float> {
 public:
     SkNx(const __m256& vec) : fVec(vec) {}
 
     SkNx() {}
     SkNx(float val) : fVec(_mm256_set1_ps(val)) {}
-    static SkNx Load(const float vals[8]) { return _mm256_loadu_ps(vals); }
+    static SkNx Load(const void* ptr) { return _mm256_loadu_ps((const float*)ptr); }
 
     SkNx(float a, float b, float c, float d,
          float e, float f, float g, float h) : fVec(_mm256_setr_ps(a,b,c,d,e,f,g,h)) {}
 
-    void store(float vals[8]) const { _mm256_storeu_ps(vals, fVec); }
+    void store(void* ptr) const { _mm256_storeu_ps((float*)ptr, fVec); }
 
     SkNx operator + (const SkNx& o) const { return _mm256_add_ps(fVec, o.fVec); }
     SkNx operator - (const SkNx& o) const { return _mm256_sub_ps(fVec, o.fVec); }
     SkNx operator * (const SkNx& o) const { return _mm256_mul_ps(fVec, o.fVec); }
     SkNx operator / (const SkNx& o) const { return _mm256_div_ps(fVec, o.fVec); }
 
     SkNx operator == (const SkNx& o) const { return _mm256_cmp_ps(fVec, o.fVec, _CMP_EQ_OQ); }
     SkNx operator != (const SkNx& o) const { return _mm256_cmp_ps(fVec, o.fVec, _CMP_NEQ_OQ); }
     SkNx operator  < (const SkNx& o) const { return _mm256_cmp_ps(fVec, o.fVec, _CMP_LT_OQ); }
     SkNx operator  > (const SkNx& o) const { return _mm256_cmp_ps(fVec, o.fVec, _CMP_GT_OQ); }
@@ skipping 42 matching lines @@
     __m128i _16 = _mm_unpacklo_epi8(src.fVec, _mm_setzero_si128()),
              lo = _mm_unpacklo_epi16(_16, _mm_setzero_si128()),
              hi = _mm_unpackhi_epi16(_16, _mm_setzero_si128());
     __m256i _32 = _mm256_insertf128_si256(_mm256_castsi128_si256(lo), hi, 1);
     return _mm256_cvtepi32_ps(_32);
 }
 
 }  // namespace
 
 #endif//SkNx_avx_DEFINED