aom_dsp/x86/synonyms_avx2.h

   1 /*
   2  * Copyright (c) 2018, Alliance for Open Media. All rights reserved
   3  *
   4  * This source code is subject to the terms of the BSD 2 Clause License and
   5  * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
   6  * was not distributed with this source code in the LICENSE file, you can
   7  * obtain it at www.aomedia.org/license/software. If the Alliance for Open
   8  * Media Patent License 1.0 was not distributed with this source code in the
   9  * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
  10  */
  11
  12 #ifndef AOM_DSP_X86_SYNONYMS_AVX2_H_
  13 #define AOM_DSP_X86_SYNONYMS_AVX2_H_
  14
  15 #include <immintrin.h>
  16
  17 #include "config/aom_config.h"
  18
  19 #include "aom/aom_integer.h"
  20
  21 /**
  22  * Various reusable shorthands for x86 SIMD intrinsics.
  23  *
  24  * Intrinsics prefixed with xx_ operate on or return 128bit XMM registers.
  25  * Intrinsics prefixed with yy_ operate on or return 256bit YMM registers.
  26  */
  27
  28 // Loads and stores to do away with the tedium of casting the address
  29 // to the right type.
  30 static INLINE __m256i yy_load_256(const void *a) {
  31   return _mm256_load_si256((const __m256i *)a);
  32 }
  33
  34 static INLINE __m256i yy_loadu_256(const void *a) {
  35   return _mm256_loadu_si256((const __m256i *)a);
  36 }
  37
  38 static INLINE void yy_store_256(void *const a, const __m256i v) {
  39   _mm256_store_si256((__m256i *)a, v);
  40 }
  41
  42 static INLINE void yy_storeu_256(void *const a, const __m256i v) {
  43   _mm256_storeu_si256((__m256i *)a, v);
  44 }
  45
  46 // The _mm256_set1_epi64x() intrinsic is undefined for some Visual Studio
  47 // compilers. The following function is equivalent to _mm256_set1_epi64x()
  48 // acting on a 32-bit integer.
  49 static INLINE __m256i yy_set1_64_from_32i(int32_t a) {
  50 #if defined(_MSC_VER) && defined(_M_IX86) && _MSC_VER < 1900
  51   return _mm256_set_epi32(0, a, 0, a, 0, a, 0, a);
  52 #else
  53   return _mm256_set1_epi64x((uint32_t)a);
  54 #endif
  55 }
  56
  57 // Some compilers don't have _mm256_set_m128i defined in immintrin.h. We
  58 // therefore define an equivalent function using a different intrinsic.
  59 // ([ hi ], [ lo ]) -> [ hi ][ lo ]
  60 static INLINE __m256i yy_set_m128i(__m128i hi, __m128i lo) {
  61   return _mm256_insertf128_si256(_mm256_castsi128_si256(lo), hi, 1);
  62 }
  63
  64 #endif  // AOM_DSP_X86_SYNONYMS_AVX2_H_