1 /* 2 * Copyright (c) 2018-2021 Arm Limited. 3 * 4 * SPDX-License-Identifier: MIT 5 * 6 * Permission is hereby granted, free of charge, to any person obtaining a copy 7 * of this software and associated documentation files (the "Software"), to 8 * deal in the Software without restriction, including without limitation the 9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or 10 * sell copies of the Software, and to permit persons to whom the Software is 11 * furnished to do so, subject to the following conditions: 12 * 13 * The above copyright notice and this permission notice shall be included in all 14 * copies or substantial portions of the Software. 15 * 16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 22 * SOFTWARE. 23 */ 24 #ifndef ARM_COMPUTE_WRAPPER_TRAITS_H 25 #define ARM_COMPUTE_WRAPPER_TRAITS_H 26 27 #include <arm_neon.h> 28 29 #if defined(ARM_COMPUTE_ENABLE_SVE) 30 #include <arm_sve.h> 31 #endif /* defined(ARM_COMPUTE_ENABLE_SVE) */ 32 33 namespace arm_compute 34 { 35 namespace wrapper 36 { 37 namespace traits 38 { 39 // *INDENT-OFF* 40 // clang-format off 41 42 /** 64-bit vector tag */ 43 struct vector_64_tag {}; 44 /** 128-bit vector tag */ 45 struct vector_128_tag {}; 46 47 /** Create the appropriate SIMD vector given its type and size in terms of elements */ 48 template <typename T, int S> struct neon_vector; 49 50 // Specializations 51 #ifndef DOXYGEN_SKIP_THIS 52 template <> struct neon_vector<uint8_t, 8>{ using scalar_type = uint8_t; using type = uint8x8_t; using tag_type = vector_64_tag; }; 53 template <> struct neon_vector<int8_t, 8>{ using scalar_type = int8_t; using type = int8x8_t; using tag_type = vector_64_tag; }; 54 template <> struct neon_vector<uint8_t, 16>{ using scalar_type = uint8_t; using type = uint8x16_t; using tag_type = vector_128_tag; }; 55 template <> struct neon_vector<int8_t, 16>{ using scalar_type = int8_t; using type = int8x16_t; using tag_type = vector_128_tag; }; 56 template <> struct neon_vector<uint16_t, 4>{ using scalar_type = uint16_t; using type = uint16x4_t; using tag_type = vector_64_tag; }; 57 template <> struct neon_vector<int16_t, 4>{ using scalar_type = int16_t; using type = int16x4_t; using tag_type = vector_64_tag; }; 58 template <> struct neon_vector<uint16_t, 8>{ using scalar_type = uint16_t; using type = uint16x8_t; using tag_type = vector_128_tag; }; 59 template <> struct neon_vector<uint16_t, 16>{ using scalar_type = uint16_t; using type = uint16x8x2_t; }; 60 template <> struct neon_vector<int16_t, 8>{ using scalar_type = int16_t; using type = int16x8_t; using tag_type = vector_128_tag; }; 61 template <> struct neon_vector<int16_t, 16>{ using scalar_type = int16_t; using type = int16x8x2_t; }; 62 template <> struct neon_vector<uint32_t, 2>{ using scalar_type = uint32_t; using type = uint32x2_t; using tag_type = vector_64_tag; }; 63 template <> struct neon_vector<int32_t, 2>{ using scalar_type = int32_t; using type = int32x2_t; using tag_type = vector_64_tag; }; 64 template <> struct neon_vector<uint32_t, 4>{ using scalar_type = uint32_t; using type = uint32x4_t; using tag_type = vector_128_tag; }; 65 template <> struct neon_vector<int32_t, 4>{ using scalar_type = int32_t; using type = int32x4_t; using tag_type = vector_128_tag; }; 66 template <> struct neon_vector<uint64_t, 1>{ using scalar_type = uint64_t;using type = uint64x1_t; using tag_type = vector_64_tag; }; 67 template <> struct neon_vector<int64_t, 1>{ using scalar_type = int64_t; using type = int64x1_t; using tag_type = vector_64_tag; }; 68 template <> struct neon_vector<uint64_t, 2>{ using scalar_type = uint64_t; using type = uint64x2_t; using tag_type = vector_128_tag; }; 69 template <> struct neon_vector<int64_t, 2>{ using scalar_type = int64_t; using type = int64x2_t; using tag_type = vector_128_tag; }; 70 template <> struct neon_vector<float_t, 2>{ using scalar_type = float_t; using type = float32x2_t; using tag_type = vector_64_tag; }; 71 template <> struct neon_vector<float_t, 4>{ using scalar_type = float_t; using type = float32x4_t; using tag_type = vector_128_tag; }; 72 73 #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC 74 template <> struct neon_vector<float16_t, 4>{ using scalar_type = float16_t; using type = float16x4_t; using tag_type = vector_64_tag; }; 75 template <> struct neon_vector<float16_t, 8>{ using scalar_type = float16_t; using type = float16x8_t; using tag_type = vector_128_tag; }; 76 #endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC 77 #endif /* DOXYGEN_SKIP_THIS */ 78 79 /** Helper type template to get the type of a neon vector */ 80 template <typename T, int S> using neon_vector_t = typename neon_vector<T, S>::type; 81 /** Helper type template to get the tag type of a neon vector */ 82 template <typename T, int S> using neon_vector_tag_t = typename neon_vector<T, S>::tag_type; 83 84 /** Vector bit-width enum class */ 85 enum class BitWidth 86 { 87 W64, /**< 64-bit width */ 88 W128, /**< 128-bit width */ 89 }; 90 91 /** Create the appropriate SIMD vector given its type and size in terms of bits */ 92 template <typename T, BitWidth BW> struct neon_bitvector; 93 // Specializations 94 #ifndef DOXYGEN_SKIP_THIS 95 template <> struct neon_bitvector<uint8_t, BitWidth::W64>{ using type = uint8x8_t; using tag_type = vector_64_tag; }; 96 template <> struct neon_bitvector<int8_t, BitWidth::W64>{ using type = int8x8_t; using tag_type = vector_64_tag; }; 97 template <> struct neon_bitvector<uint8_t, BitWidth::W128>{ using type = uint8x16_t; using tag_type = vector_128_tag; }; 98 template <> struct neon_bitvector<int8_t, BitWidth::W128>{ using type = int8x16_t; using tag_type = vector_128_tag; }; 99 template <> struct neon_bitvector<uint16_t, BitWidth::W64>{ using type = uint16x4_t; using tag_type = vector_64_tag; }; 100 template <> struct neon_bitvector<int16_t, BitWidth::W64>{ using type = int16x4_t; using tag_type = vector_64_tag; }; 101 template <> struct neon_bitvector<uint16_t, BitWidth::W128>{ using type = uint16x8_t; using tag_type = vector_128_tag; }; 102 template <> struct neon_bitvector<int16_t, BitWidth::W128>{ using type = int16x8_t; using tag_type = vector_128_tag; }; 103 template <> struct neon_bitvector<uint32_t, BitWidth::W64>{ using type = uint32x2_t; using tag_type = vector_64_tag; }; 104 template <> struct neon_bitvector<int32_t, BitWidth::W64>{ using type = int32x2_t; using tag_type = vector_64_tag; }; 105 template <> struct neon_bitvector<uint32_t, BitWidth::W128>{ using type = uint32x4_t; using tag_type = vector_128_tag; }; 106 template <> struct neon_bitvector<int32_t, BitWidth::W128>{ using type = int32x4_t; using tag_type = vector_128_tag; }; 107 template <> struct neon_bitvector<uint64_t, BitWidth::W64>{ using type = uint64x1_t; using tag_type = vector_64_tag; }; 108 template <> struct neon_bitvector<int64_t, BitWidth::W64>{ using type = int64x1_t; using tag_type = vector_64_tag; }; 109 template <> struct neon_bitvector<uint64_t, BitWidth::W128>{ using type = uint64x2_t; using tag_type = vector_128_tag; }; 110 template <> struct neon_bitvector<int64_t, BitWidth::W128>{ using type = int64x2_t; using tag_type = vector_128_tag; }; 111 template <> struct neon_bitvector<float_t, BitWidth::W64>{ using type = float32x2_t; using tag_type = vector_64_tag; }; 112 template <> struct neon_bitvector<float_t, BitWidth::W128>{ using type = float32x4_t; using tag_type = vector_128_tag; }; 113 #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC 114 template <> struct neon_bitvector<float16_t, BitWidth::W64>{ using type = float16x4_t; using tag_type = vector_64_tag; }; 115 template <> struct neon_bitvector<float16_t, BitWidth::W128>{ using type = float16x8_t; using tag_type = vector_128_tag; }; 116 #endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC 117 118 119 #if defined(ARM_COMPUTE_ENABLE_SVE) 120 /** Create the appropriate SVE vector given its type */ 121 template <typename T> struct sve_vector; 122 123 template <> struct sve_vector<uint8_t>{ using scalar_type = uint8_t; using type = svuint8_t; }; 124 template <> struct sve_vector<int8_t>{ using scalar_type = int8_t; using type = svint8_t; }; 125 #endif /* defined(ARM_COMPUTE_ENABLE_SVE) */ 126 127 #endif /* DOXYGEN_SKIP_THIS */ 128 129 /** Helper type template to get the type of a neon vector */ 130 template <typename T, BitWidth BW> using neon_bitvector_t = typename neon_bitvector<T, BW>::type; 131 /** Helper type template to get the tag type of a neon vector */ 132 template <typename T, BitWidth BW> using neon_bitvector_tag_t = typename neon_bitvector<T, BW>::tag_type; 133 134 /** Promote a type */ 135 template <typename T> struct promote { }; 136 template <> struct promote<uint8_t> { using type = uint16_t; }; 137 template <> struct promote<int8_t> { using type = int16_t; }; 138 template <> struct promote<uint16_t> { using type = uint32_t; }; 139 template <> struct promote<int16_t> { using type = int32_t; }; 140 template <> struct promote<uint32_t> { using type = uint64_t; }; 141 template <> struct promote<int32_t> { using type = int64_t; }; 142 template <> struct promote<float> { using type = float; }; 143 template <> struct promote<half> { using type = half; }; 144 145 /** Get promoted type */ 146 template <typename T> 147 using promote_t = typename promote<T>::type; 148 149 // clang-format on 150 // *INDENT-ON* 151 } // namespace traits 152 } // namespace wrapper 153 } // namespace arm_compute 154 #endif /* ARM_COMPUTE_WRAPPER_TRAITS_H */ 155