| Quellcodebibliothek Statistik Leitseite products/Sources/formale Sprachen/C/Firefox/third_party/simde/simde/x86/ (Browser von der Mozilla Stiftung Version 136.0.1©) Datei vom 10.2.2025 mit Größe 260 kB |
|
Quelle sse2.h Sprache: C |
|||||||||||||||
|
/* SPDX-License-Identifier: MIT * * Permission is hereby granted, free of charge, to any person * obtaining a copy of this software and associated documentation * files (the "Software"), to deal in the Software without * restriction, including without limitation the rights to use, copy, * modify, merge, publish, distribute, sublicense, and/or sell copies * of the Software, and to permit persons to whom the Software is * furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be * included in all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. * * Copyright: * 2017-2020 Evan Nemerson <evan@nemerson.com> * 2015-2017 John W. Ratcliff <jratcliffscarab@gmail.com> * 2015 Brandon Rowlett <browlett@nvidia.com> * 2015 Ken Fast <kfast@gdeb.com> * 2017 Hasindu Gamaarachchi <hasindu@unsw.edu.au> * 2018 Jeff Daily <jeff.daily@amd.com> */ #if !defined(SIMDE_X86_SSE2_H) #define SIMDE_X86_SSE2_H #include "sse.h" HEDLEY_DIAGNOSTIC_PUSH SIMDE_DISABLE_UNWANTED_DIAGNOSTICS SIMDE_BEGIN_DECLS_ typedef union { #if defined(SIMDE_VECTOR_SUBSCRIPT) SIMDE_ALIGN_TO_16 int8_t i8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 int16_t i16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 int32_t i32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 int64_t i64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint8_t u8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint16_t u16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint32_t u32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint64_t u64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; #if defined(SIMDE_HAVE_INT128_) SIMDE_ALIGN_TO_16 simde_int128 i128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 simde_uint128 u128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; #endif SIMDE_ALIGN_TO_16 simde_float32 f32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 simde_float64 f64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 int_fast32_t i32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint_fast32_t u32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; #else SIMDE_ALIGN_TO_16 int8_t i8[16]; SIMDE_ALIGN_TO_16 int16_t i16[8]; SIMDE_ALIGN_TO_16 int32_t i32[4]; SIMDE_ALIGN_TO_16 int64_t i64[2]; SIMDE_ALIGN_TO_16 uint8_t u8[16]; SIMDE_ALIGN_TO_16 uint16_t u16[8]; SIMDE_ALIGN_TO_16 uint32_t u32[4]; SIMDE_ALIGN_TO_16 uint64_t u64[2]; #if defined(SIMDE_HAVE_INT128_) SIMDE_ALIGN_TO_16 simde_int128 i128[1]; SIMDE_ALIGN_TO_16 simde_uint128 u128[1]; #endif SIMDE_ALIGN_TO_16 simde_float32 f32[4]; SIMDE_ALIGN_TO_16 simde_float64 f64[2]; SIMDE_ALIGN_TO_16 int_fast32_t i32f[16 / sizeof(int_fast32_t)]; SIMDE_ALIGN_TO_16 uint_fast32_t u32f[16 / sizeof(uint_fast32_t)]; #endif SIMDE_ALIGN_TO_16 simde__m64_private m64_private[2]; SIMDE_ALIGN_TO_16 simde__m64 m64[2]; #if defined(SIMDE_X86_SSE2_NATIVE) SIMDE_ALIGN_TO_16 __m128i n; #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) SIMDE_ALIGN_TO_16 int8x16_t neon_i8; SIMDE_ALIGN_TO_16 int16x8_t neon_i16; SIMDE_ALIGN_TO_16 int32x4_t neon_i32; SIMDE_ALIGN_TO_16 int64x2_t neon_i64; SIMDE_ALIGN_TO_16 uint8x16_t neon_u8; SIMDE_ALIGN_TO_16 uint16x8_t neon_u16; SIMDE_ALIGN_TO_16 uint32x4_t neon_u32; SIMDE_ALIGN_TO_16 uint64x2_t neon_u64; #if defined(__ARM_FP16_FORMAT_IEEE) SIMDE_ALIGN_TO_16 float16x8_t neon_f16; #endif SIMDE_ALIGN_TO_16 float32x4_t neon_f32; #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) SIMDE_ALIGN_TO_16 float64x2_t neon_f64; #endif #elif defined(SIMDE_MIPS_MSA_NATIVE) v16i8 msa_i8; v8i16 msa_i16; v4i32 msa_i32; v2i64 msa_i64; v16u8 msa_u8; v8u16 msa_u16; v4u32 msa_u32; v2u64 msa_u64; #elif defined(SIMDE_WASM_SIMD128_NATIVE) SIMDE_ALIGN_TO_16 v128_t wasm_v128; #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed char) altivec_i8; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed short) altivec_i16; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32; #if defined(__UINT_FAST32_TYPE__) && (defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(__INT_FAST32_TYPE__) altivec_i32f; #else SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32f; #endif SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) altivec_u8; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned short) altivec_u16; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32; #if defined(__UINT_FAST32_TYPE__) && (defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(__UINT_FAST32_TYPE__) altivec_u32f; #else SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32f; #endif SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(float) altivec_f32; #if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed long long) altivec_i64; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long) altivec_u64; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(double) altivec_f64; #endif #endif } simde__m128i_private; typedef union { #if defined(SIMDE_VECTOR_SUBSCRIPT) SIMDE_ALIGN_TO_16 int8_t i8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 int16_t i16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 int32_t i32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 int64_t i64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint8_t u8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint16_t u16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint32_t u32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint64_t u64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 simde_float32 f32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 simde_float64 f64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 int_fast32_t i32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; SIMDE_ALIGN_TO_16 uint_fast32_t u32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; #else SIMDE_ALIGN_TO_16 int8_t i8[16]; SIMDE_ALIGN_TO_16 int16_t i16[8]; SIMDE_ALIGN_TO_16 int32_t i32[4]; SIMDE_ALIGN_TO_16 int64_t i64[2]; SIMDE_ALIGN_TO_16 uint8_t u8[16]; SIMDE_ALIGN_TO_16 uint16_t u16[8]; SIMDE_ALIGN_TO_16 uint32_t u32[4]; SIMDE_ALIGN_TO_16 uint64_t u64[2]; SIMDE_ALIGN_TO_16 simde_float32 f32[4]; SIMDE_ALIGN_TO_16 simde_float64 f64[2]; SIMDE_ALIGN_TO_16 int_fast32_t i32f[16 / sizeof(int_fast32_t)]; SIMDE_ALIGN_TO_16 uint_fast32_t u32f[16 / sizeof(uint_fast32_t)]; #endif SIMDE_ALIGN_TO_16 simde__m64_private m64_private[2]; SIMDE_ALIGN_TO_16 simde__m64 m64[2]; #if defined(SIMDE_X86_SSE2_NATIVE) SIMDE_ALIGN_TO_16 __m128d n; #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) SIMDE_ALIGN_TO_16 int8x16_t neon_i8; SIMDE_ALIGN_TO_16 int16x8_t neon_i16; SIMDE_ALIGN_TO_16 int32x4_t neon_i32; SIMDE_ALIGN_TO_16 int64x2_t neon_i64; SIMDE_ALIGN_TO_16 uint8x16_t neon_u8; SIMDE_ALIGN_TO_16 uint16x8_t neon_u16; SIMDE_ALIGN_TO_16 uint32x4_t neon_u32; SIMDE_ALIGN_TO_16 uint64x2_t neon_u64; SIMDE_ALIGN_TO_16 float32x4_t neon_f32; #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) SIMDE_ALIGN_TO_16 float64x2_t neon_f64; #endif #elif defined(SIMDE_MIPS_MSA_NATIVE) v16i8 msa_i8; v8i16 msa_i16; v4i32 msa_i32; v2i64 msa_i64; v16u8 msa_u8; v8u16 msa_u16; v4u32 msa_u32; v2u64 msa_u64; #elif defined(SIMDE_WASM_SIMD128_NATIVE) SIMDE_ALIGN_TO_16 v128_t wasm_v128; #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed char) altivec_i8; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed short) altivec_i16; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32; #if defined(__INT_FAST32_TYPE__) && (defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined( SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(__INT_FAST32_TYPE__) altivec_i32f; #else SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32f; #endif SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) altivec_u8; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned short) altivec_u16; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32; #if defined(__UINT_FAST32_TYPE__) && (defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(__UINT_FAST32_TYPE__) altivec_u32f; #else SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32f; #endif SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(float) altivec_f32; #if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed long long) altivec_i64; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long) altivec_u64; SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(double) altivec_f64; #endif #endif } simde__m128d_private; #if defined(SIMDE_X86_SSE2_NATIVE) typedef __m128i simde__m128i; typedef __m128d simde__m128d; #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) typedef int64x2_t simde__m128i; # if defined(SIMDE_ARM_NEON_A64V8_NATIVE) typedef float64x2_t simde__m128d; # elif defined(SIMDE_VECTOR_SUBSCRIPT) typedef simde_float64 simde__m128d SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; # else typedef simde__m128d_private simde__m128d; # endif #elif defined(SIMDE_WASM_SIMD128_NATIVE) typedef v128_t simde__m128i; typedef v128_t simde__m128d; #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI typedef SIMDE_POWER_ALTIVEC_VECTOR(float) simde__m128i; #if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE typedef SIMDE_POWER_ALTIVEC_VECTOR(double) simde__m128d; #else typedef simde__m128d_private simde__m128d; #endif #elif defined(SIMDE_VECTOR_SUBSCRIPT) typedef int64_t simde__m128i SIMDE_ALIGN_TO_16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS; typedef simde_float64 simde__m128d SIMDE_ALIGN_TO_16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS #else typedef simde__m128i_private simde__m128i; typedef simde__m128d_private simde__m128d; #endif #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) typedef simde__m128i __m128i; typedef simde__m128d __m128d; #endif HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128i), "simde__m128i size incorrect"); HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128i_private), "simde__m128i_private size HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128d), "simde__m128d size incorrect"); HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128d_private), "simde__m128d_private size #if defined(SIMDE_CHECK_ALIGNMENT) && defined(SIMDE_ALIGN_OF) HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128i) == 16, "simde__m128i is not 16-by HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128i_private) == 16, "simde__m128i_priv HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128d) == 16, "simde__m128d is not 16-by HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128d_private) == 16, "simde__m128d_priv #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde__m128i_from_private(simde__m128i_private v) { simde__m128i r; simde_memcpy(&r, &v, sizeof(r)); return r; } SIMDE_FUNCTION_ATTRIBUTES simde__m128i_private simde__m128i_to_private(simde__m128i v) { simde__m128i_private r; simde_memcpy(&r, &v, sizeof(r)); return r; } SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde__m128d_from_private(simde__m128d_private v) { simde__m128d r; simde_memcpy(&r, &v, sizeof(r)); return r; } SIMDE_FUNCTION_ATTRIBUTES simde__m128d_private simde__m128d_to_private(simde__m128d v) { simde__m128d_private r; simde_memcpy(&r, &v, sizeof(r)); return r; } #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int8x16_t, neon, i8) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int16x8_t, neon, i16) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int32x4_t, neon, i32) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int64x2_t, neon, i64) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint8x16_t, neon, u8) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint16x8_t, neon, u16) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint32x4_t, neon, u32) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint64x2_t, neon, u64) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, float32x4_t, neon, f32) #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, float64x2_t, neon, f64) #endif #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, SIMDE_POWER_ALTIVEC_VECTOR(signed ch SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, SIMDE_POWER_ALTIVEC_VECTOR(signed sh SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, SIMDE_POWER_ALTIVEC_VECTOR(signed in SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, SIMDE_POWER_ALTIVEC_VECTOR(unsigned SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, SIMDE_POWER_ALTIVEC_VECTOR(unsigned SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, SIMDE_POWER_ALTIVEC_VECTOR(unsigned #if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, SIMDE_POWER_ALTIVEC_VECTOR(unsigned SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, SIMDE_POWER_ALTIVEC_VECTOR(signed lo #endif #endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */ #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int8x16_t, neon, i8) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int16x8_t, neon, i16) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int32x4_t, neon, i32) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int64x2_t, neon, i64) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint8x16_t, neon, u8) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint16x8_t, neon, u16) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint32x4_t, neon, u32) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint64x2_t, neon, u64) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, float32x4_t, neon, f32) #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, float64x2_t, neon, f64) #endif #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, SIMDE_POWER_ALTIVEC_VECTOR(signed ch SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, SIMDE_POWER_ALTIVEC_VECTOR(signed sh SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, SIMDE_POWER_ALTIVEC_VECTOR(signed in SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, SIMDE_POWER_ALTIVEC_VECTOR(unsigned SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, SIMDE_POWER_ALTIVEC_VECTOR(unsigned SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, SIMDE_POWER_ALTIVEC_VECTOR(unsigned #if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, SIMDE_POWER_ALTIVEC_VECTOR(unsigned SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, SIMDE_POWER_ALTIVEC_VECTOR(signed lo #if defined(SIMDE_BUG_GCC_95782) SIMDE_FUNCTION_ATTRIBUTES SIMDE_POWER_ALTIVEC_VECTOR(double) simde__m128d_to_altivec_f64(simde__m128d value) { simde__m128d_private r_ = simde__m128d_to_private(value); return r_.altivec_f64; } SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde__m128d_from_altivec_f64(SIMDE_POWER_ALTIVEC_VECTOR(double) value) { simde__m128d_private r_; r_.altivec_f64 = value; return simde__m128d_from_private(r_); } #else SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, SIMDE_POWER_ALTIVEC_VECTOR(double), #endif #endif #elif defined(SIMDE_WASM_SIMD128_NATIVE) SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, v128_t, wasm, v128); SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, v128_t, wasm, v128); #endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */ SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_set_pd (simde_float64 e1, simde_float64 e0) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_set_pd(e1, e0); #else simde__m128d_private r_; #if defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_f64x2_make(e0, e1); #elif defined(SIMDE_ARM_NEON_A64V8_NATIVE) SIMDE_ALIGN_TO_16 simde_float64 data[2] = { e0, e1 }; r_.neon_f64 = vld1q_f64(data); #else r_.f64[0] = e0; r_.f64[1] = e1; #endif return simde__m128d_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_set_pd(e1, e0) simde_mm_set_pd(e1, e0) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_set1_pd (simde_float64 a) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_set1_pd(a); #else simde__m128d_private r_; #if defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_f64x2_splat(a); #elif defined(SIMDE_ARM_NEON_A64V8_NATIVE) r_.neon_f64 = vdupq_n_f64(a); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI r_.altivec_f64 = vec_splats(HEDLEY_STATIC_CAST(double, a)); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i64) / sizeof(r_.i64[0])) ; i++) { r_.f64[i] = a; } #endif return simde__m128d_from_private(r_); #endif } #define simde_mm_set_pd1(a) simde_mm_set1_pd(a) #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_set1_pd(a) simde_mm_set1_pd(a) #define _mm_set_pd1(a) simde_mm_set1_pd(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_x_mm_abs_pd(simde__m128d a) { #if defined(SIMDE_X86_SSE2_NATIVE) simde_float64 mask_; uint64_t u64_ = UINT64_C(0x7FFFFFFFFFFFFFFF); simde_memcpy(&mask_, &u64_, sizeof(u64_)); return _mm_and_pd(_mm_set1_pd(mask_), a); #else simde__m128d_private r_, a_ = simde__m128d_to_private(a); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) r_.neon_f64 = vabsq_f64(a_.neon_f64); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI r_.altivec_f64 = vec_abs(a_.altivec_f64); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_f64x2_abs(a_.wasm_v128); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.f64) / sizeof(r_.f64[0])) ; i++) { r_.f64[i] = simde_math_fabs(a_.f64[i]); } #endif return simde__m128d_from_private(r_); #endif } SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_x_mm_not_pd(simde__m128d a) { #if defined(SIMDE_X86_AVX512VL_NATIVE) __m128i ai = _mm_castpd_si128(a); return _mm_castsi128_pd(_mm_ternarylogic_epi64(ai, ai, ai, 0x55)); #else simde__m128d_private r_, a_ = simde__m128d_to_private(a); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i32 = vmvnq_s32(a_.neon_i32); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) r_.altivec_f64 = vec_nor(a_.altivec_f64, a_.altivec_f64); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) r_.altivec_i32 = vec_nor(a_.altivec_i32, a_.altivec_i32); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_v128_not(a_.wasm_v128); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i32f = ~a_.i32f; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])) ; i++) { r_.i32f[i] = ~(a_.i32f[i]); } #endif return simde__m128d_from_private(r_); #endif } SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_x_mm_select_pd(simde__m128d a, simde__m128d b, simde__m128d mask) { /* This function is for when you want to blend two elements together * according to a mask. It is similar to _mm_blendv_pd, except that * it is undefined whether the blend is based on the highest bit in * each lane (like blendv) or just bitwise operations. This allows * us to implement the function efficiently everywhere. * * Basically, you promise that all the lanes in mask are either 0 or * ~0. */ #if defined(SIMDE_X86_SSE4_1_NATIVE) return _mm_blendv_pd(a, b, mask); #else simde__m128d_private r_, a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b), mask_ = simde__m128d_to_private(mask); #if defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i64 = a_.i64 ^ ((a_.i64 ^ b_.i64) & mask_.i64); #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i64 = vbslq_s64(mask_.neon_u64, b_.neon_i64, a_.neon_i64); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i64) / sizeof(r_.i64[0])) ; i++) { r_.i64[i] = a_.i64[i] ^ ((a_.i64[i] ^ b_.i64[i]) & mask_.i64[i]); } #endif return simde__m128d_from_private(r_); #endif } SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_add_epi8 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_add_epi8(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i8 = vaddq_s8(a_.neon_i8, b_.neon_i8); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) r_.altivec_i8 = vec_add(a_.altivec_i8, b_.altivec_i8); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i8x16_add(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i8 = a_.i8 + b_.i8; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i8) / sizeof(r_.i8[0])) ; i++) { r_.i8[i] = a_.i8[i] + b_.i8[i]; } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_add_epi8(a, b) simde_mm_add_epi8(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_add_epi16 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_add_epi16(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i16 = vaddq_s16(a_.neon_i16, b_.neon_i16); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) r_.altivec_i16 = vec_add(a_.altivec_i16, b_.altivec_i16); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i16x8_add(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i16 = a_.i16 + b_.i16; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i16) / sizeof(r_.i16[0])) ; i++) { r_.i16[i] = a_.i16[i] + b_.i16[i]; } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_add_epi16(a, b) simde_mm_add_epi16(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_add_epi32 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_add_epi32(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i32 = vaddq_s32(a_.neon_i32, b_.neon_i32); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) r_.altivec_i32 = vec_add(a_.altivec_i32, b_.altivec_i32); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i32x4_add(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i32 = a_.i32 + b_.i32; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i32) / sizeof(r_.i32[0])) ; i++) { r_.i32[i] = a_.i32[i] + b_.i32[i]; } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_add_epi32(a, b) simde_mm_add_epi32(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_add_epi64 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_add_epi64(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i64 = vaddq_s64(a_.neon_i64, b_.neon_i64); #elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE) r_.altivec_i64 = vec_add(a_.altivec_i64, b_.altivec_i64); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i64x2_add(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i64 = a_.i64 + b_.i64; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i64) / sizeof(r_.i64[0])) ; i++) { r_.i64[i] = a_.i64[i] + b_.i64[i]; } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_add_epi64(a, b) simde_mm_add_epi64(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_add_pd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_add_pd(a, b); #else simde__m128d_private r_, a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) r_.neon_f64 = vaddq_f64(a_.neon_f64, b_.neon_f64); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_f64x2_add(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) r_.altivec_f64 = vec_add(a_.altivec_f64, b_.altivec_f64); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_f64x2_add(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.f64 = a_.f64 + b_.f64; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.f64) / sizeof(r_.f64[0])) ; i++) { r_.f64[i] = a_.f64[i] + b_.f64[i]; } #endif return simde__m128d_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_add_pd(a, b) simde_mm_add_pd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_move_sd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_move_sd(a, b); #else simde__m128d_private r_, a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) r_.neon_f64 = vsetq_lane_f64(vgetq_lane_f64(b_.neon_f64, 0), a_.neon_f64, 0); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) #if defined(HEDLEY_IBM_VERSION) r_.altivec_f64 = vec_xxpermdi(a_.altivec_f64, b_.altivec_f64, 1); #else r_.altivec_f64 = vec_xxpermdi(b_.altivec_f64, a_.altivec_f64, 1); #endif #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i64x2_shuffle(a_.wasm_v128, b_.wasm_v128, 2, 1); #elif defined(SIMDE_SHUFFLE_VECTOR_) r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, b_.f64, 2, 1); #else r_.f64[0] = b_.f64[0]; r_.f64[1] = a_.f64[1]; #endif return simde__m128d_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_move_sd(a, b) simde_mm_move_sd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_x_mm_broadcastlow_pd(simde__m128d a) { /* This function broadcasts the first element in the input vector to * all lanes. It is used to avoid generating spurious exceptions in * *_sd functions since there may be garbage in the upper lanes. */ #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_castsi128_pd(_mm_shuffle_epi32(_mm_castpd_si128(a), 0x44)); #else simde__m128d_private r_, a_ = simde__m128d_to_private(a); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) r_.neon_f64 = vdupq_laneq_f64(a_.neon_f64, 0); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) r_.altivec_f64 = vec_splat(a_.altivec_f64, 0); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_f64x2_splat(a_.f64[0]); #elif defined(SIMDE_SHUFFLE_VECTOR_) r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, a_.f64, 0, 0); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.f64) / sizeof(r_.f64[0])) ; i++) { r_.f64[i] = a_.f64[0]; } #endif return simde__m128d_from_private(r_); #endif } SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_add_sd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_add_sd(a, b); #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS) return simde_mm_move_sd(a, simde_mm_add_pd(a, b)); #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) return simde_mm_move_sd(a, simde_mm_add_pd(simde_x_mm_broadcastlow_pd(a), simde_x_m #else simde__m128d_private r_, a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); r_.f64[0] = a_.f64[0] + b_.f64[0]; r_.f64[1] = a_.f64[1]; return simde__m128d_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_add_sd(a, b) simde_mm_add_sd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m64 simde_mm_add_si64 (simde__m64 a, simde__m64 b) { #if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) return _mm_add_si64(a, b); #else simde__m64_private r_, a_ = simde__m64_to_private(a), b_ = simde__m64_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i64 = vadd_s64(a_.neon_i64, b_.neon_i64); #else r_.i64[0] = a_.i64[0] + b_.i64[0]; #endif return simde__m64_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_add_si64(a, b) simde_mm_add_si64(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_adds_epi8 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_adds_epi8(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i8 = vqaddq_s8(a_.neon_i8, b_.neon_i8); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i8x16_add_sat(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) r_.altivec_i8 = vec_adds(a_.altivec_i8, b_.altivec_i8); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i8) / sizeof(r_.i8[0])) ; i++) { r_.i8[i] = simde_math_adds_i8(a_.i8[i], b_.i8[i]); } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_adds_epi8(a, b) simde_mm_adds_epi8(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_adds_epi16 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_adds_epi16(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i16 = vqaddq_s16(a_.neon_i16, b_.neon_i16); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i16x8_add_sat(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) r_.altivec_i16 = vec_adds(a_.altivec_i16, b_.altivec_i16); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i16) / sizeof(r_.i16[0])) ; i++) { r_.i16[i] = simde_math_adds_i16(a_.i16[i], b_.i16[i]); } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_adds_epi16(a, b) simde_mm_adds_epi16(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_adds_epu8 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_adds_epu8(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_u8 = vqaddq_u8(a_.neon_u8, b_.neon_u8); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_u8x16_add_sat(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) r_.altivec_u8 = vec_adds(a_.altivec_u8, b_.altivec_u8); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.u8) / sizeof(r_.u8[0])) ; i++) { r_.u8[i] = simde_math_adds_u8(a_.u8[i], b_.u8[i]); } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_adds_epu8(a, b) simde_mm_adds_epu8(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_adds_epu16 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_adds_epu16(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_u16 = vqaddq_u16(a_.neon_u16, b_.neon_u16); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_u16x8_add_sat(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) r_.altivec_u16 = vec_adds(a_.altivec_u16, b_.altivec_u16); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.u16) / sizeof(r_.u16[0])) ; i++) { r_.u16[i] = simde_math_adds_u16(a_.u16[i], b_.u16[i]); } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_adds_epu16(a, b) simde_mm_adds_epu16(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_and_pd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_and_pd(a, b); #else simde__m128d_private r_, a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i32 = vandq_s32(a_.neon_i32, b_.neon_i32); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_v128_and(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) r_.altivec_f64 = vec_and(a_.altivec_f64, b_.altivec_f64); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i32f = a_.i32f & b_.i32f; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])) ; i++) { r_.i32f[i] = a_.i32f[i] & b_.i32f[i]; } #endif return simde__m128d_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_and_pd(a, b) simde_mm_and_pd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_and_si128 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_and_si128(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i32 = vandq_s32(b_.neon_i32, a_.neon_i32); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) r_.altivec_u32f = vec_and(a_.altivec_u32f, b_.altivec_u32f); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_v128_and(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i32f = a_.i32f & b_.i32f; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])) ; i++) { r_.i32f[i] = a_.i32f[i] & b_.i32f[i]; } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_and_si128(a, b) simde_mm_and_si128(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_andnot_pd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_andnot_pd(a, b); #else simde__m128d_private r_, a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i32 = vbicq_s32(b_.neon_i32, a_.neon_i32); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_v128_andnot(b_.wasm_v128, a_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI r_.altivec_f64 = vec_andc(b_.altivec_f64, a_.altivec_f64); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) r_.altivec_i32f = vec_andc(b_.altivec_i32f, a_.altivec_i32f); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i32f = ~a_.i32f & b_.i32f; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.u64) / sizeof(r_.u64[0])) ; i++) { r_.u64[i] = ~a_.u64[i] & b_.u64[i]; } #endif return simde__m128d_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_andnot_pd(a, b) simde_mm_andnot_pd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_andnot_si128 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_andnot_si128(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i32 = vbicq_s32(b_.neon_i32, a_.neon_i32); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI r_.altivec_i32 = vec_andc(b_.altivec_i32, a_.altivec_i32); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_v128_andnot(b_.wasm_v128, a_.wasm_v128); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i32f = ~a_.i32f & b_.i32f; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])) ; i++) { r_.i32f[i] = ~(a_.i32f[i]) & b_.i32f[i]; } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_andnot_si128(a, b) simde_mm_andnot_si128(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_xor_pd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_xor_pd(a, b); #else simde__m128d_private r_, a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i32f = a_.i32f ^ b_.i32f; #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_v128_xor(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i64 = veorq_s64(a_.neon_i64, b_.neon_i64); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])) ; i++) { r_.i32f[i] = a_.i32f[i] ^ b_.i32f[i]; } #endif return simde__m128d_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_xor_pd(a, b) simde_mm_xor_pd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_avg_epu8 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_avg_epu8(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_u8 = vrhaddq_u8(b_.neon_u8, a_.neon_u8); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_u8x16_avgr(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI r_.altivec_u8 = vec_avg(a_.altivec_u8, b_.altivec_u8); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && de uint16_t wa SIMDE_VECTOR(32); uint16_t wb SIMDE_VECTOR(32); uint16_t wr SIMDE_VECTOR(32); SIMDE_CONVERT_VECTOR_(wa, a_.u8); SIMDE_CONVERT_VECTOR_(wb, b_.u8); wr = (wa + wb + 1) >> 1; SIMDE_CONVERT_VECTOR_(r_.u8, wr); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.u8) / sizeof(r_.u8[0])) ; i++) { r_.u8[i] = (a_.u8[i] + b_.u8[i] + 1) >> 1; } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_avg_epu8(a, b) simde_mm_avg_epu8(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_avg_epu16 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_avg_epu16(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_u16 = vrhaddq_u16(b_.neon_u16, a_.neon_u16); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_u16x8_avgr(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI r_.altivec_u16 = vec_avg(a_.altivec_u16, b_.altivec_u16); #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && de uint32_t wa SIMDE_VECTOR(32); uint32_t wb SIMDE_VECTOR(32); uint32_t wr SIMDE_VECTOR(32); SIMDE_CONVERT_VECTOR_(wa, a_.u16); SIMDE_CONVERT_VECTOR_(wb, b_.u16); wr = (wa + wb + 1) >> 1; SIMDE_CONVERT_VECTOR_(r_.u16, wr); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.u16) / sizeof(r_.u16[0])) ; i++) { r_.u16[i] = (a_.u16[i] + b_.u16[i] + 1) >> 1; } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_avg_epu16(a, b) simde_mm_avg_epu16(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_setzero_si128 (void) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_setzero_si128(); #else simde__m128i_private r_; #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_i32 = vdupq_n_s32(0); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI r_.altivec_i32 = vec_splats(HEDLEY_STATIC_CAST(signed int, 0)); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i32x4_splat(INT32_C(0)); #elif defined(SIMDE_VECTOR_SUBSCRIPT) r_.i32 = __extension__ (__typeof__(r_.i32)) { 0, 0, 0, 0 }; #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])) ; i++) { r_.i32f[i] = 0; } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_setzero_si128() (simde_mm_setzero_si128()) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_bslli_si128 (simde__m128i a, const int imm8) SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255) { simde__m128i_private r_, a_ = simde__m128i_to_private(a); if (HEDLEY_UNLIKELY((imm8 & ~15))) { return simde_mm_setzero_si128(); } #if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) && defined(SIMDE_ENDIAN_ORDER) r_.altivec_i8 = #if (SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_LITTLE) vec_slo #else /* SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_BIG */ vec_sro #endif (a_.altivec_i8, vec_splats(HEDLEY_STATIC_CAST(unsigned char, imm8 * 8))); #elif defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE) r_.altivec_i8 = vec_srb(a_.altivec_i8, vec_splats(HEDLEY_STATIC_CAST(unsigned char, ( #elif defined(SIMDE_HAVE_INT128_) && (SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_LITTLE) r_.u128[0] = a_.u128[0] << (imm8 * 8); #else r_ = simde__m128i_to_private(simde_mm_setzero_si128()); for (int i = imm8 ; i < HEDLEY_STATIC_CAST(int, sizeof(r_.i8) / sizeof(r_.i8[0])) ; i++) { r_.i8[i] = a_.i8[i - imm8]; } #endif return simde__m128i_from_private(r_); } #if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI) #define simde_mm_bslli_si128(a, imm8) _mm_slli_si128(a, imm8) #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) && !defined(__clang__) #define simde_mm_bslli_si128(a, imm8) \ simde__m128i_from_neon_i8(((imm8) <= 0) ? simde__m128i_to_neon_i8(a) : (((imm8) > 15) ? (vdu #elif defined(SIMDE_WASM_SIMD128_NATIVE) #define simde_mm_bslli_si128(a, imm8) __extension__ ({ \ simde__m128i_from_wasm_v128( \ wasm_i8x16_shuffle(wasm_i32x4_splat(INT32_C(0)), \ simde__m128i_to_wasm_v128((a)), \ ((imm8)&0xF0) ? 0 : 16 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 17 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 18 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 19 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 20 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 21 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 22 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 23 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 24 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 25 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 26 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 27 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 28 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 29 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 30 - ((imm8)&0xF), \ ((imm8)&0xF0) ? 0 : 31 - ((imm8)&0xF))); }) #elif defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) && !defin #define simde_mm_bslli_si128(a, imm8) (__extension__ ({ \ const simde__m128i_private simde_tmp_a_ = simde__m128i_to_private(a); \ const simde__m128i_private simde_tmp_z_ = simde__m128i_to_private(simde_mm_setzero_s simde__m128i_private simde_tmp_r_; \ if (HEDLEY_UNLIKELY(imm8 > 15)) { \ simde_tmp_r_ = simde__m128i_to_private(simde_mm_setzero_si128()); \ } else { \ simde_tmp_r_.i8 = \ SIMDE_SHUFFLE_VECTOR_(8, 16, \ simde_tmp_z_.i8, \ (simde_tmp_a_).i8, \ HEDLEY_STATIC_CAST(int8_t, (16 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (17 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (18 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (19 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (20 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (21 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (22 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (23 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (24 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (25 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (26 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (27 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (28 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (29 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (30 - imm8) & 31), \ HEDLEY_STATIC_CAST(int8_t, (31 - imm8) & 31)); \ } \ simde__m128i_from_private(simde_tmp_r_); })) #endif #define simde_mm_slli_si128(a, imm8) simde_mm_bslli_si128(a, imm8) #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_bslli_si128(a, imm8) simde_mm_bslli_si128(a, imm8) #define _mm_slli_si128(a, imm8) simde_mm_bslli_si128(a, imm8) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_bsrli_si128 (simde__m128i a, const int imm8) SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255) { simde__m128i_private r_, a_ = simde__m128i_to_private(a); if (HEDLEY_UNLIKELY((imm8 & ~15))) { return simde_mm_setzero_si128(); } #if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) && defined(SIMDE_ENDIAN_ORDER) r_.altivec_i8 = #if (SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_LITTLE) vec_sro #else /* SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_BIG */ vec_slo #endif (a_.altivec_i8, vec_splats(HEDLEY_STATIC_CAST(unsigned char, imm8 * 8))); #elif defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE) r_.altivec_i8 = vec_slb(a_.altivec_i8, vec_splats(HEDLEY_STATIC_CAST(unsigned char, ( #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i8) / sizeof(r_.i8[0])) ; i++) { const int e = HEDLEY_STATIC_CAST(int, i) + imm8; r_.i8[i] = (e < 16) ? a_.i8[e] : 0; } #endif return simde__m128i_from_private(r_); } #if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI) #define simde_mm_bsrli_si128(a, imm8) _mm_srli_si128(a, imm8) #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) && !defined(__clang__) #define simde_mm_bsrli_si128(a, imm8) \ simde__m128i_from_neon_i8(((imm8 < 0) || (imm8 > 15)) ? vdupq_n_s8(0) : (vextq_s8(simde__m12 #elif defined(SIMDE_WASM_SIMD128_NATIVE) #define simde_mm_bsrli_si128(a, imm8) (__extension__ ({ \ const simde__m128i_private simde_tmp_a_ = simde__m128i_to_private(a); \ const simde__m128i_private simde_tmp_z_ = simde__m128i_to_private(simde_mm_setzero_s simde__m128i_private simde_tmp_r_ = simde__m128i_to_private(a); \ if (HEDLEY_UNLIKELY(imm8 > 15)) { \ simde_tmp_r_ = simde__m128i_to_private(simde_mm_setzero_si128()); \ } else { \ simde_tmp_r_.wasm_v128 = \ wasm_i8x16_shuffle( \ simde_tmp_z_.wasm_v128, \ simde_tmp_a_.wasm_v128, \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 16) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 17) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 18) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 19) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 20) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 21) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 22) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 23) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 24) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 25) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 26) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 27) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 28) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 29) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 30) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 31) & 31)); \ } \ simde__m128i_from_private(simde_tmp_r_); })) #elif defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) && !defin #define simde_mm_bsrli_si128(a, imm8) (__extension__ ({ \ const simde__m128i_private simde_tmp_a_ = simde__m128i_to_private(a); \ const simde__m128i_private simde_tmp_z_ = simde__m128i_to_private(simde_mm_setzero_s simde__m128i_private simde_tmp_r_ = simde__m128i_to_private(a); \ if (HEDLEY_UNLIKELY(imm8 > 15)) { \ simde_tmp_r_ = simde__m128i_to_private(simde_mm_setzero_si128()); \ } else { \ simde_tmp_r_.i8 = \ SIMDE_SHUFFLE_VECTOR_(8, 16, \ simde_tmp_z_.i8, \ (simde_tmp_a_).i8, \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 16) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 17) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 18) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 19) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 20) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 21) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 22) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 23) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 24) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 25) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 26) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 27) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 28) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 29) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 30) & 31), \ HEDLEY_STATIC_CAST(int8_t, (imm8 + 31) & 31)); \ } \ simde__m128i_from_private(simde_tmp_r_); })) #endif #define simde_mm_srli_si128(a, imm8) simde_mm_bsrli_si128((a), (imm8)) #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_bsrli_si128(a, imm8) simde_mm_bsrli_si128((a), (imm8)) #define _mm_srli_si128(a, imm8) simde_mm_bsrli_si128((a), (imm8)) #endif SIMDE_FUNCTION_ATTRIBUTES void simde_mm_clflush (void const* p) { #if defined(SIMDE_X86_SSE2_NATIVE) _mm_clflush(p); #else (void) p; #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_clflush(p) simde_mm_clflush(p) #endif SIMDE_FUNCTION_ATTRIBUTES int simde_mm_comieq_sd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_comieq_sd(a, b); #else simde__m128d_private a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return !!vgetq_lane_u64(vceqq_f64(a_.neon_f64, b_.neon_f64), 0); #elif defined(SIMDE_WASM_SIMD128_NATIVE) return wasm_f64x2_extract_lane(a_.wasm_v128, 0) == wasm_f64x2_extract_lane(b_.wasm_v #else return a_.f64[0] == b_.f64[0]; #endif #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_comieq_sd(a, b) simde_mm_comieq_sd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES int simde_mm_comige_sd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_comige_sd(a, b); #else simde__m128d_private a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return !!vgetq_lane_u64(vcgeq_f64(a_.neon_f64, b_.neon_f64), 0); #elif defined(SIMDE_WASM_SIMD128_NATIVE) return wasm_f64x2_extract_lane(a_.wasm_v128, 0) >= wasm_f64x2_extract_lane(b_.wasm_v1 #else return a_.f64[0] >= b_.f64[0]; #endif #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_comige_sd(a, b) simde_mm_comige_sd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES int simde_mm_comigt_sd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_comigt_sd(a, b); #else simde__m128d_private a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return !!vgetq_lane_u64(vcgtq_f64(a_.neon_f64, b_.neon_f64), 0); #elif defined(SIMDE_WASM_SIMD128_NATIVE) return wasm_f64x2_extract_lane(a_.wasm_v128, 0) > wasm_f64x2_extract_lane(b_.wasm_v12 #else return a_.f64[0] > b_.f64[0]; #endif #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_comigt_sd(a, b) simde_mm_comigt_sd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES int simde_mm_comile_sd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_comile_sd(a, b); #else simde__m128d_private a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return !!vgetq_lane_u64(vcleq_f64(a_.neon_f64, b_.neon_f64), 0); #elif defined(SIMDE_WASM_SIMD128_NATIVE) return wasm_f64x2_extract_lane(a_.wasm_v128, 0) <= wasm_f64x2_extract_lane(b_.wasm_v1 #else return a_.f64[0] <= b_.f64[0]; #endif #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_comile_sd(a, b) simde_mm_comile_sd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES int simde_mm_comilt_sd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_comilt_sd(a, b); #else simde__m128d_private a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return !!vgetq_lane_u64(vcltq_f64(a_.neon_f64, b_.neon_f64), 0); #elif defined(SIMDE_WASM_SIMD128_NATIVE) return wasm_f64x2_extract_lane(a_.wasm_v128, 0) < wasm_f64x2_extract_lane(b_.wasm_v12 #else return a_.f64[0] < b_.f64[0]; #endif #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_comilt_sd(a, b) simde_mm_comilt_sd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES int simde_mm_comineq_sd (simde__m128d a, simde__m128d b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_comineq_sd(a, b); #else simde__m128d_private a_ = simde__m128d_to_private(a), b_ = simde__m128d_to_private(b); #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) return !vgetq_lane_u64(vceqq_f64(a_.neon_f64, b_.neon_f64), 0); #elif defined(SIMDE_WASM_SIMD128_NATIVE) return wasm_f64x2_extract_lane(a_.wasm_v128, 0) != wasm_f64x2_extract_lane(b_.wasm_v #else return a_.f64[0] != b_.f64[0]; #endif #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_comineq_sd(a, b) simde_mm_comineq_sd(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_x_mm_copysign_pd(simde__m128d dest, simde__m128d src) { simde__m128d_private r_, dest_ = simde__m128d_to_private(dest), src_ = simde__m128d_to_private(src); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) #if defined(SIMDE_ARM_NEON_A64V8_NATIVE) uint64x2_t sign_pos = vreinterpretq_u64_f64(vdupq_n_f64(-SIMDE_FLOAT64_C(0.0))); #else simde_float64 dbl_nz = -SIMDE_FLOAT64_C(0.0); uint64_t u64_nz; simde_memcpy(&u64_nz, &dbl_nz, sizeof(u64_nz)); uint64x2_t sign_pos = vdupq_n_u64(u64_nz); #endif r_.neon_u64 = vbslq_u64(sign_pos, src_.neon_u64, dest_.neon_u64); #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) #if defined(SIMDE_BUG_VEC_CPSGN_REVERSED_ARGS) r_.altivec_f64 = vec_cpsgn(dest_.altivec_f64, src_.altivec_f64); #else r_.altivec_f64 = vec_cpsgn(src_.altivec_f64, dest_.altivec_f64); #endif #elif defined(simde_math_copysign) SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.f64) / sizeof(r_.f64[0])) ; i++) { r_.f64[i] = simde_math_copysign(dest_.f64[i], src_.f64[i]); } #else simde__m128d sgnbit = simde_mm_set1_pd(-SIMDE_FLOAT64_C(0.0)); return simde_mm_xor_pd(simde_mm_and_pd(sgnbit, src), simde_mm_andnot_pd(sgnbit, dest #endif return simde__m128d_from_private(r_); } SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_x_mm_xorsign_pd(simde__m128d dest, simde__m128d src) { return simde_mm_xor_pd(simde_mm_and_pd(simde_mm_set1_pd(-0.0), src), dest); } SIMDE_FUNCTION_ATTRIBUTES simde__m128 simde_mm_castpd_ps (simde__m128d a) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_castpd_ps(a); #elif defined(SIMDE_ARM_NEON_A64V8_NATIVE) return vreinterpretq_f32_f64(a); #else simde__m128 r; simde_memcpy(&r, &a, sizeof(a)); return r; #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_castpd_ps(a) simde_mm_castpd_ps(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_castpd_si128 (simde__m128d a) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_castpd_si128(a); #elif defined(SIMDE_ARM_NEON_A64V8_NATIVE) return vreinterpretq_s64_f64(a); #else simde__m128i r; simde_memcpy(&r, &a, sizeof(a)); return r; #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_castpd_si128(a) simde_mm_castpd_si128(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_castps_pd (simde__m128 a) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_castps_pd(a); #elif defined(SIMDE_ARM_NEON_A64V8_NATIVE) return vreinterpretq_f64_f32(a); #else simde__m128d r; simde_memcpy(&r, &a, sizeof(a)); return r; #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_castps_pd(a) simde_mm_castps_pd(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_castps_si128 (simde__m128 a) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_castps_si128(a); #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) return simde__m128i_from_neon_i32(simde__m128_to_private(a).neon_i32); #else simde__m128i r; simde_memcpy(&r, &a, sizeof(a)); return r; #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_castps_si128(a) simde_mm_castps_si128(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128d simde_mm_castsi128_pd (simde__m128i a) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_castsi128_pd(a); #elif defined(SIMDE_ARM_NEON_A64V8_NATIVE) return vreinterpretq_f64_s64(a); #else simde__m128d r; simde_memcpy(&r, &a, sizeof(a)); return r; #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_castsi128_pd(a) simde_mm_castsi128_pd(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128 simde_mm_castsi128_ps (simde__m128i a) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_castsi128_ps(a); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) return HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), a); #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) return simde__m128_from_neon_i32(simde__m128i_to_private(a).neon_i32); #else simde__m128 r; simde_memcpy(&r, &a, sizeof(a)); return r; #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_castsi128_ps(a) simde_mm_castsi128_ps(a) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_cmpeq_epi8 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_cmpeq_epi8(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_u8 = vceqq_s8(b_.neon_i8, a_.neon_i8); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i8x16_eq(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI r_.altivec_i8 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(signed char), vec #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i8 = HEDLEY_REINTERPRET_CAST(__typeof__(r_.i8), (a_.i8 == b_.i8)); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i8) / sizeof(r_.i8[0])) ; i++) { r_.i8[i] = (a_.i8[i] == b_.i8[i]) ? ~INT8_C(0) : INT8_C(0); } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_cmpeq_epi8(a, b) simde_mm_cmpeq_epi8(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_cmpeq_epi16 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_cmpeq_epi16(a, b); #else simde__m128i_private r_, a_ = simde__m128i_to_private(a), b_ = simde__m128i_to_private(b); #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) r_.neon_u16 = vceqq_s16(b_.neon_i16, a_.neon_i16); #elif defined(SIMDE_WASM_SIMD128_NATIVE) r_.wasm_v128 = wasm_i16x8_eq(a_.wasm_v128, b_.wasm_v128); #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATI r_.altivec_i16 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(signed short), v #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) r_.i16 = (a_.i16 == b_.i16); #else SIMDE_VECTORIZE for (size_t i = 0 ; i < (sizeof(r_.i16) / sizeof(r_.i16[0])) ; i++) { r_.i16[i] = (a_.i16[i] == b_.i16[i]) ? ~INT16_C(0) : INT16_C(0); } #endif return simde__m128i_from_private(r_); #endif } #if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES) #define _mm_cmpeq_epi16(a, b) simde_mm_cmpeq_epi16(a, b) #endif SIMDE_FUNCTION_ATTRIBUTES simde__m128i simde_mm_cmpeq_epi32 (simde__m128i a, simde__m128i b) { #if defined(SIMDE_X86_SSE2_NATIVE) return _mm_cmpeq_epi32(a, b); #else --> -------------------- --> maximum size reached --> --------------------
¤ Dauer der Verarbeitung: 0.38 Sekunden (vorverarbeitet) ¤*© Formatika GbR, Deutschland |
|
||||||||||||||
2026-04-02