SSL sse.h

/* 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>
 */

#if !defined(SIMDE_X86_SSE_H)
#define SIMDE_X86_SSE_H

#include "mmx.h"

#if defined(_WIN32) && !defined(SIMDE_X86_SSE_NATIVE) && defined(_MSC_VER)
  #define NOMINMAX
  #include <windows.h>
#endif

#if defined(__ARM_ACLE)
  #include <arm_acle.h>
#endif

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 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 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_SSE_NATIVE)
    SIMDE_ALIGN_TO_16 __m128         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_NATIVE)
    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;
    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;
    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(unsigned long long) altivec_u64;
      SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed long long)   altivec_i64;
      SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(double)             altivec_f64;
    #endif
  #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
    v16i8 lsx_i8;
    v8i16 lsx_i16;
    v4i32 lsx_i32;
    v2i64 lsx_i64;
    v16u8 lsx_u8;
    v8u16 lsx_u16;
    v4u32 lsx_u32;
    v2u64 lsx_u64;
    v4f32 lsx_f32;
    v2f64 lsx_f64;
  #endif
} simde__m128_private;

#if defined(SIMDE_X86_SSE_NATIVE)
  typedef __m128 simde__m128;
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
   typedef float32x4_t simde__m128;
#elif defined(SIMDE_WASM_SIMD128_NATIVE)
   typedef v128_t simde__m128;
#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE)
   typedef SIMDE_POWER_ALTIVEC_VECTOR(float) simde__m128;
#elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
  typedef v4f32 simde__m128;
#elif defined(SIMDE_VECTOR_SUBSCRIPT)
  typedef simde_float32 simde__m128 SIMDE_ALIGN_TO_16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
#else
  typedef simde__m128_private simde__m128;
#endif

#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  typedef simde__m128 __m128;
#endif

HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128), "simde__m128 size incorrect");
HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128_private), "simde__m128_private size incorrect");
#if defined(SIMDE_CHECK_ALIGNMENT) && defined(SIMDE_ALIGN_OF)
HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128) == 16, "simde__m128 is not 16-byte aligned");
HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128_private) == 16, "simde__m128_private is not 16-byte aligned");
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde__m128_from_private(simde__m128_private v) {
  simde__m128 r;
  simde_memcpy(&r, &v, sizeof(r));
  return r;
}

SIMDE_FUNCTION_ATTRIBUTES
simde__m128_private
simde__m128_to_private(simde__m128 v) {
  simde__m128_private r;
  simde_memcpy(&r, &v, sizeof(r));
  return r;
}

#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int8x16_t, neon, i8)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int16x8_t, neon, i16)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int32x4_t, neon, i32)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int64x2_t, neon, i64)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint8x16_t, neon, u8)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint16x8_t, neon, u16)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint32x4_t, neon, u32)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint64x2_t, neon, u64)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, float32x4_t, neon, f32)
  #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
    SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, float64x2_t, neon, f64)
  #endif
#endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */

#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(signed char), altivec, i8)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(signed short), altivec, i16)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(signed int), altivec, i32)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(unsigned char), altivec, u8)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(unsigned short), altivec, u16)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(unsigned int), altivec, u32)

  #if defined(SIMDE_BUG_GCC_95782)
    SIMDE_FUNCTION_ATTRIBUTES
    SIMDE_POWER_ALTIVEC_VECTOR(float)
    simde__m128_to_altivec_f32(simde__m128 value) {
      simde__m128_private r_ = simde__m128_to_private(value);
      return r_.altivec_f32;
    }

    SIMDE_FUNCTION_ATTRIBUTES
    simde__m128
    simde__m128_from_altivec_f32(SIMDE_POWER_ALTIVEC_VECTOR(float) value) {
      simde__m128_private r_;
      r_.altivec_f32 = value;
      return simde__m128_from_private(r_);
    }
  #else
    SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(float), altivec, f32)
  #endif

  #if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE)
    SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(signed long long), altivec, i64)
    SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long), altivec, u64)
  #endif
#elif defined(SIMDE_WASM_SIMD128_NATIVE)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v128_t, wasm, v128);
#endif /* defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) */

#if defined(SIMDE_LOONGARCH_LSX_NATIVE)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v16i8, lsx, i8)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v8i16, lsx, i16)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v4i32, lsx, i32)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v2i64, lsx, i64)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v16u8, lsx, u8)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v8u16, lsx, u16)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v4u32, lsx, u32)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v2u64, lsx, u64)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v4f32, lsx, f32)
  SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v2f64, lsx, f64)
#endif /* defined(SIMDE_LOONGARCH_LSX_NATIVE) */

enum {
  #if defined(SIMDE_X86_SSE_NATIVE)
    SIMDE_MM_ROUND_NEAREST     = _MM_ROUND_NEAREST,
    SIMDE_MM_ROUND_DOWN        = _MM_ROUND_DOWN,
    SIMDE_MM_ROUND_UP          = _MM_ROUND_UP,
    SIMDE_MM_ROUND_TOWARD_ZERO = _MM_ROUND_TOWARD_ZERO
  #else
    SIMDE_MM_ROUND_NEAREST     = 0x0000,
    SIMDE_MM_ROUND_DOWN        = 0x2000,
    SIMDE_MM_ROUND_UP          = 0x4000,
    SIMDE_MM_ROUND_TOWARD_ZERO = 0x6000
  #endif
};

#if defined(_MM_FROUND_TO_NEAREST_INT)
#  define SIMDE_MM_FROUND_TO_NEAREST_INT _MM_FROUND_TO_NEAREST_INT
#  define SIMDE_MM_FROUND_TO_NEG_INF     _MM_FROUND_TO_NEG_INF
#  define SIMDE_MM_FROUND_TO_POS_INF     _MM_FROUND_TO_POS_INF
#  define SIMDE_MM_FROUND_TO_ZERO        _MM_FROUND_TO_ZERO
#  define SIMDE_MM_FROUND_CUR_DIRECTION  _MM_FROUND_CUR_DIRECTION

#  define SIMDE_MM_FROUND_RAISE_EXC      _MM_FROUND_RAISE_EXC
#  define SIMDE_MM_FROUND_NO_EXC         _MM_FROUND_NO_EXC
#else
#  define SIMDE_MM_FROUND_TO_NEAREST_INT 0x00
#  define SIMDE_MM_FROUND_TO_NEG_INF     0x01
#  define SIMDE_MM_FROUND_TO_POS_INF     0x02
#  define SIMDE_MM_FROUND_TO_ZERO        0x03
#  define SIMDE_MM_FROUND_CUR_DIRECTION  0x04

#  define SIMDE_MM_FROUND_RAISE_EXC      0x00
#  define SIMDE_MM_FROUND_NO_EXC         0x08
#endif

#define SIMDE_MM_FROUND_NINT \
  (SIMDE_MM_FROUND_TO_NEAREST_INT | SIMDE_MM_FROUND_RAISE_EXC)
#define SIMDE_MM_FROUND_FLOOR \
  (SIMDE_MM_FROUND_TO_NEG_INF | SIMDE_MM_FROUND_RAISE_EXC)
#define SIMDE_MM_FROUND_CEIL \
  (SIMDE_MM_FROUND_TO_POS_INF | SIMDE_MM_FROUND_RAISE_EXC)
#define SIMDE_MM_FROUND_TRUNC \
  (SIMDE_MM_FROUND_TO_ZERO | SIMDE_MM_FROUND_RAISE_EXC)
#define SIMDE_MM_FROUND_RINT \
  (SIMDE_MM_FROUND_CUR_DIRECTION | SIMDE_MM_FROUND_RAISE_EXC)
#define SIMDE_MM_FROUND_NEARBYINT \
  (SIMDE_MM_FROUND_CUR_DIRECTION | SIMDE_MM_FROUND_NO_EXC)

#if defined(SIMDE_X86_SSE4_1_ENABLE_NATIVE_ALIASES) && !defined(_MM_FROUND_TO_NEAREST_INT)
#  define _MM_FROUND_TO_NEAREST_INT SIMDE_MM_FROUND_TO_NEAREST_INT
#  define _MM_FROUND_TO_NEG_INF SIMDE_MM_FROUND_TO_NEG_INF
#  define _MM_FROUND_TO_POS_INF SIMDE_MM_FROUND_TO_POS_INF
#  define _MM_FROUND_TO_ZERO SIMDE_MM_FROUND_TO_ZERO
#  define _MM_FROUND_CUR_DIRECTION SIMDE_MM_FROUND_CUR_DIRECTION
#  define _MM_FROUND_RAISE_EXC SIMDE_MM_FROUND_RAISE_EXC
#  define _MM_FROUND_NINT SIMDE_MM_FROUND_NINT
#  define _MM_FROUND_FLOOR SIMDE_MM_FROUND_FLOOR
#  define _MM_FROUND_CEIL SIMDE_MM_FROUND_CEIL
#  define _MM_FROUND_TRUNC SIMDE_MM_FROUND_TRUNC
#  define _MM_FROUND_RINT SIMDE_MM_FROUND_RINT
#  define _MM_FROUND_NEARBYINT SIMDE_MM_FROUND_NEARBYINT
#endif

#if defined(_MM_EXCEPT_INVALID)
#  define SIMDE_MM_EXCEPT_INVALID _MM_EXCEPT_INVALID
#else
#  define SIMDE_MM_EXCEPT_INVALID (0x0001)
#endif
#if defined(_MM_EXCEPT_DENORM)
#  define SIMDE_MM_EXCEPT_DENORM _MM_EXCEPT_DENORM
#else
#  define SIMDE_MM_EXCEPT_DENORM (0x0002)
#endif
#if defined(_MM_EXCEPT_DIV_ZERO)
#  define SIMDE_MM_EXCEPT_DIV_ZERO _MM_EXCEPT_DIV_ZERO
#else
#  define SIMDE_MM_EXCEPT_DIV_ZERO (0x0004)
#endif
#if defined(_MM_EXCEPT_OVERFLOW)
#  define SIMDE_MM_EXCEPT_OVERFLOW _MM_EXCEPT_OVERFLOW
#else
#  define SIMDE_MM_EXCEPT_OVERFLOW (0x0008)
#endif
#if defined(_MM_EXCEPT_UNDERFLOW)
#  define SIMDE_MM_EXCEPT_UNDERFLOW _MM_EXCEPT_UNDERFLOW
#else
#  define SIMDE_MM_EXCEPT_UNDERFLOW (0x0010)
#endif
#if defined(_MM_EXCEPT_INEXACT)
#  define SIMDE_MM_EXCEPT_INEXACT _MM_EXCEPT_INEXACT
#else
#  define SIMDE_MM_EXCEPT_INEXACT (0x0020)
#endif
#if defined(_MM_EXCEPT_MASK)
#  define SIMDE_MM_EXCEPT_MASK _MM_EXCEPT_MASK
#else
#  define SIMDE_MM_EXCEPT_MASK \
     (SIMDE_MM_EXCEPT_INVALID | SIMDE_MM_EXCEPT_DENORM | \
      SIMDE_MM_EXCEPT_DIV_ZERO | SIMDE_MM_EXCEPT_OVERFLOW | \
      SIMDE_MM_EXCEPT_UNDERFLOW | SIMDE_MM_EXCEPT_INEXACT)
#endif
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _MM_EXCEPT_INVALID SIMDE_MM_EXCEPT_INVALID
  #define _MM_EXCEPT_DENORM SIMDE_MM_EXCEPT_DENORM
  #define _MM_EXCEPT_DIV_ZERO SIMDE_MM_EXCEPT_DIV_ZERO
  #define _MM_EXCEPT_OVERFLOW SIMDE_MM_EXCEPT_OVERFLOW
  #define _MM_EXCEPT_UNDERFLOW SIMDE_MM_EXCEPT_UNDERFLOW
  #define _MM_EXCEPT_INEXACT SIMDE_MM_EXCEPT_INEXACT
  #define _MM_EXCEPT_MASK SIMDE_MM_EXCEPT_MASK
#endif

#if defined(_MM_MASK_INVALID)
#  define SIMDE_MM_MASK_INVALID _MM_MASK_INVALID
#else
#  define SIMDE_MM_MASK_INVALID (0x0080)
#endif
#if defined(_MM_MASK_DENORM)
#  define SIMDE_MM_MASK_DENORM _MM_MASK_DENORM
#else
#  define SIMDE_MM_MASK_DENORM (0x0100)
#endif
#if defined(_MM_MASK_DIV_ZERO)
#  define SIMDE_MM_MASK_DIV_ZERO _MM_MASK_DIV_ZERO
#else
#  define SIMDE_MM_MASK_DIV_ZERO (0x0200)
#endif
#if defined(_MM_MASK_OVERFLOW)
#  define SIMDE_MM_MASK_OVERFLOW _MM_MASK_OVERFLOW
#else
#  define SIMDE_MM_MASK_OVERFLOW (0x0400)
#endif
#if defined(_MM_MASK_UNDERFLOW)
#  define SIMDE_MM_MASK_UNDERFLOW _MM_MASK_UNDERFLOW
#else
#  define SIMDE_MM_MASK_UNDERFLOW (0x0800)
#endif
#if defined(_MM_MASK_INEXACT)
#  define SIMDE_MM_MASK_INEXACT _MM_MASK_INEXACT
#else
#  define SIMDE_MM_MASK_INEXACT (0x1000)
#endif
#if defined(_MM_MASK_MASK)
#  define SIMDE_MM_MASK_MASK _MM_MASK_MASK
#else
#  define SIMDE_MM_MASK_MASK \
     (SIMDE_MM_MASK_INVALID | SIMDE_MM_MASK_DENORM | \
      SIMDE_MM_MASK_DIV_ZERO | SIMDE_MM_MASK_OVERFLOW | \
      SIMDE_MM_MASK_UNDERFLOW | SIMDE_MM_MASK_INEXACT)
#endif
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _MM_MASK_INVALID SIMDE_MM_MASK_INVALID
  #define _MM_MASK_DENORM SIMDE_MM_MASK_DENORM
  #define _MM_MASK_DIV_ZERO SIMDE_MM_MASK_DIV_ZERO
  #define _MM_MASK_OVERFLOW SIMDE_MM_MASK_OVERFLOW
  #define _MM_MASK_UNDERFLOW SIMDE_MM_MASK_UNDERFLOW
  #define _MM_MASK_INEXACT SIMDE_MM_MASK_INEXACT
  #define _MM_MASK_MASK SIMDE_MM_MASK_MASK
#endif

#if defined(_MM_FLUSH_ZERO_MASK)
#  define SIMDE_MM_FLUSH_ZERO_MASK _MM_FLUSH_ZERO_MASK
#else
#  define SIMDE_MM_FLUSH_ZERO_MASK (0x8000)
#endif
#if defined(_MM_FLUSH_ZERO_ON)
#  define SIMDE_MM_FLUSH_ZERO_ON _MM_FLUSH_ZERO_ON
#else
#  define SIMDE_MM_FLUSH_ZERO_ON (0x8000)
#endif
#if defined(_MM_FLUSH_ZERO_OFF)
#  define SIMDE_MM_FLUSH_ZERO_OFF _MM_FLUSH_ZERO_OFF
#else
#  define SIMDE_MM_FLUSH_ZERO_OFF (0x0000)
#endif
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _MM_FLUSH_ZERO_MASK SIMDE_MM_FLUSH_ZERO_MASK
  #define _MM_FLUSH_ZERO_ON SIMDE_MM_FLUSH_ZERO_ON
  #define _MM_FLUSH_ZERO_OFF SIMDE_MM_FLUSH_ZERO_OFF
#endif

SIMDE_FUNCTION_ATTRIBUTES
unsigned int
SIMDE_MM_GET_ROUNDING_MODE(void) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _MM_GET_ROUNDING_MODE();
  #elif defined(SIMDE_HAVE_FENV_H)
    unsigned int vfe_mode;

    switch (fegetround()) {
      #if defined(FE_TONEAREST)
        case FE_TONEAREST:
          vfe_mode = SIMDE_MM_ROUND_NEAREST;
          break;
      #endif

      #if defined(FE_TOWARDZERO)
        case FE_TOWARDZERO:
          vfe_mode = SIMDE_MM_ROUND_DOWN;
          break;
      #endif

      #if defined(FE_UPWARD)
        case FE_UPWARD:
          vfe_mode = SIMDE_MM_ROUND_UP;
          break;
      #endif

      #if defined(FE_DOWNWARD)
        case FE_DOWNWARD:
          vfe_mode = SIMDE_MM_ROUND_TOWARD_ZERO;
          break;
      #endif

      default:
        vfe_mode = SIMDE_MM_ROUND_NEAREST;
        break;
    }

    return vfe_mode;
  #else
    return SIMDE_MM_ROUND_NEAREST;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _MM_GET_ROUNDING_MODE() SIMDE_MM_GET_ROUNDING_MODE()
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
SIMDE_MM_SET_ROUNDING_MODE(unsigned int a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _MM_SET_ROUNDING_MODE(a);
  #elif defined(SIMDE_HAVE_FENV_H)
    int fe_mode = FE_TONEAREST;

    switch (a) {
      #if defined(FE_TONEAREST)
        case SIMDE_MM_ROUND_NEAREST:
          fe_mode = FE_TONEAREST;
          break;
      #endif

      #if defined(FE_TOWARDZERO)
        case SIMDE_MM_ROUND_TOWARD_ZERO:
          fe_mode = FE_TOWARDZERO;
          break;
      #endif

      #if defined(FE_DOWNWARD)
        case SIMDE_MM_ROUND_DOWN:
          fe_mode = FE_DOWNWARD;
          break;
      #endif

      #if defined(FE_UPWARD)
        case SIMDE_MM_ROUND_UP:
          fe_mode = FE_UPWARD;
          break;
      #endif

      default:
        return;
    }

    fesetround(fe_mode);
  #else
    (void) a;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _MM_SET_ROUNDING_MODE(a) SIMDE_MM_SET_ROUNDING_MODE(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
uint32_t
SIMDE_MM_GET_FLUSH_ZERO_MODE (void) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_getcsr() & _MM_FLUSH_ZERO_MASK;
  #else
    return SIMDE_MM_FLUSH_ZERO_OFF;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _MM_SET_FLUSH_ZERO_MODE(a) SIMDE_MM_SET_FLUSH_ZERO_MODE(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
SIMDE_MM_SET_FLUSH_ZERO_MODE (uint32_t a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _MM_SET_FLUSH_ZERO_MODE(a);
  #else
    (void) a;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _MM_SET_FLUSH_ZERO_MODE(a) SIMDE_MM_SET_FLUSH_ZERO_MODE(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
uint32_t
simde_mm_getcsr (void) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_getcsr();
  #else
    return SIMDE_MM_GET_ROUNDING_MODE();
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _mm_getcsr() simde_mm_getcsr()
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_setcsr (uint32_t a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_setcsr(a);
  #else
    SIMDE_MM_SET_ROUNDING_MODE(HEDLEY_STATIC_CAST(unsigned int, a));
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _mm_setcsr(a) simde_mm_setcsr(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_round_ps (simde__m128 a, int rounding, int lax_rounding)
    SIMDE_REQUIRE_CONSTANT_RANGE(rounding, 0, 15)
    SIMDE_REQUIRE_CONSTANT_RANGE(lax_rounding, 0, 1) {
  simde__m128_private
    r_,
    a_ = simde__m128_to_private(a);

  (void) lax_rounding;

  /* For architectures which lack a current direction SIMD instruction.
   *
   * Note that NEON actually has a current rounding mode instruction,
   * but in ARMv8+ the rounding mode is ignored and nearest is always
   * used, so we treat ARMv7 as having a rounding mode but ARMv8 as
   * not. */
  #if \
      defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || \
      defined(SIMDE_ARM_NEON_A32V8)
    if ((rounding & 7) == SIMDE_MM_FROUND_CUR_DIRECTION)
      rounding = HEDLEY_STATIC_CAST(int, SIMDE_MM_GET_ROUNDING_MODE()) << 13;
  #endif

  switch (rounding & ~SIMDE_MM_FROUND_NO_EXC) {
    case SIMDE_MM_FROUND_CUR_DIRECTION:
      #if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
        r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_round(a_.altivec_f32));
      #elif defined(SIMDE_ARM_NEON_A32V8_NATIVE) && !defined(SIMDE_BUG_GCC_95399)
        r_.neon_f32 = vrndiq_f32(a_.neon_f32);
      #elif defined(SIMDE_WASM_SIMD128_NATIVE)
        r_.wasm_v128 = wasm_f32x4_nearest(a_.wasm_v128);
      #elif defined(simde_math_nearbyintf)
        SIMDE_VECTORIZE
        for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
          r_.f32[i] = simde_math_nearbyintf(a_.f32[i]);
        }
      #else
        HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
      #endif
      break;

    case SIMDE_MM_FROUND_TO_NEAREST_INT:
      #if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
        r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_rint(a_.altivec_f32));
      #elif defined(SIMDE_ARM_NEON_A32V8_NATIVE)
        r_.neon_f32 = vrndnq_f32(a_.neon_f32);
      #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
        r_.lsx_i64 = __lsx_vfrintrne_s(a_.lsx_f32);
      #elif defined(SIMDE_WASM_SIMD128_NATIVE)
        r_.wasm_v128 = wasm_f32x4_nearest(a_.wasm_v128);
      #elif defined(simde_math_roundevenf)
        SIMDE_VECTORIZE
        for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
          r_.f32[i] = simde_math_roundevenf(a_.f32[i]);
        }
      #else
        HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
      #endif
      break;

    case SIMDE_MM_FROUND_TO_NEG_INF:
      #if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
        r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_floor(a_.altivec_f32));
      #elif defined(SIMDE_ARM_NEON_A32V8_NATIVE)
        r_.neon_f32 = vrndmq_f32(a_.neon_f32);
      #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
        r_.lsx_i64 = __lsx_vfrintrm_s(a_.lsx_f32);
      #elif defined(SIMDE_WASM_SIMD128_NATIVE)
        r_.wasm_v128 = wasm_f32x4_floor(a_.wasm_v128);
      #elif defined(simde_math_floorf)
        SIMDE_VECTORIZE
        for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
          r_.f32[i] = simde_math_floorf(a_.f32[i]);
        }
      #else
        HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
      #endif
      break;

    case SIMDE_MM_FROUND_TO_POS_INF:
      #if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
        r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_ceil(a_.altivec_f32));
      #elif defined(SIMDE_ARM_NEON_A32V8_NATIVE)
        r_.neon_f32 = vrndpq_f32(a_.neon_f32);
      #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
        r_.lsx_i64 = __lsx_vfrintrp_s(a_.lsx_f32);
      #elif defined(SIMDE_WASM_SIMD128_NATIVE)
        r_.wasm_v128 = wasm_f32x4_ceil(a_.wasm_v128);
      #elif defined(simde_math_ceilf)
        SIMDE_VECTORIZE
        for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
          r_.f32[i] = simde_math_ceilf(a_.f32[i]);
        }
      #else
        HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
      #endif
      break;

    case SIMDE_MM_FROUND_TO_ZERO:
      #if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
        r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_trunc(a_.altivec_f32));
      #elif defined(SIMDE_ARM_NEON_A32V8_NATIVE)
        r_.neon_f32 = vrndq_f32(a_.neon_f32);
      #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
        r_.lsx_i64 = __lsx_vfrintrz_s(a_.lsx_f32);
      #elif defined(SIMDE_WASM_SIMD128_NATIVE)
        r_.wasm_v128 = wasm_f32x4_trunc(a_.wasm_v128);
      #elif defined(simde_math_truncf)
        SIMDE_VECTORIZE
        for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
          r_.f32[i] = simde_math_truncf(a_.f32[i]);
        }
      #else
        HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
      #endif
      break;

    default:
      HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
  }

  return simde__m128_from_private(r_);
}
#if defined(SIMDE_X86_SSE4_1_NATIVE)
  #define simde_mm_round_ps(a, rounding) _mm_round_ps((a), (rounding))
#else
  #define simde_mm_round_ps(a, rounding) simde_x_mm_round_ps((a), (rounding), 0)
#endif
#if defined(SIMDE_X86_SSE4_1_ENABLE_NATIVE_ALIASES)
  #define _mm_round_ps(a, rounding) simde_mm_round_ps((a), (rounding))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_set_ps (simde_float32 e3, simde_float32 e2, simde_float32 e1, simde_float32 e0) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_set_ps(e3, e2, e1, e0);
  #else
    simde__m128_private r_;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      SIMDE_ALIGN_TO_16 simde_float32 data[4] = { e0, e1, e2, e3 };
      r_.neon_f32 = vld1q_f32(data);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_make(e0, e1, e2, e3);
    #else
      r_.f32[0] = e0;
      r_.f32[1] = e1;
      r_.f32[2] = e2;
      r_.f32[3] = e3;
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_set_ps(e3, e2, e1, e0) simde_mm_set_ps(e3, e2, e1, e0)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_set_ps1 (simde_float32 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_set_ps1(a);
  #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
    return vdupq_n_f32(a);
  #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
    (void) a;
    return vec_splats(a);
  #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
    return (simde__m128)__lsx_vldrepl_w(&a, 0);
  #elif defined(SIMDE_WASM_SIMD128_NATIVE)
    return wasm_f32x4_splat(a);
  #else
    return simde_mm_set_ps(a, a, a, a);
  #endif
}
#define simde_mm_set1_ps(a) simde_mm_set_ps1(a)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_set_ps1(a) simde_mm_set_ps1(a)
#  define _mm_set1_ps(a) simde_mm_set1_ps(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_move_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_move_ss(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vsetq_lane_f32(vgetq_lane_f32(b_.neon_f32, 0), a_.neon_f32, 0);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      static const SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) m = { ~0U, 0U, 0U, 0U };
      r_.altivec_f32 = vec_sel(a_.altivec_f32, b_.altivec_f32, m);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_i8x16_shuffle(b_.wasm_v128, a_.wasm_v128, 0, 1, 2, 3, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vextrins_w(a_.lsx_i64, b_.lsx_i64, 0);
    #elif defined(SIMDE_SHUFFLE_VECTOR_)
      r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 4, 1, 2, 3);
    #else
      r_.f32[0] = b_.f32[0];
      r_.f32[1] = a_.f32[1];
      r_.f32[2] = a_.f32[2];
      r_.f32[3] = a_.f32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_move_ss(a, b) simde_mm_move_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_broadcastlow_ps(simde__m128 a) {
  /* This function broadcasts the first element in the inpu vector to
   * all lanes.  It is used to avoid generating spurious exceptions in
   * *_ss functions since there may be garbage in the upper lanes. */

  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_shuffle_ps(a, a, 0);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
      r_.neon_f32 = vdupq_laneq_f32(a_.neon_f32, 0);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = vec_splat(a_.altivec_f32, 0);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vreplvei_w(a_.lsx_i64, 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_splat(a_.f32[0]);
    #elif defined(SIMDE_SHUFFLE_VECTOR_)
      r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, a_.f32, 0, 0, 0, 0);
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = a_.f32[0];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_add_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_add_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vaddq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_add(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = vec_add(a_.altivec_f32, b_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_f32 = __lsx_vfadd_s(a_.lsx_f32, b_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.f32 = a_.f32 + b_.f32;
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = a_.f32[i] + b_.f32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_add_ps(a, b) simde_mm_add_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_add_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_add_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_add_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_add_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32_t b0 = vgetq_lane_f32(b_.neon_f32, 0);
      float32x4_t value = vsetq_lane_f32(b0, vdupq_n_f32(0), 0);
      // the upper values in the result must be the remnants of <a>.
      r_.neon_f32 = vaddq_f32(a_.neon_f32, value);
    #else
      r_.f32[0] = a_.f32[0] + b_.f32[0];
      r_.f32[1] = a_.f32[1];
      r_.f32[2] = a_.f32[2];
      r_.f32[3] = a_.f32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_add_ss(a, b) simde_mm_add_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_and_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_and_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_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_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vand_v(a_.lsx_i64, b_.lsx_i64);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32 = a_.i32 & b_.i32;
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = vec_and(a_.altivec_f32, b_.altivec_f32);
    #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__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_and_ps(a, b) simde_mm_and_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_andnot_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_andnot_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_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_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
      r_.altivec_f32 = vec_andc(b_.altivec_f32, a_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vandn_v(a_.lsx_i64, b_.lsx_i64);
    #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__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_andnot_ps(a, b) simde_mm_andnot_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_xor_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_xor_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_i32 = veorq_s32(a_.neon_i32, b_.neon_i32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_xor(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_i32 = vec_xor(a_.altivec_i32, b_.altivec_i32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vxor_v(a_.lsx_i64, b_.lsx_i64);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32f = a_.i32f ^ b_.i32f;
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.u32) / sizeof(r_.u32[0])) ; i++) {
        r_.u32[i] = a_.u32[i] ^ b_.u32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_xor_ps(a, b) simde_mm_xor_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_or_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_or_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_i32 = vorrq_s32(a_.neon_i32, b_.neon_i32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_or(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_i32 = vec_or(a_.altivec_i32, b_.altivec_i32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vor_v(a_.lsx_i64, b_.lsx_i64);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32f = a_.i32f | b_.i32f;
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.u32) / sizeof(r_.u32[0])) ; i++) {
        r_.u32[i] = a_.u32[i] | b_.u32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_or_ps(a, b) simde_mm_or_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_not_ps(simde__m128 a) {
  #if defined(SIMDE_X86_AVX512VL_NATIVE)
    __m128i ai = _mm_castps_si128(a);
    return _mm_castsi128_ps(_mm_ternarylogic_epi32(ai, ai, ai, 0x55));
  #elif defined(SIMDE_X86_SSE2_NATIVE)
    /* Note: we use ints instead of floats because we don't want cmpeq
     * to return false for (NaN, NaN) */
    __m128i ai = _mm_castps_si128(a);
    return _mm_castsi128_ps(_mm_andnot_si128(ai, _mm_cmpeq_epi32(ai, ai)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_i32 = vmvnq_s32(a_.neon_i32);
    #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_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vnor_v(a_.lsx_i64, a_.lsx_i64);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32 = ~a_.i32;
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.i32) / sizeof(r_.i32[0])) ; i++) {
        r_.i32[i] = ~(a_.i32[i]);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_select_ps(simde__m128 a, simde__m128 b, simde__m128 mask) {
  /* This function is for when you want to blend two elements together
   * according to a mask.  It is similar to _mm_blendv_ps, 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_ps(a, b, mask);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b),
      mask_ = simde__m128_to_private(mask);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_i32 = vbslq_s32(mask_.neon_u32, b_.neon_i32, a_.neon_i32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_bitselect(b_.wasm_v128, a_.wasm_v128, mask_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE)
      r_.altivec_i32 = vec_sel(a_.altivec_i32, b_.altivec_i32, mask_.altivec_u32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vbitsel_v(a_.lsx_i64, b_.lsx_i64, mask_.lsx_i64);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32 = a_.i32 ^ ((a_.i32 ^ b_.i32) & mask_.i32);
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.i32) / sizeof(r_.i32[0])) ; i++) {
        r_.i32[i] = a_.i32[i] ^ ((a_.i32[i] ^ b_.i32[i]) & mask_.i32[i]);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_avg_pu16 (simde__m64 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_avg_pu16(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_u16 = vrhadd_u16(b_.neon_u16, a_.neon_u16);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && defined(SIMDE_CONVERT_VECTOR_) && !defined(SIMDE_BUG_GCC_100761)
      uint32_t wa SIMDE_VECTOR(16);
      uint32_t wb SIMDE_VECTOR(16);
      uint32_t wr SIMDE_VECTOR(16);
      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__m64_from_private(r_);
  #endif
}
#define simde_m_pavgw(a, b) simde_mm_avg_pu16(a, b)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_avg_pu16(a, b) simde_mm_avg_pu16(a, b)
#  define _m_pavgw(a, b) simde_mm_avg_pu16(a, b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_avg_pu8 (simde__m64 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_avg_pu8(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_u8 = vrhadd_u8(b_.neon_u8, a_.neon_u8);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && defined(SIMDE_CONVERT_VECTOR_) && !defined(SIMDE_BUG_GCC_100761)
      uint16_t wa SIMDE_VECTOR(16);
      uint16_t wb SIMDE_VECTOR(16);
      uint16_t wr SIMDE_VECTOR(16);
      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__m64_from_private(r_);
  #endif
}
#define simde_m_pavgb(a, b) simde_mm_avg_pu8(a, b)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_avg_pu8(a, b) simde_mm_avg_pu8(a, b)
#  define _m_pavgb(a, b) simde_mm_avg_pu8(a, b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_abs_ps(simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    simde_float32 mask_;
    uint32_t u32_ = UINT32_C(0x7FFFFFFF);
    simde_memcpy(&mask_, &u32_, sizeof(u32_));
    return _mm_and_ps(_mm_set1_ps(mask_), a);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vabsq_f32(a_.neon_f32);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
      r_.altivec_f32 = vec_abs(a_.altivec_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_abs(a_.wasm_v128);
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = simde_math_fabsf(a_.f32[i]);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpeq_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmpeq_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_u32 = vceqq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_eq(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_cmpeq(a_.altivec_f32, b_.altivec_f32));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vfcmp_ceq_s(a_.lsx_f32, b_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32 = HEDLEY_REINTERPRET_CAST(__typeof__(r_.i32), a_.f32 == b_.f32);
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.u32[i] = (a_.f32[i] == b_.f32[i]) ? ~UINT32_C(0) : UINT32_C(0);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpeq_ps(a, b) simde_mm_cmpeq_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpeq_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmpeq_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_cmpeq_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_cmpeq_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    r_.u32[0] = (a_.f32[0] == b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
    SIMDE_VECTORIZE
    for (size_t i = 1 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
      r_.u32[i] = a_.u32[i];
    }

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpeq_ss(a, b) simde_mm_cmpeq_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpge_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmpge_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_u32 = vcgeq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_ge(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_cmpge(a_.altivec_f32, b_.altivec_f32));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vfcmp_cle_s(b_.lsx_f32, a_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32 = HEDLEY_REINTERPRET_CAST(__typeof__(r_.i32), (a_.f32 >= b_.f32));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.u32[i] = (a_.f32[i] >= b_.f32[i]) ? ~UINT32_C(0) : UINT32_C(0);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpge_ps(a, b) simde_mm_cmpge_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpge_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
    return _mm_cmpge_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_cmpge_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_cmpge_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    r_.u32[0] = (a_.f32[0] >= b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
    SIMDE_VECTORIZE
    for (size_t i = 1 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
      r_.u32[i] = a_.u32[i];
    }

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpge_ss(a, b) simde_mm_cmpge_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpgt_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmpgt_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_u32 = vcgtq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_gt(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_cmpgt(a_.altivec_f32, b_.altivec_f32));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vfcmp_clt_s(b_.lsx_f32, a_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32 = HEDLEY_REINTERPRET_CAST(__typeof__(r_.i32), (a_.f32 > b_.f32));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.u32[i] = (a_.f32[i] > b_.f32[i]) ? ~UINT32_C(0) : UINT32_C(0);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpgt_ps(a, b) simde_mm_cmpgt_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpgt_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
    return _mm_cmpgt_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_cmpgt_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_cmpgt_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    r_.u32[0] = (a_.f32[0] > b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
    SIMDE_VECTORIZE
    for (size_t i = 1 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
      r_.u32[i] = a_.u32[i];
    }

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpgt_ss(a, b) simde_mm_cmpgt_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmple_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmple_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_u32 = vcleq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_le(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_cmple(a_.altivec_f32, b_.altivec_f32));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vfcmp_cle_s(a_.lsx_f32, b_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32 = HEDLEY_REINTERPRET_CAST(__typeof__(r_.i32), (a_.f32 <= b_.f32));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.u32[i] = (a_.f32[i] <= b_.f32[i]) ? ~UINT32_C(0) : UINT32_C(0);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmple_ps(a, b) simde_mm_cmple_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmple_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmple_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_cmple_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_cmple_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    r_.u32[0] = (a_.f32[0] <= b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
    SIMDE_VECTORIZE
    for (size_t i = 1 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
      r_.u32[i] = a_.u32[i];
    }

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmple_ss(a, b) simde_mm_cmple_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmplt_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmplt_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_u32 = vcltq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_lt(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_cmplt(a_.altivec_f32, b_.altivec_f32));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vfcmp_clt_s(a_.lsx_f32, b_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32 = HEDLEY_REINTERPRET_CAST(__typeof__(r_.i32), (a_.f32 < b_.f32));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.u32[i] = (a_.f32[i] < b_.f32[i]) ? ~UINT32_C(0) : UINT32_C(0);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmplt_ps(a, b) simde_mm_cmplt_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmplt_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmplt_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_cmplt_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_cmplt_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    r_.u32[0] = (a_.f32[0] < b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
    SIMDE_VECTORIZE
    for (size_t i = 1 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
      r_.u32[i] = a_.u32[i];
    }

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmplt_ss(a, b) simde_mm_cmplt_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpneq_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmpneq_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_u32 = vmvnq_u32(vceqq_f32(a_.neon_f32, b_.neon_f32));
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_ne(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_cmpeq(a_.altivec_f32, b_.altivec_f32));
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), vec_nor(r_.altivec_f32, r_.altivec_f32));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vfcmp_cune_s(a_.lsx_f32, b_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.i32 = HEDLEY_REINTERPRET_CAST(__typeof__(r_.i32), (a_.f32 != b_.f32));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.u32[i] = (a_.f32[i] != b_.f32[i]) ? ~UINT32_C(0) : UINT32_C(0);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpneq_ps(a, b) simde_mm_cmpneq_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpneq_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmpneq_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_cmpneq_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_cmpneq_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    r_.u32[0] = (a_.f32[0] != b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
    SIMDE_VECTORIZE
    for (size_t i = 1 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
      r_.u32[i] = a_.u32[i];
    }

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpneq_ss(a, b) simde_mm_cmpneq_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpnge_ps (simde__m128 a, simde__m128 b) {
  return simde_mm_cmplt_ps(a, b);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpnge_ps(a, b) simde_mm_cmpnge_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpnge_ss (simde__m128 a, simde__m128 b) {
  return simde_mm_cmplt_ss(a, b);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpnge_ss(a, b) simde_mm_cmpnge_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpngt_ps (simde__m128 a, simde__m128 b) {
  return simde_mm_cmple_ps(a, b);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpngt_ps(a, b) simde_mm_cmpngt_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpngt_ss (simde__m128 a, simde__m128 b) {
  return simde_mm_cmple_ss(a, b);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpngt_ss(a, b) simde_mm_cmpngt_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpnle_ps (simde__m128 a, simde__m128 b) {
  return simde_mm_cmpgt_ps(a, b);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpnle_ps(a, b) simde_mm_cmpnle_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpnle_ss (simde__m128 a, simde__m128 b) {
  return simde_mm_cmpgt_ss(a, b);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpnle_ss(a, b) simde_mm_cmpnle_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpnlt_ps (simde__m128 a, simde__m128 b) {
  return simde_mm_cmpge_ps(a, b);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpnlt_ps(a, b) simde_mm_cmpnlt_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpnlt_ss (simde__m128 a, simde__m128 b) {
  return simde_mm_cmpge_ss(a, b);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpnlt_ss(a, b) simde_mm_cmpnlt_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpord_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmpord_ps(a, b);
  #elif defined(SIMDE_WASM_SIMD128_NATIVE)
    return wasm_v128_and(wasm_f32x4_eq(a, a), wasm_f32x4_eq(b, b));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      /* Note: NEON does not have ordered compare builtin
        Need to compare a eq a and b eq b to check for NaN
        Do AND of results to get final */
      uint32x4_t ceqaa = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t ceqbb = vceqq_f32(b_.neon_f32, b_.neon_f32);
      r_.neon_u32 = vandq_u32(ceqaa, ceqbb);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_and(wasm_f32x4_eq(a_.wasm_v128, a_.wasm_v128), wasm_f32x4_eq(b_.wasm_v128, b_.wasm_v128));
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float),
          vec_and(vec_cmpeq(a_.altivec_f32, a_.altivec_f32), vec_cmpeq(b_.altivec_f32, b_.altivec_f32)));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vfcmp_cun_s(a_.lsx_f32, b_.lsx_f32);
      r_.lsx_i64 = __lsx_vnor_v(r_.lsx_i64, r_.lsx_i64);
    #elif defined(simde_math_isnanf)
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.u32[i] = (simde_math_isnanf(a_.f32[i]) || simde_math_isnanf(b_.f32[i])) ? UINT32_C(0) : ~UINT32_C(0);
      }
    #else
      HEDLEY_UNREACHABLE();
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpord_ps(a, b) simde_mm_cmpord_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpunord_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmpunord_ps(a, b);
  #elif defined(SIMDE_WASM_SIMD128_NATIVE)
    return wasm_v128_or(wasm_f32x4_ne(a, a), wasm_f32x4_ne(b, b));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t ceqaa = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t ceqbb = vceqq_f32(b_.neon_f32, b_.neon_f32);
      r_.neon_u32 = vmvnq_u32(vandq_u32(ceqaa, ceqbb));
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_or(wasm_f32x4_ne(a_.wasm_v128, a_.wasm_v128), wasm_f32x4_ne(b_.wasm_v128, b_.wasm_v128));
    #elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float),
          vec_nand(vec_cmpeq(a_.altivec_f32, a_.altivec_f32), vec_cmpeq(b_.altivec_f32, b_.altivec_f32)));
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float),
          vec_and(vec_cmpeq(a_.altivec_f32, a_.altivec_f32), vec_cmpeq(b_.altivec_f32, b_.altivec_f32)));
      r_.altivec_f32 = vec_nor(r_.altivec_f32, r_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vfcmp_cun_s(a_.lsx_f32, b_.lsx_f32);
    #elif defined(simde_math_isnanf)
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.u32[i] = (simde_math_isnanf(a_.f32[i]) || simde_math_isnanf(b_.f32[i])) ? ~UINT32_C(0) : UINT32_C(0);
      }
    #else
      HEDLEY_UNREACHABLE();
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpunord_ps(a, b) simde_mm_cmpunord_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpunord_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
    return _mm_cmpunord_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_cmpunord_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_cmpunord_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(simde_math_isnanf)
      r_.u32[0] = (simde_math_isnanf(a_.f32[0]) || simde_math_isnanf(b_.f32[0])) ? ~UINT32_C(0) : UINT32_C(0);
      SIMDE_VECTORIZE
      for (size_t i = 1 ; i < (sizeof(r_.u32) / sizeof(r_.u32[0])) ; i++) {
        r_.u32[i] = a_.u32[i];
      }
    #else
      HEDLEY_UNREACHABLE();
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpunord_ss(a, b) simde_mm_cmpunord_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_comieq_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_comieq_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
      uint32x4_t a_eq_b = vceqq_f32(a_.neon_f32, b_.neon_f32);
      return !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_eq_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      return wasm_f32x4_extract_lane(a_.wasm_v128, 0) == wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #else
      return a_.f32[0] == b_.f32[0];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_comieq_ss(a, b) simde_mm_comieq_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_comige_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_comige_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
      uint32x4_t a_ge_b = vcgeq_f32(a_.neon_f32, b_.neon_f32);
      return !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_ge_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      return wasm_f32x4_extract_lane(a_.wasm_v128, 0) >= wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #else
      return a_.f32[0] >= b_.f32[0];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_comige_ss(a, b) simde_mm_comige_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_comigt_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_comigt_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
      uint32x4_t a_gt_b = vcgtq_f32(a_.neon_f32, b_.neon_f32);
      return !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_gt_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      return wasm_f32x4_extract_lane(a_.wasm_v128, 0) > wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #else
      return a_.f32[0] > b_.f32[0];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_comigt_ss(a, b) simde_mm_comigt_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_comile_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_comile_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
      uint32x4_t a_le_b = vcleq_f32(a_.neon_f32, b_.neon_f32);
      return !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_le_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      return wasm_f32x4_extract_lane(a_.wasm_v128, 0) <= wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #else
      return a_.f32[0] <= b_.f32[0];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_comile_ss(a, b) simde_mm_comile_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_comilt_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_comilt_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
      uint32x4_t a_lt_b = vcltq_f32(a_.neon_f32, b_.neon_f32);
      return !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_lt_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      return wasm_f32x4_extract_lane(a_.wasm_v128, 0) < wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #else
      return a_.f32[0] < b_.f32[0];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_comilt_ss(a, b) simde_mm_comilt_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_comineq_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_comineq_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
      uint32x4_t a_neq_b = vmvnq_u32(vceqq_f32(a_.neon_f32, b_.neon_f32));
      return !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_neq_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      return wasm_f32x4_extract_lane(a_.wasm_v128, 0) != wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #else
      return a_.f32[0] != b_.f32[0];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_comineq_ss(a, b) simde_mm_comineq_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_copysign_ps(simde__m128 dest, simde__m128 src) {
  simde__m128_private
    r_,
    dest_ = simde__m128_to_private(dest),
    src_ = simde__m128_to_private(src);

  #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
    const uint32x4_t sign_pos = vreinterpretq_u32_f32(vdupq_n_f32(-SIMDE_FLOAT32_C(0.0)));
    r_.neon_u32 = vbslq_u32(sign_pos, src_.neon_u32, dest_.neon_u32);
  #elif defined(SIMDE_WASM_SIMD128_NATIVE)
    const v128_t sign_pos = wasm_f32x4_splat(-0.0f);
    r_.wasm_v128 = wasm_v128_bitselect(src_.wasm_v128, dest_.wasm_v128, sign_pos);
  #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
    #if defined(SIMDE_BUG_VEC_CPSGN_REVERSED_ARGS)
      r_.altivec_f32 = vec_cpsgn(dest_.altivec_f32, src_.altivec_f32);
    #else
      r_.altivec_f32 = vec_cpsgn(src_.altivec_f32, dest_.altivec_f32);
    #endif
  #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
    const SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) sign_pos = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(unsigned int), vec_splats(-0.0f));
    r_.altivec_f32 = vec_sel(dest_.altivec_f32, src_.altivec_f32, sign_pos);
  #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
    const v4f32 sign_pos = {-0.0f, -0.0f, -0.0f, -0.0f};
    r_.lsx_i64 = __lsx_vbitsel_v(dest_.lsx_i64, src_.lsx_i64, (v2i64)sign_pos);
  #elif defined(SIMDE_IEEE754_STORAGE)
    (void) src_;
    (void) dest_;
    simde__m128 sign_pos = simde_mm_set1_ps(-0.0f);
    r_ = simde__m128_to_private(simde_mm_xor_ps(dest, simde_mm_and_ps(simde_mm_xor_ps(dest, src), sign_pos)));
  #else
    SIMDE_VECTORIZE
    for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
      r_.f32[i] = simde_math_copysignf(dest_.f32[i], src_.f32[i]);
    }
  #endif

  return simde__m128_from_private(r_);
}

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_xorsign_ps(simde__m128 dest, simde__m128 src) {
  return simde_mm_xor_ps(simde_mm_and_ps(simde_mm_set1_ps(-0.0f), src), dest);
}

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvt_pi2ps (simde__m128 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvt_pi2ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);
    simde__m64_private b_ = simde__m64_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vcombine_f32(vcvt_f32_s32(b_.neon_i32), vget_high_f32(a_.neon_f32));
    #elif defined(SIMDE_CONVERT_VECTOR_)
      SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, b_.i32);
      r_.m64_private[1] = a_.m64_private[1];
    #else
      r_.f32[0] = (simde_float32) b_.i32[0];
      r_.f32[1] = (simde_float32) b_.i32[1];
      r_.i32[2] = a_.i32[2];
      r_.i32[3] = a_.i32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvt_pi2ps(a, b) simde_mm_cvt_pi2ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_cvt_ps2pi (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvt_ps2pi(a);
  #else
    simde__m64_private r_;
    simde__m128_private a_;

  #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
    a_ = simde__m128_to_private(simde_mm_round_ps(a, SIMDE_MM_FROUND_CUR_DIRECTION));
    r_.neon_i32 = vcvt_s32_f32(vget_low_f32(a_.neon_f32));
  #elif defined(SIMDE_CONVERT_VECTOR_) && SIMDE_NATURAL_VECTOR_SIZE_GE(128) && !defined(SIMDE_BUG_GCC_100761)
    a_ = simde__m128_to_private(simde_mm_round_ps(a, SIMDE_MM_FROUND_CUR_DIRECTION));
    SIMDE_CONVERT_VECTOR_(r_.i32, a_.m64_private[0].f32);
  #else
    a_ = simde__m128_to_private(a);

    SIMDE_VECTORIZE
    for (size_t i = 0 ; i < (sizeof(r_.i32) / sizeof(r_.i32[0])) ; i++) {
      r_.i32[i] = HEDLEY_STATIC_CAST(int32_t, simde_math_nearbyintf(a_.f32[i]));
    }
  #endif

    return simde__m64_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvt_ps2pi(a) simde_mm_cvt_ps2pi((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvt_si2ss (simde__m128 a, int32_t b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cvt_si2ss(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vsetq_lane_f32(HEDLEY_STATIC_CAST(float, b), a_.neon_f32, 0);
    #else
      r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b);
      r_.i32[1] = a_.i32[1];
      r_.i32[2] = a_.i32[2];
      r_.i32[3] = a_.i32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvt_si2ss(a, b) simde_mm_cvt_si2ss((a), b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
int32_t
simde_mm_cvt_ss2si (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cvt_ss2si(a);
  #elif defined(SIMDE_ARM_NEON_A32V8_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE) && !defined(SIMDE_BUG_GCC_95399)
    return vgetq_lane_s32(vcvtnq_s32_f32(simde__m128_to_neon_f32(a)), 0);
  #else
    simde__m128_private a_ = simde__m128_to_private(simde_mm_round_ps(a, SIMDE_MM_FROUND_CUR_DIRECTION));
    #if !defined(SIMDE_FAST_CONVERSION_RANGE)
      return ((a_.f32[0] > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) &&
          (a_.f32[0] < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX))) ?
        SIMDE_CONVERT_FTOI(int32_t, a_.f32[0]) : INT32_MIN;
    #else
      return SIMDE_CONVERT_FTOI(int32_t, a_.f32[0]);
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvt_ss2si(a) simde_mm_cvt_ss2si((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvtpi16_ps (simde__m64 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtpi16_ps(a);
  #else
    simde__m128_private r_;
    simde__m64_private a_ = simde__m64_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vcvtq_f32_s32(vmovl_s16(a_.neon_i16));
    #elif defined(SIMDE_CONVERT_VECTOR_)
      SIMDE_CONVERT_VECTOR_(r_.f32, a_.i16);
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        simde_float32 v = a_.i16[i];
        r_.f32[i] = v;
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtpi16_ps(a) simde_mm_cvtpi16_ps(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvtpi32_ps (simde__m128 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtpi32_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);
    simde__m64_private b_ = simde__m64_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vcombine_f32(vcvt_f32_s32(b_.neon_i32), vget_high_f32(a_.neon_f32));
    #elif defined(SIMDE_CONVERT_VECTOR_)
      SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, b_.i32);
      r_.m64_private[1] = a_.m64_private[1];
    #else
      r_.f32[0] = (simde_float32) b_.i32[0];
      r_.f32[1] = (simde_float32) b_.i32[1];
      r_.i32[2] = a_.i32[2];
      r_.i32[3] = a_.i32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtpi32_ps(a, b) simde_mm_cvtpi32_ps((a), b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvtpi32x2_ps (simde__m64 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtpi32x2_ps(a, b);
  #else
    simde__m128_private r_;
    simde__m64_private
      a_ = simde__m64_to_private(a),
      b_ = simde__m64_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vcvtq_f32_s32(vcombine_s32(a_.neon_i32, b_.neon_i32));
    #elif defined(SIMDE_CONVERT_VECTOR_)
      SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, a_.i32);
      SIMDE_CONVERT_VECTOR_(r_.m64_private[1].f32, b_.i32);
    #else
      r_.f32[0] = (simde_float32) a_.i32[0];
      r_.f32[1] = (simde_float32) a_.i32[1];
      r_.f32[2] = (simde_float32) b_.i32[0];
      r_.f32[3] = (simde_float32) b_.i32[1];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtpi32x2_ps(a, b) simde_mm_cvtpi32x2_ps(a, b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvtpi8_ps (simde__m64 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtpi8_ps(a);
  #else
    simde__m128_private r_;
    simde__m64_private a_ = simde__m64_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vcvtq_f32_s32(vmovl_s16(vget_low_s16(vmovl_s8(a_.neon_i8))));
    #else
      r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[0]);
      r_.f32[1] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[1]);
      r_.f32[2] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[2]);
      r_.f32[3] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[3]);
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtpi8_ps(a) simde_mm_cvtpi8_ps(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_cvtps_pi16 (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtps_pi16(a);
  #else
    simde__m64_private r_;
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V8_NATIVE) && !defined(SIMDE_BUG_GCC_95399)
      r_.neon_i16 = vmovn_s32(vcvtq_s32_f32(vrndiq_f32(a_.neon_f32)));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.i16) / sizeof(r_.i16[0])) ; i++) {
        r_.i16[i] = SIMDE_CONVERT_FTOI(int16_t, simde_math_roundf(a_.f32[i]));
      }
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtps_pi16(a) simde_mm_cvtps_pi16((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_cvtps_pi32 (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtps_pi32(a);
  #else
    simde__m64_private r_;
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V8_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE) && !defined(SIMDE_BUG_GCC_95399)
      r_.neon_i32 = vcvt_s32_f32(vget_low_f32(vrndiq_f32(a_.neon_f32)));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.i32) / sizeof(r_.i32[0])) ; i++) {
        simde_float32 v = simde_math_roundf(a_.f32[i]);
        #if !defined(SIMDE_FAST_CONVERSION_RANGE)
          r_.i32[i] = ((v > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) && (v < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX))) ?
            SIMDE_CONVERT_FTOI(int32_t, v) : INT32_MIN;
        #else
          r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
        #endif
      }
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtps_pi32(a) simde_mm_cvtps_pi32((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_cvtps_pi8 (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtps_pi8(a);
  #else
    simde__m64_private r_;
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V8_NATIVE) && !defined(SIMDE_BUG_GCC_95471)
      /* Clamp the input to [INT8_MIN, INT8_MAX], round, convert to i32, narrow to
      * i16, combine with an all-zero vector of i16 (which will become the upper
      * half), narrow to i8. */
      float32x4_t max = vdupq_n_f32(HEDLEY_STATIC_CAST(simde_float32, INT8_MAX));
      float32x4_t min = vdupq_n_f32(HEDLEY_STATIC_CAST(simde_float32, INT8_MIN));
      float32x4_t values = vrndnq_f32(vmaxq_f32(vminq_f32(max, a_.neon_f32), min));
      r_.neon_i8 = vmovn_s16(vcombine_s16(vmovn_s32(vcvtq_s32_f32(values)), vdup_n_s16(0)));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(a_.f32) / sizeof(a_.f32[0])) ; i++) {
        if (a_.f32[i] > HEDLEY_STATIC_CAST(simde_float32, INT8_MAX))
          r_.i8[i] = INT8_MAX;
        else if (a_.f32[i] <  HEDLEY_STATIC_CAST(simde_float32, INT8_MIN))
          r_.i8[i] = INT8_MIN;
        else
          r_.i8[i] = SIMDE_CONVERT_FTOI(int8_t, simde_math_roundf(a_.f32[i]));
      }
      /* Note: the upper half is undefined */
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtps_pi8(a) simde_mm_cvtps_pi8((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvtpu16_ps (simde__m64 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtpu16_ps(a);
  #else
    simde__m128_private r_;
    simde__m64_private a_ = simde__m64_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vcvtq_f32_u32(vmovl_u16(a_.neon_u16));
    #elif defined(SIMDE_CONVERT_VECTOR_)
      SIMDE_CONVERT_VECTOR_(r_.f32, a_.u16);
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = (simde_float32) a_.u16[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtpu16_ps(a) simde_mm_cvtpu16_ps(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvtpu8_ps (simde__m64 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtpu8_ps(a);
  #else
    simde__m128_private r_;
    simde__m64_private a_ = simde__m64_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vcvtq_f32_u32(vmovl_u16(vget_low_u16(vmovl_u8(a_.neon_u8))));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = HEDLEY_STATIC_CAST(simde_float32, a_.u8[i]);
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtpu8_ps(a) simde_mm_cvtpu8_ps(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvtsi32_ss (simde__m128 a, int32_t b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cvtsi32_ss(a, b);
  #else
    simde__m128_private r_;
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vsetq_lane_f32(HEDLEY_STATIC_CAST(float32_t, b), a_.neon_f32, 0);
    #else
      r_ = a_;
      r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b);
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtsi32_ss(a, b) simde_mm_cvtsi32_ss((a), b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cvtsi64_ss (simde__m128 a, int64_t b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ARCH_AMD64)
    #if !defined(__PGI)
      return _mm_cvtsi64_ss(a, b);
    #else
      return _mm_cvtsi64x_ss(a, b);
    #endif
  #else
    simde__m128_private r_;
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vsetq_lane_f32(HEDLEY_STATIC_CAST(float32_t, b), a_.neon_f32, 0);
    #else
      r_ = a_;
      r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b);
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && !defined(SIMDE_ARCH_AMD64))
#  define _mm_cvtsi64_ss(a, b) simde_mm_cvtsi64_ss((a), b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde_float32
simde_mm_cvtss_f32 (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cvtss_f32(a);
  #else
    simde__m128_private a_ = simde__m128_to_private(a);
    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      return vgetq_lane_f32(a_.neon_f32, 0);
    #else
      return a_.f32[0];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtss_f32(a) simde_mm_cvtss_f32((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int32_t
simde_mm_cvtss_si32 (simde__m128 a) {
  return simde_mm_cvt_ss2si(a);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtss_si32(a) simde_mm_cvtss_si32((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int64_t
simde_mm_cvtss_si64 (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ARCH_AMD64)
    #if !defined(__PGI)
      return _mm_cvtss_si64(a);
    #else
      return _mm_cvtss_si64x(a);
    #endif
  #else
    simde__m128_private a_ = simde__m128_to_private(a);
    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      return SIMDE_CONVERT_FTOI(int64_t, simde_math_roundf(vgetq_lane_f32(a_.neon_f32, 0)));
    #else
      return SIMDE_CONVERT_FTOI(int64_t, simde_math_roundf(a_.f32[0]));
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && !defined(SIMDE_ARCH_AMD64))
#  define _mm_cvtss_si64(a) simde_mm_cvtss_si64((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_cvtt_ps2pi (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_cvtt_ps2pi(a);
  #else
    simde__m64_private r_;
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE)
      r_.neon_i32 = vcvt_s32_f32(vget_low_f32(a_.neon_f32));
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        simde_float32 v = a_.f32[i];
        #if !defined(SIMDE_FAST_CONVERSION_RANGE)
          r_.i32[i] = ((v > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) && (v < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX))) ?
            SIMDE_CONVERT_FTOI(int32_t, v) : INT32_MIN;
        #else
          r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
        #endif
      }
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#define simde_mm_cvttps_pi32(a) simde_mm_cvtt_ps2pi(a)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtt_ps2pi(a) simde_mm_cvtt_ps2pi((a))
#  define _mm_cvttps_pi32(a) simde_mm_cvttps_pi32((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int32_t
simde_mm_cvtt_ss2si (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cvtt_ss2si(a);
  #else
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE)
      return SIMDE_CONVERT_FTOI(int32_t, vgetq_lane_f32(a_.neon_f32, 0));
    #else
      simde_float32 v = a_.f32[0];
      #if !defined(SIMDE_FAST_CONVERSION_RANGE)
        return ((v > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) && (v < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX))) ?
          SIMDE_CONVERT_FTOI(int32_t, v) : INT32_MIN;
      #else
        return SIMDE_CONVERT_FTOI(int32_t, v);
      #endif
    #endif
  #endif
}
#define simde_mm_cvttss_si32(a) simde_mm_cvtt_ss2si((a))
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cvtt_ss2si(a) simde_mm_cvtt_ss2si((a))
#  define _mm_cvttss_si32(a) simde_mm_cvtt_ss2si((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int64_t
simde_mm_cvttss_si64 (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ARCH_AMD64) && !defined(_MSC_VER)
    #if defined(__PGI)
      return _mm_cvttss_si64x(a);
    #else
      return _mm_cvttss_si64(a);
    #endif
  #else
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      return SIMDE_CONVERT_FTOI(int64_t, vgetq_lane_f32(a_.neon_f32, 0));
    #else
      return SIMDE_CONVERT_FTOI(int64_t, a_.f32[0]);
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && !defined(SIMDE_ARCH_AMD64))
#  define _mm_cvttss_si64(a) simde_mm_cvttss_si64((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_cmpord_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_cmpord_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_cmpord_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_cmpord_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(simde_math_isnanf)
      r_.u32[0] = (simde_math_isnanf(simde_mm_cvtss_f32(a)) || simde_math_isnanf(simde_mm_cvtss_f32(b))) ? UINT32_C(0) : ~UINT32_C(0);
      SIMDE_VECTORIZE
      for (size_t i = 1 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.u32[i] = a_.u32[i];
      }
    #else
      HEDLEY_UNREACHABLE();
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_cmpord_ss(a, b) simde_mm_cmpord_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_div_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_div_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
      r_.neon_f32 = vdivq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32x4_t recip0 = vrecpeq_f32(b_.neon_f32);
      float32x4_t recip1 = vmulq_f32(recip0, vrecpsq_f32(recip0, b_.neon_f32));
      r_.neon_f32 = vmulq_f32(a_.neon_f32, recip1);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 =  wasm_f32x4_div(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
      r_.altivec_f32 = vec_div(a_.altivec_f32, b_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LASX_NATIVE)
      r_.lsx_f32 = __lsx_vfdiv_s(a_.lsx_f32, b_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.f32 = a_.f32 / b_.f32;
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = a_.f32[i] / b_.f32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_div_ps(a, b) simde_mm_div_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_div_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_div_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_div_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_div_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32_t value =
              vgetq_lane_f32(simde__m128_to_private(simde_mm_div_ps(a, b)).neon_f32, 0);
      r_.neon_f32 = vsetq_lane_f32(value, a_.neon_f32, 0);
    #else
      r_.f32[0] = a_.f32[0] / b_.f32[0];
      SIMDE_VECTORIZE
      for (size_t i = 1 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = a_.f32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_div_ss(a, b) simde_mm_div_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int16_t
simde_mm_extract_pi16 (simde__m64 a, const int imm8)
    SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 3) {
  simde__m64_private a_ = simde__m64_to_private(a);
  return a_.i16[imm8];
}
#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && !defined(HEDLEY_PGI_VERSION) && !defined(SIMDE_BUG_CLANG_44589)
  #define simde_mm_extract_pi16(a, imm8) HEDLEY_STATIC_CAST(int16_t, _mm_extract_pi16(a, imm8))
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
  #define simde_mm_extract_pi16(a, imm8) vget_lane_s16(simde__m64_to_private(a).neon_i16, imm8)
#endif
#define simde_m_pextrw(a, imm8) simde_mm_extract_pi16(a, imm8)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_extract_pi16(a, imm8) simde_mm_extract_pi16((a), (imm8))
#  define _m_pextrw(a, imm8) simde_mm_extract_pi16((a), (imm8))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_insert_pi16 (simde__m64 a, int16_t i, const int imm8)
    SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 3) {
  simde__m64_private
    a_ = simde__m64_to_private(a);

  a_.i16[imm8] = i;

  return simde__m64_from_private(a_);
}
#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI) && !defined(SIMDE_BUG_CLANG_44589)
  #define simde_mm_insert_pi16(a, i, imm8) _mm_insert_pi16(a, i, imm8)
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
  #define simde_mm_insert_pi16(a, i, imm8) simde__m64_from_neon_i16(vset_lane_s16((i), simde__m64_to_neon_i16(a), (imm8)))
#endif
#define simde_m_pinsrw(a, i, imm8) (simde_mm_insert_pi16(a, i, imm8))
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_insert_pi16(a, i, imm8) simde_mm_insert_pi16(a, i, imm8)
#  define _m_pinsrw(a, i, imm8) simde_mm_insert_pi16(a, i, imm8)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_load_ps (simde_float32 const mem_addr[HEDLEY_ARRAY_PARAM(4)]) {
#if defined(SIMDE_X86_SSE_NATIVE)
  return _mm_load_ps(mem_addr);
#else
  simde__m128_private r_;

  #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
    r_.neon_f32 = vld1q_f32(mem_addr);
  #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
    r_.altivec_f32 = vec_vsx_ld(0, mem_addr);
  #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
    r_.altivec_f32 = vec_ld(0, mem_addr);
  #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
    r_.lsx_i64 = __lsx_vld(mem_addr, 0);
  #elif defined(SIMDE_WASM_SIMD128_NATIVE)
    r_.wasm_v128 = wasm_v128_load(mem_addr);
  #else
    simde_memcpy(&r_, SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m128), sizeof(r_));
  #endif

  return simde__m128_from_private(r_);
#endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_load_ps(mem_addr) simde_mm_load_ps(mem_addr)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_load1_ps (simde_float32 const* mem_addr) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_load_ps1(mem_addr);
  #else
    simde__m128_private r_;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vld1q_dup_f32(mem_addr);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vldrepl_w(mem_addr, 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_load32_splat(mem_addr);
    #else
      r_ = simde__m128_to_private(simde_mm_set1_ps(*mem_addr));
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#define simde_mm_load_ps1(mem_addr) simde_mm_load1_ps(mem_addr)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_load_ps1(mem_addr) simde_mm_load1_ps(mem_addr)
#  define _mm_load1_ps(mem_addr) simde_mm_load1_ps(mem_addr)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_load_ss (simde_float32 const* mem_addr) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_load_ss(mem_addr);
  #else
    simde__m128_private r_;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vsetq_lane_f32(*mem_addr, vdupq_n_f32(0), 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_load32_zero(mem_addr);
    #else
      r_.f32[0] = *mem_addr;
      r_.i32[1] = 0;
      r_.i32[2] = 0;
      r_.i32[3] = 0;
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_load_ss(mem_addr) simde_mm_load_ss(mem_addr)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_loadh_pi (simde__m128 a, simde__m64 const* mem_addr) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_loadh_pi(a, HEDLEY_REINTERPRET_CAST(__m64 const*, mem_addr));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

  #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
    r_.neon_f32 = vcombine_f32(vget_low_f32(a_.neon_f32), vld1_f32(HEDLEY_REINTERPRET_CAST(const float32_t*, mem_addr)));
  #elif defined(SIMDE_WASM_SIMD128_NATIVE)
    r_.wasm_v128 = wasm_v128_load64_lane(mem_addr, a_.wasm_v128, 1);
  #else
    simde__m64_private b_ = *HEDLEY_REINTERPRET_CAST(simde__m64_private const*, mem_addr);
    r_.f32[0] = a_.f32[0];
    r_.f32[1] = a_.f32[1];
    r_.f32[2] = b_.f32[0];
    r_.f32[3] = b_.f32[1];
  #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #if HEDLEY_HAS_WARNING("-Wold-style-cast")
    #define _mm_loadh_pi(a, mem_addr) simde_mm_loadh_pi((a), HEDLEY_REINTERPRET_CAST(simde__m64 const*, (mem_addr)))
  #else
    #define _mm_loadh_pi(a, mem_addr) simde_mm_loadh_pi((a), (simde__m64 const*) (mem_addr))
  #endif
#endif

/* The SSE documentation says that there are no alignment requirements
   for mem_addr.  Unfortunately they used the __m64 type for the argument
   which is supposed to be 8-byte aligned, so some compilers (like clang
   with -Wcast-align) will generate a warning if you try to cast, say,
   a simde_float32* to a simde__m64* for this function.

   I think the choice of argument type is unfortunate, but I do think we
   need to stick to it here.  If there is demand I can always add something
   like simde_x_mm_loadl_f32(simde__m128, simde_float32 mem_addr[2]) */
SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_loadl_pi (simde__m128 a, simde__m64 const* mem_addr) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_loadl_pi(a, HEDLEY_REINTERPRET_CAST(__m64 const*, mem_addr));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vcombine_f32(vld1_f32(
        HEDLEY_REINTERPRET_CAST(const float32_t*, mem_addr)), vget_high_f32(a_.neon_f32));
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_load64_lane(mem_addr, a_.wasm_v128, 0);
    #else
      simde__m64_private b_;
      simde_memcpy(&b_, mem_addr, sizeof(b_));
      r_.i32[0] = b_.i32[0];
      r_.i32[1] = b_.i32[1];
      r_.i32[2] = a_.i32[2];
      r_.i32[3] = a_.i32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #if HEDLEY_HAS_WARNING("-Wold-style-cast")
    #define _mm_loadl_pi(a, mem_addr) simde_mm_loadl_pi((a), HEDLEY_REINTERPRET_CAST(simde__m64 const*, (mem_addr)))
  #else
    #define _mm_loadl_pi(a, mem_addr) simde_mm_loadl_pi((a), (simde__m64 const*) (mem_addr))
  #endif
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_loadr_ps (simde_float32 const mem_addr[HEDLEY_ARRAY_PARAM(4)]) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_loadr_ps(mem_addr);
  #else
    simde__m128_private
      r_,
      v_ = simde__m128_to_private(simde_mm_load_ps(mem_addr));

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vrev64q_f32(v_.neon_f32);
      r_.neon_f32 = vextq_f32(r_.neon_f32, r_.neon_f32, 2);
    #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) && defined(__PPC64__)
      r_.altivec_f32 = vec_reve(v_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vshuf4i_w(v_.lsx_i64, 0x1b);
    #elif defined(SIMDE_SHUFFLE_VECTOR_)
      r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, v_.f32, v_.f32, 3, 2, 1, 0);
    #else
      r_.f32[0] = v_.f32[3];
      r_.f32[1] = v_.f32[2];
      r_.f32[2] = v_.f32[1];
      r_.f32[3] = v_.f32[0];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_loadr_ps(mem_addr) simde_mm_loadr_ps(mem_addr)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_loadu_ps (simde_float32 const mem_addr[HEDLEY_ARRAY_PARAM(4)]) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_loadu_ps(mem_addr);
  #else
    simde__m128_private r_;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vld1q_f32(HEDLEY_REINTERPRET_CAST(const float32_t*, mem_addr));
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_load(mem_addr);
    #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) && defined(__PPC64__)
      r_.altivec_f32 = vec_vsx_ld(0, mem_addr);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vld(mem_addr, 0);
    #else
      simde_memcpy(&r_, mem_addr, sizeof(r_));
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_loadu_ps(mem_addr) simde_mm_loadu_ps(mem_addr)
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_maskmove_si64 (simde__m64 a, simde__m64 mask, int8_t* mem_addr) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    _mm_maskmove_si64(a, mask, HEDLEY_REINTERPRET_CAST(char*, mem_addr));
  #else
    simde__m64_private
      a_ = simde__m64_to_private(a),
      mask_ = simde__m64_to_private(mask);

    SIMDE_VECTORIZE
    for (size_t i = 0 ; i < (sizeof(a_.i8) / sizeof(a_.i8[0])) ; i++)
      if (mask_.i8[i] < 0)
        mem_addr[i] = a_.i8[i];
  #endif
}
#define simde_m_maskmovq(a, mask, mem_addr) simde_mm_maskmove_si64(a, mask, mem_addr)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_maskmove_si64(a, mask, mem_addr) simde_mm_maskmove_si64((a), (mask), SIMDE_CHECKED_REINTERPRET_CAST(int8_t*, char*, (mem_addr)))
#  define _m_maskmovq(a, mask, mem_addr) simde_mm_maskmove_si64((a), (mask), SIMDE_CHECKED_REINTERPRET_CAST(int8_t*, char*, (mem_addr)))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_max_pi16 (simde__m64 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_max_pi16(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_i16 = vmax_s16(a_.neon_i16, b_.neon_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]) ? a_.i16[i] : b_.i16[i];
      }
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#define simde_m_pmaxsw(a, b) simde_mm_max_pi16(a, b)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_max_pi16(a, b) simde_mm_max_pi16(a, b)
#  define _m_pmaxsw(a, b) simde_mm_max_pi16(a, b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_max_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_max_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_FAST_NANS)
      r_.neon_f32 = vmaxq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vbslq_f32(vcgtq_f32(a_.neon_f32, b_.neon_f32), a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE) && defined(SIMDE_FAST_NANS)
      r_.wasm_v128 = wasm_f32x4_max(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_v128_bitselect(a_.wasm_v128, b_.wasm_v128, wasm_f32x4_gt(a_.wasm_v128, b_.wasm_v128));
    #elif (defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)) && defined(SIMDE_FAST_NANS)
      r_.altivec_f32 = vec_max(a_.altivec_f32, b_.altivec_f32);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
      r_.altivec_f32 = vec_sel(b_.altivec_f32, a_.altivec_f32, vec_cmpgt(a_.altivec_f32, b_.altivec_f32));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE) && defined(SIMDE_FAST_NANS)
      r_.lsx_f32 = __lsx_vfmax_s(a_.lsx_f32, b_.lsx_f32);
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = (a_.f32[i] > b_.f32[i]) ? a_.f32[i] : b_.f32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_max_ps(a, b) simde_mm_max_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_max_pu8 (simde__m64 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_max_pu8(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_u8 = vmax_u8(a_.neon_u8, b_.neon_u8);
    #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]) ? a_.u8[i] : b_.u8[i];
      }
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#define simde_m_pmaxub(a, b) simde_mm_max_pu8(a, b)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_max_pu8(a, b) simde_mm_max_pu8(a, b)
#  define _m_pmaxub(a, b) simde_mm_max_pu8(a, b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_max_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_max_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_max_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_max_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32_t value = vgetq_lane_f32(maxq_f32(a_.neon_f32, b_.neon_f32), 0);
      r_.neon_f32 = vsetq_lane_f32(value, a_.neon_f32, 0);
    #else
      r_.f32[0] = (a_.f32[0] > b_.f32[0]) ? a_.f32[0] : b_.f32[0];
      r_.f32[1] = a_.f32[1];
      r_.f32[2] = a_.f32[2];
      r_.f32[3] = a_.f32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_max_ss(a, b) simde_mm_max_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_min_pi16 (simde__m64 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_min_pi16(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_i16 = vmin_s16(a_.neon_i16, b_.neon_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]) ? a_.i16[i] : b_.i16[i];
      }
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#define simde_m_pminsw(a, b) simde_mm_min_pi16(a, b)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_min_pi16(a, b) simde_mm_min_pi16(a, b)
#  define _m_pminsw(a, b) simde_mm_min_pi16(a, b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_min_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_min_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_FAST_NANS) && defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vminq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_pmin(b_.wasm_v128, a_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
      #if defined(SIMDE_FAST_NANS)
        r_.altivec_f32 = vec_min(a_.altivec_f32, b_.altivec_f32);
      #else
        r_.altivec_f32 = vec_sel(b_.altivec_f32, a_.altivec_f32, vec_cmpgt(b_.altivec_f32, a_.altivec_f32));
      #endif
    #elif defined(SIMDE_FAST_NANS) && defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_f32 = __lsx_vfmin_s(a_.lsx_f32, b_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      uint32_t SIMDE_VECTOR(16) m = HEDLEY_REINTERPRET_CAST(__typeof__(m), a_.f32 < b_.f32);
      r_.f32 =
        HEDLEY_REINTERPRET_CAST(
          __typeof__(r_.f32),
          ( (HEDLEY_REINTERPRET_CAST(__typeof__(m), a_.f32) &  m) |
            (HEDLEY_REINTERPRET_CAST(__typeof__(m), b_.f32) & ~m)
          )
        );
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = (a_.f32[i] < b_.f32[i]) ? a_.f32[i] : b_.f32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_min_ps(a, b) simde_mm_min_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_min_pu8 (simde__m64 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_min_pu8(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_u8 = vmin_u8(a_.neon_u8, b_.neon_u8);
    #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]) ? a_.u8[i] : b_.u8[i];
      }
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#define simde_m_pminub(a, b) simde_mm_min_pu8(a, b)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_min_pu8(a, b) simde_mm_min_pu8(a, b)
#  define _m_pminub(a, b) simde_mm_min_pu8(a, b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_min_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_min_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_min_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_min_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32_t value = vgetq_lane_f32(vminq_f32(a_.neon_f32, b_.neon_f32), 0);
      r_.neon_f32 = vsetq_lane_f32(value, a_.neon_f32, 0);
    #else
      r_.f32[0] = (a_.f32[0] < b_.f32[0]) ? a_.f32[0] : b_.f32[0];
      r_.f32[1] = a_.f32[1];
      r_.f32[2] = a_.f32[2];
      r_.f32[3] = a_.f32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_min_ss(a, b) simde_mm_min_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_movehl_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_movehl_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32x2_t a32 = vget_high_f32(a_.neon_f32);
      float32x2_t b32 = vget_high_f32(b_.neon_f32);
      r_.neon_f32 = vcombine_f32(b32, a32);
    #elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float),
          vec_mergel(b_.altivec_i64, a_.altivec_i64));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vilvh_d(a_.lsx_i64, b_.lsx_i64);
    #elif defined(SIMDE_SHUFFLE_VECTOR_)
      r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 6, 7, 2, 3);
    #else
      r_.f32[0] = b_.f32[2];
      r_.f32[1] = b_.f32[3];
      r_.f32[2] = a_.f32[2];
      r_.f32[3] = a_.f32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_movehl_ps(a, b) simde_mm_movehl_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_movelh_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_movelh_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32x2_t a10 = vget_low_f32(a_.neon_f32);
      float32x2_t b10 = vget_low_f32(b_.neon_f32);
      r_.neon_f32 = vcombine_f32(a10, b10);
    #elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_13_NATIVE)
      r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float),
          vec_mergeh(a_.altivec_i64, b_.altivec_i64));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vilvl_d(b_.lsx_i64, a_.lsx_i64);
    #elif defined(SIMDE_SHUFFLE_VECTOR_)
      r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 0, 1, 4, 5);
    #else
      r_.f32[0] = a_.f32[0];
      r_.f32[1] = a_.f32[1];
      r_.f32[2] = b_.f32[0];
      r_.f32[3] = b_.f32[1];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_movelh_ps(a, b) simde_mm_movelh_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_movemask_pi8 (simde__m64 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_movemask_pi8(a);
  #else
    simde__m64_private a_ = simde__m64_to_private(a);
    int r = 0;

    #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
      uint8x8_t input = a_.neon_u8;
      const int8_t xr[8] = {-7, -6, -5, -4, -3, -2, -1, 0};
      const uint8x8_t mask_and = vdup_n_u8(0x80);
      const int8x8_t mask_shift = vld1_s8(xr);
      const uint8x8_t mask_result = vshl_u8(vand_u8(input, mask_and), mask_shift);
      uint8x8_t lo = mask_result;
      r = vaddv_u8(lo);
    #else
      const size_t nmemb = sizeof(a_.i8) / sizeof(a_.i8[0]);
      SIMDE_VECTORIZE_REDUCTION(|:r)
      for (size_t i = 0 ; i < nmemb ; i++) {
        r |= (a_.u8[nmemb - 1 - i] >> 7) << (nmemb - 1 - i);
      }
    #endif

    return r;
  #endif
}
#define simde_m_pmovmskb(a) simde_mm_movemask_pi8(a)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_movemask_pi8(a) simde_mm_movemask_pi8(a)
#  define _m_pmovmskb(a) simde_mm_movemask_pi8(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_movemask_ps (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_movemask_ps(a);
  #else
    int r = 0;
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      // Shift out everything but the sign bits with a 32-bit unsigned shift right.
      uint64x2_t high_bits = vreinterpretq_u64_u32(vshrq_n_u32(a_.neon_u32, 31));
      // Merge the two pairs together with a 64-bit unsigned shift right + add.
      uint8x16_t paired = vreinterpretq_u8_u64(vsraq_n_u64(high_bits, high_bits, 31));
      // Extract the result.
      return vgetq_lane_u8(paired, 0) | (vgetq_lane_u8(paired, 8) << 2);
    #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      static const uint32_t md[4] = {
        1 << 0, 1 << 1, 1 << 2, 1 << 3
      };

      uint32x4_t extended = vreinterpretq_u32_s32(vshrq_n_s32(a_.neon_i32, 31));
      uint32x4_t masked = vandq_u32(vld1q_u32(md), extended);
      #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
        return HEDLEY_STATIC_CAST(int32_t, vaddvq_u32(masked));
      #else
        uint64x2_t t64 = vpaddlq_u32(masked);
        return
          HEDLEY_STATIC_CAST(int, vgetq_lane_u64(t64, 0)) +
          HEDLEY_STATIC_CAST(int, vgetq_lane_u64(t64, 1));
      #endif
    #elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE) && defined(SIMDE_BUG_CLANG_50932)
      SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) idx = { 96, 64, 32, 0, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
      SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) res = HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(unsigned char), vec_bperm(HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(unsigned __int128), a_.altivec_u64), idx));
      return HEDLEY_STATIC_CAST(int32_t, vec_extract(HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(signed int), res), 2));
    #elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
      SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) idx = { 96, 64, 32, 0, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
      SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) res = vec_bperm(a_.altivec_u8, idx);
      return HEDLEY_STATIC_CAST(int32_t, vec_extract(HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(signed int), res), 2));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      v2i64 t64 = __lsx_vmskltz_w(a_.lsx_i64);
      r = __lsx_vpickve2gr_wu(t64, 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      return HEDLEY_STATIC_CAST(int32_t, wasm_i32x4_bitmask(a_.wasm_v128));
    #else
      SIMDE_VECTORIZE_REDUCTION(|:r)
      for (size_t i = 0 ; i < sizeof(a_.u32) / sizeof(a_.u32[0]) ; i++) {
        r |= (a_.u32[i] >> ((sizeof(a_.u32[i]) * CHAR_BIT) - 1)) << i;
      }
    #endif

    return r;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_movemask_ps(a) simde_mm_movemask_ps((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_mul_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_mul_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vmulq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_mul(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.f32 = a_.f32 * b_.f32;
    #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
      r_.altivec_f32 = vec_mul(a_.altivec_f32, b_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_f32 = __lsx_vfmul_s(a_.lsx_f32, b_.lsx_f32);
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = a_.f32[i] * b_.f32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_mul_ps(a, b) simde_mm_mul_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_mul_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_mul_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_mul_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_mul_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    r_.f32[0] = a_.f32[0] * b_.f32[0];
    r_.f32[1] = a_.f32[1];
    r_.f32[2] = a_.f32[2];
    r_.f32[3] = a_.f32[3];

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_mul_ss(a, b) simde_mm_mul_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_mulhi_pu16 (simde__m64 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_mulhi_pu16(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)
      const uint32x4_t t1 = vmull_u16(a_.neon_u16, b_.neon_u16);
      const uint32x4_t t2 = vshrq_n_u32(t1, 16);
      const uint16x4_t t3 = vmovn_u32(t2);
      r_.neon_u16 = t3;
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.u16) / sizeof(r_.u16[0])) ; i++) {
        r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t, ((HEDLEY_STATIC_CAST(uint32_t, a_.u16[i]) * HEDLEY_STATIC_CAST(uint32_t, b_.u16[i])) >> UINT32_C(16)));
      }
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#define simde_m_pmulhuw(a, b) simde_mm_mulhi_pu16(a, b)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_mulhi_pu16(a, b) simde_mm_mulhi_pu16(a, b)
#  define _m_pmulhuw(a, b) simde_mm_mulhi_pu16(a, b)
#endif

#if defined(SIMDE_X86_SSE_NATIVE) && defined(HEDLEY_GCC_VERSION)
  #define SIMDE_MM_HINT_NTA  HEDLEY_STATIC_CAST(enum _mm_hint, 0)
  #define SIMDE_MM_HINT_T0   HEDLEY_STATIC_CAST(enum _mm_hint, 1)
  #define SIMDE_MM_HINT_T1   HEDLEY_STATIC_CAST(enum _mm_hint, 2)
  #define SIMDE_MM_HINT_T2   HEDLEY_STATIC_CAST(enum _mm_hint, 3)
  #define SIMDE_MM_HINT_ENTA HEDLEY_STATIC_CAST(enum _mm_hint, 4)
  #define SIMDE_MM_HINT_ET0  HEDLEY_STATIC_CAST(enum _mm_hint, 5)
  #define SIMDE_MM_HINT_ET1  HEDLEY_STATIC_CAST(enum _mm_hint, 6)
  #define SIMDE_MM_HINT_ET2  HEDLEY_STATIC_CAST(enum _mm_hint, 7)
#else
  #define SIMDE_MM_HINT_NTA  0
  #define SIMDE_MM_HINT_T0   1
  #define SIMDE_MM_HINT_T1   2
  #define SIMDE_MM_HINT_T2   3
  #define SIMDE_MM_HINT_ENTA 4
  #define SIMDE_MM_HINT_ET0  5
  #define SIMDE_MM_HINT_ET1  6
  #define SIMDE_MM_HINT_ET2  7
#endif

#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  HEDLEY_DIAGNOSTIC_PUSH
  #if HEDLEY_HAS_WARNING("-Wreserved-id-macro")
    _Pragma("clang diagnostic ignored \"-Wreserved-id-macro\"")
  #endif
  #undef  _MM_HINT_NTA
  #define _MM_HINT_NTA  SIMDE_MM_HINT_NTA
  #undef  _MM_HINT_T0
  #define _MM_HINT_T0   SIMDE_MM_HINT_T0
  #undef  _MM_HINT_T1
  #define _MM_HINT_T1   SIMDE_MM_HINT_T1
  #undef  _MM_HINT_T2
  #define _MM_HINT_T2   SIMDE_MM_HINT_T2
  #undef  _MM_HINT_ENTA
  #define _MM_HINT_ETNA SIMDE_MM_HINT_ENTA
  #undef  _MM_HINT_ET0
  #define _MM_HINT_ET0  SIMDE_MM_HINT_ET0
  #undef  _MM_HINT_ET1
  #define _MM_HINT_ET1  SIMDE_MM_HINT_ET1
  #undef  _MM_HINT_ET1
  #define _MM_HINT_ET2  SIMDE_MM_HINT_ET2
  HEDLEY_DIAGNOSTIC_POP
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_prefetch (const void* p, int i) {
  #if \
      HEDLEY_HAS_BUILTIN(__builtin_prefetch) || \
      HEDLEY_GCC_VERSION_CHECK(3,4,0) || \
      HEDLEY_INTEL_VERSION_CHECK(13,0,0)
    switch(i) {
      case SIMDE_MM_HINT_NTA:
        __builtin_prefetch(p, 0, 0);
        break;
      case SIMDE_MM_HINT_T0:
        __builtin_prefetch(p, 0, 3);
        break;
      case SIMDE_MM_HINT_T1:
        __builtin_prefetch(p, 0, 2);
        break;
      case SIMDE_MM_HINT_T2:
        __builtin_prefetch(p, 0, 1);
        break;
      case SIMDE_MM_HINT_ENTA:
        __builtin_prefetch(p, 1, 0);
        break;
      case SIMDE_MM_HINT_ET0:
        __builtin_prefetch(p, 1, 3);
        break;
      case SIMDE_MM_HINT_ET1:
        __builtin_prefetch(p, 1, 2);
        break;
      case SIMDE_MM_HINT_ET2:
        __builtin_prefetch(p, 0, 1);
        break;
    }
  #elif defined(__ARM_ACLE)
    #if (__ARM_ACLE >= 101)
      switch(i) {
        case SIMDE_MM_HINT_NTA:
          __pldx(0, 0, 1, p);
          break;
        case SIMDE_MM_HINT_T0:
          __pldx(0, 0, 0, p);
          break;
        case SIMDE_MM_HINT_T1:
          __pldx(0, 1, 0, p);
          break;
        case SIMDE_MM_HINT_T2:
          __pldx(0, 2, 0, p);
          break;
        case SIMDE_MM_HINT_ENTA:
          __pldx(1, 0, 1, p);
          break;
        case SIMDE_MM_HINT_ET0:
          __pldx(1, 0, 0, p);
          break;
        case SIMDE_MM_HINT_ET1:
          __pldx(1, 1, 0, p);
          break;
        case SIMDE_MM_HINT_ET2:
          __pldx(1, 2, 0, p);
          break;
      }
    #else
      (void) i;
      __pld(p)
    #endif
  #elif HEDLEY_PGI_VERSION_CHECK(10,0,0)
    (void) i;
    #pragma mem prefetch p
  #elif HEDLEY_CRAY_VERSION_CHECK(8,1,0)
    switch (i) {
      case SIMDE_MM_HINT_NTA:
        #pragma _CRI prefetch (nt) p
        break;
      case SIMDE_MM_HINT_T0:
      case SIMDE_MM_HINT_T1:
      case SIMDE_MM_HINT_T2:
        #pragma _CRI prefetch p
        break;
      case SIMDE_MM_HINT_ENTA:
        #pragma _CRI prefetch (write, nt) p
        break;
      case SIMDE_MM_HINT_ET0:
      case SIMDE_MM_HINT_ET1:
      case SIMDE_MM_HINT_ET2:
        #pragma _CRI prefetch (write) p
        break;
    }
  #elif HEDLEY_IBM_VERSION_CHECK(11,0,0)
    switch(i) {
      case SIMDE_MM_HINT_NTA:
        __prefetch_by_load(p, 0, 0);
        break;
      case SIMDE_MM_HINT_T0:
        __prefetch_by_load(p, 0, 3);
        break;
      case SIMDE_MM_HINT_T1:
        __prefetch_by_load(p, 0, 2);
        break;
      case SIMDE_MM_HINT_T2:
        __prefetch_by_load(p, 0, 1);
        break;
      case SIMDE_MM_HINT_ENTA:
        __prefetch_by_load(p, 1, 0);
        break;
      case SIMDE_MM_HINT_ET0:
        __prefetch_by_load(p, 1, 3);
        break;
      case SIMDE_MM_HINT_ET1:
        __prefetch_by_load(p, 1, 2);
        break;
      case SIMDE_MM_HINT_ET2:
        __prefetch_by_load(p, 0, 1);
        break;
    }
  #elif HEDLEY_MSVC_VERSION
    (void) i;
    (void) p;
  #endif
}
#if defined(SIMDE_X86_SSE_NATIVE)
  #if defined(__clang__) && !SIMDE_DETECT_CLANG_VERSION_CHECK(10,0,0) /* https://reviews.llvm.org/D71718 */
    #define simde_mm_prefetch(p, i) \
      (__extension__({ \
        HEDLEY_DIAGNOSTIC_PUSH \
        HEDLEY_DIAGNOSTIC_DISABLE_CAST_QUAL \
        _mm_prefetch((p), (i)); \
        HEDLEY_DIAGNOSTIC_POP \
      }))
  #else
    #define simde_mm_prefetch(p, i) _mm_prefetch(p, i)
  #endif
#endif
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
  #define _mm_prefetch(p, i) simde_mm_prefetch(p, i)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_negate_ps(simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return simde_mm_xor_ps(a, _mm_set1_ps(SIMDE_FLOAT32_C(-0.0)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vnegq_f32(a_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_neg(a_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
      r_.altivec_f32 = vec_neg(a_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      const v4f32 f32 = {0.0f, 0.0f, 0.0f, 0.0f};
      r_.lsx_f32 = __lsx_vfsub_s(f32, a_.lsx_f32);
    #elif defined(SIMDE_VECTOR_NEGATE)
      r_.f32 = -a_.f32;
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = -a_.f32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_rcp_ps (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_rcp_ps(a);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32x4_t recip = vrecpeq_f32(a_.neon_f32);

      #if SIMDE_ACCURACY_PREFERENCE > 0
        for (int i = 0; i < SIMDE_ACCURACY_PREFERENCE ; ++i) {
          recip = vmulq_f32(recip, vrecpsq_f32(recip, a_.neon_f32));
        }
      #endif

      r_.neon_f32 = recip;
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_div(simde_mm_set1_ps(1.0f), a_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = vec_re(a_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_f32 = __lsx_vfrecip_s(a_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
      r_.f32 = 1.0f / a_.f32;
    #elif defined(SIMDE_IEEE754_STORAGE)
      /* https://stackoverflow.com/questions/12227126/division-as-multiply-and-lut-fast-float-division-reciprocal/12228234#12228234 */
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        int32_t ix;
        simde_float32 fx = a_.f32[i];
        simde_memcpy(&ix, &fx, sizeof(ix));
        int32_t x = INT32_C(0x7EF311C3) - ix;
        simde_float32 temp;
        simde_memcpy(&temp, &x, sizeof(temp));
        r_.f32[i] = temp * (SIMDE_FLOAT32_C(2.0) - temp * fx);
      }
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = 1.0f / a_.f32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_rcp_ps(a) simde_mm_rcp_ps((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_rcp_ss (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_rcp_ss(a);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_rcp_ps(a));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_rcp_ps(simde_x_mm_broadcastlow_ps(a)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    r_.f32[0] = 1.0f / a_.f32[0];
    r_.f32[1] = a_.f32[1];
    r_.f32[2] = a_.f32[2];
    r_.f32[3] = a_.f32[3];

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_rcp_ss(a) simde_mm_rcp_ss((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_rsqrt_ps (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_rsqrt_ps(a);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vrsqrteq_f32(a_.neon_f32);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = vec_rsqrte(a_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_f32 = __lsx_vfrsqrt_s(a_.lsx_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_div(simde_mm_set1_ps(1.0f), wasm_f32x4_sqrt(a_.wasm_v128));
    #elif defined(SIMDE_IEEE754_STORAGE)
      /* https://basesandframes.files.wordpress.com/2020/04/even_faster_math_functions_green_2020.pdf
        Pages 100 - 103 */
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        #if SIMDE_ACCURACY_PREFERENCE <= 0
          r_.i32[i] = INT32_C(0x5F37624F) - (a_.i32[i] >> 1);
        #else
          simde_float32 x = a_.f32[i];
          simde_float32 xhalf = SIMDE_FLOAT32_C(0.5) * x;
          int32_t ix;

          simde_memcpy(&ix, &x, sizeof(ix));

          #if SIMDE_ACCURACY_PREFERENCE == 1
            ix = INT32_C(0x5F375A82) - (ix >> 1);
          #else
            ix = INT32_C(0x5F37599E) - (ix >> 1);
          #endif

          simde_memcpy(&x, &ix, sizeof(x));

          #if SIMDE_ACCURACY_PREFERENCE >= 2
            x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
          #endif
          x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);

          r_.f32[i] = x;
        #endif
      }
    #elif defined(simde_math_sqrtf)
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = 1.0f / simde_math_sqrtf(a_.f32[i]);
      }
    #else
      HEDLEY_UNREACHABLE();
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_rsqrt_ps(a) simde_mm_rsqrt_ps((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_rsqrt_ss (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_rsqrt_ss(a);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_rsqrt_ps(a));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_rsqrt_ps(simde_x_mm_broadcastlow_ps(a)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

  #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vsetq_lane_f32(vgetq_lane_f32(simde_mm_rsqrt_ps(a).neon_f32, 0), a_.neon_f32, 0);
  #elif defined(SIMDE_IEEE754_STORAGE)
    {
      #if SIMDE_ACCURACY_PREFERENCE <= 0
        r_.i32[0] = INT32_C(0x5F37624F) - (a_.i32[0] >> 1);
      #else
        simde_float32 x = a_.f32[0];
        simde_float32 xhalf = SIMDE_FLOAT32_C(0.5) * x;
        int32_t ix;

        simde_memcpy(&ix, &x, sizeof(ix));

        #if SIMDE_ACCURACY_PREFERENCE == 1
          ix = INT32_C(0x5F375A82) - (ix >> 1);
        #else
          ix = INT32_C(0x5F37599E) - (ix >> 1);
        #endif

        simde_memcpy(&x, &ix, sizeof(x));

        #if SIMDE_ACCURACY_PREFERENCE >= 2
          x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
        #endif
        x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);

        r_.f32[0] = x;
      #endif
    }
    r_.f32[1] = a_.f32[1];
    r_.f32[2] = a_.f32[2];
    r_.f32[3] = a_.f32[3];
  #elif defined(simde_math_sqrtf)
    r_.f32[0] = 1.0f / simde_math_sqrtf(a_.f32[0]);
    r_.f32[1] = a_.f32[1];
    r_.f32[2] = a_.f32[2];
    r_.f32[3] = a_.f32[3];
  #else
    HEDLEY_UNREACHABLE();
  #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_rsqrt_ss(a) simde_mm_rsqrt_ss((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_sad_pu8 (simde__m64 a, simde__m64 b) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    return _mm_sad_pu8(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)
      uint64x1_t t = vpaddl_u32(vpaddl_u16(vpaddl_u8(vabd_u8(a_.neon_u8, b_.neon_u8))));
      r_.neon_u16 = vset_lane_u16(HEDLEY_STATIC_CAST(uint64_t, vget_lane_u64(t, 0)), vdup_n_u16(0), 0);
    #else
      uint16_t sum = 0;

      SIMDE_VECTORIZE_REDUCTION(+:sum)
      for (size_t i = 0 ; i < (sizeof(r_.u8) / sizeof(r_.u8[0])) ; i++) {
        sum += HEDLEY_STATIC_CAST(uint8_t, simde_math_abs(a_.u8[i] - b_.u8[i]));
      }

      r_.i16[0] = HEDLEY_STATIC_CAST(int16_t, sum);
      r_.i16[1] = 0;
      r_.i16[2] = 0;
      r_.i16[3] = 0;
    #endif

    return simde__m64_from_private(r_);
  #endif
}
#define simde_m_psadbw(a, b) simde_mm_sad_pu8(a, b)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_sad_pu8(a, b) simde_mm_sad_pu8(a, b)
#  define _m_psadbw(a, b) simde_mm_sad_pu8(a, b)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_set_ss (simde_float32 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_set_ss(a);
  #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
    return vsetq_lane_f32(a, vdupq_n_f32(SIMDE_FLOAT32_C(0.0)), 0);
  #else
    return simde_mm_set_ps(SIMDE_FLOAT32_C(0.0), SIMDE_FLOAT32_C(0.0), SIMDE_FLOAT32_C(0.0), a);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_set_ss(a) simde_mm_set_ss(a)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_setr_ps (simde_float32 e3, simde_float32 e2, simde_float32 e1, simde_float32 e0) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_setr_ps(e3, e2, e1, e0);
  #else
    return simde_mm_set_ps(e0, e1, e2, e3);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_setr_ps(e3, e2, e1, e0) simde_mm_setr_ps(e3, e2, e1, e0)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_setzero_ps (void) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_setzero_ps();
  #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
    return vdupq_n_f32(SIMDE_FLOAT32_C(0.0));
  #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
    return vec_splats(SIMDE_FLOAT32_C(0.0));
  #elif defined(SIMDE_WASM_SIMD128_NATIVE)
    return wasm_f32x4_const(0.f, 0.f, 0.f, 0.f);
  #else
    simde__m128 r;
    simde_memset(&r, 0, sizeof(r));
    return r;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_setzero_ps() simde_mm_setzero_ps()
#endif

#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
HEDLEY_DIAGNOSTIC_PUSH
SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_undefined_ps (void) {
  simde__m128_private r_;

  #if defined(SIMDE_HAVE_UNDEFINED128)
    r_.n = _mm_undefined_ps();
  #elif !defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
    r_ = simde__m128_to_private(simde_mm_setzero_ps());
  #endif

  return simde__m128_from_private(r_);
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_undefined_ps() simde_mm_undefined_ps()
#endif

#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
HEDLEY_DIAGNOSTIC_POP
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_x_mm_setone_ps (void) {
  simde__m128 t = simde_mm_setzero_ps();
  return simde_mm_cmpeq_ps(t, t);
}

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_sfence (void) {
    /* TODO: Use Hedley. */
  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_sfence();
  #elif defined(__GNUC__) && ((__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 7))
    __atomic_thread_fence(__ATOMIC_SEQ_CST);
  #elif !defined(__INTEL_COMPILER) && defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L) && !defined(__STDC_NO_ATOMICS__)
    #if defined(__GNUC__) && (__GNUC__ == 4) && (__GNUC_MINOR__ < 9)
      __atomic_thread_fence(__ATOMIC_SEQ_CST);
    #else
      atomic_thread_fence(memory_order_seq_cst);
    #endif
  #elif defined(_MSC_VER)
    MemoryBarrier();
  #elif HEDLEY_HAS_EXTENSION(c_atomic)
    __c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
  #elif defined(__GNUC__) && ((__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 1))
    __sync_synchronize();
  #elif defined(_OPENMP)
    #pragma omp critical(simde_mm_sfence_)
    { }
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_sfence() simde_mm_sfence()
#endif

#define SIMDE_MM_SHUFFLE(z, y, x, w) (((z) << 6) | ((y) << 4) | ((x) << 2) | (w))
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _MM_SHUFFLE(z, y, x, w) SIMDE_MM_SHUFFLE(z, y, x, w)
#endif

#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
#  define simde_mm_shuffle_pi16(a, imm8) _mm_shuffle_pi16(a, imm8)
#elif defined(SIMDE_SHUFFLE_VECTOR_)
#  define simde_mm_shuffle_pi16(a, imm8) (__extension__ ({ \
      const simde__m64_private simde_tmp_a_ = simde__m64_to_private(a); \
      simde__m64_from_private((simde__m64_private) { .i16 = \
        SIMDE_SHUFFLE_VECTOR_(16, 8, \
          (simde_tmp_a_).i16, \
          (simde_tmp_a_).i16, \
          (((imm8)     ) & 3), \
          (((imm8) >> 2) & 3), \
          (((imm8) >> 4) & 3), \
          (((imm8) >> 6) & 3)) }); }))
#else
SIMDE_FUNCTION_ATTRIBUTES
simde__m64
simde_mm_shuffle_pi16 (simde__m64 a, const int imm8)
    SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255) {
  simde__m64_private r_;
  simde__m64_private a_ = simde__m64_to_private(a);

  for (size_t i = 0 ; i < sizeof(r_.i16) / sizeof(r_.i16[0]) ; i++) {
    r_.i16[i] = a_.i16[(imm8 >> (i * 2)) & 3];
  }

HEDLEY_DIAGNOSTIC_PUSH
#if HEDLEY_HAS_WARNING("-Wconditional-uninitialized")
#  pragma clang diagnostic ignored "-Wconditional-uninitialized"
#endif
  return simde__m64_from_private(r_);
HEDLEY_DIAGNOSTIC_POP
}
#endif
#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
#  define simde_m_pshufw(a, imm8) _m_pshufw(a, imm8)
#else
#  define simde_m_pshufw(a, imm8) simde_mm_shuffle_pi16(a, imm8)
#endif
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_shuffle_pi16(a, imm8) simde_mm_shuffle_pi16(a, imm8)
#  define _m_pshufw(a, imm8) simde_mm_shuffle_pi16(a, imm8)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_shuffle_ps (simde__m128 a, simde__m128 b, const int imm8)
    SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255) {
  simde__m128_private
    r_,
    a_ = simde__m128_to_private(a),
    b_ = simde__m128_to_private(b);

  r_.f32[0] = a_.f32[(imm8 >> 0) & 3];
  r_.f32[1] = a_.f32[(imm8 >> 2) & 3];
  r_.f32[2] = b_.f32[(imm8 >> 4) & 3];
  r_.f32[3] = b_.f32[(imm8 >> 6) & 3];

  return simde__m128_from_private(r_);
}
#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
#  define simde_mm_shuffle_ps(a, b, imm8) _mm_shuffle_ps(a, b, imm8)
#elif defined(SIMDE_WASM_SIMD128_NATIVE)
  #define simde_mm_shuffle_ps(a, b, imm8) (__extension__ ({ \
    simde__m128_from_private((simde__m128_private) { .wasm_v128 = \
      wasm_i32x4_shuffle( \
        simde__m128_to_private(a).wasm_v128, \
        simde__m128_to_private(b).wasm_v128, \
        (((imm8)     ) & 3), \
        (((imm8) >> 2) & 3), \
        (((imm8) >> 4) & 3) + 4, \
        (((imm8) >> 6) & 3) + 4) }); }))
#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_STATEMENT_EXPR_)
  #define simde_mm_shuffle_ps(a, b, imm8) \
    (__extension__({ \
      float32x4_t simde_mm_shuffle_ps_a_ = simde__m128_to_neon_f32(a); \
      float32x4_t simde_mm_shuffle_ps_b_ = simde__m128_to_neon_f32(b); \
      float32x4_t simde_mm_shuffle_ps_r_; \
      \
      simde_mm_shuffle_ps_r_ = vmovq_n_f32(vgetq_lane_f32(simde_mm_shuffle_ps_a_, (imm8) &&nbsp;(0x3))); \
      simde_mm_shuffle_ps_r_ = vsetq_lane_f32(vgetq_lane_f32(simde_mm_shuffle_ps_a_, ((imm8) >> 2) & 0x3), simde_mm_shuffle_ps_r_, 1); \
      simde_mm_shuffle_ps_r_ = vsetq_lane_f32(vgetq_lane_f32(simde_mm_shuffle_ps_b_, ((imm8) >> 4) & 0x3), simde_mm_shuffle_ps_r_, 2); \
                               vsetq_lane_f32(vgetq_lane_f32(simde_mm_shuffle_ps_b_, ((imm8) >> 6) & 0x3), simde_mm_shuffle_ps_r_, 3); \
    }))
#elif defined(SIMDE_SHUFFLE_VECTOR_)
  #define simde_mm_shuffle_ps(a, b, imm8) (__extension__ ({ \
      simde__m128_from_private((simde__m128_private) { .f32 = \
        SIMDE_SHUFFLE_VECTOR_(32, 16, \
          simde__m128_to_private(a).f32, \
          simde__m128_to_private(b).f32, \
          (((imm8)     ) & 3), \
          (((imm8) >> 2) & 3), \
          (((imm8) >> 4) & 3) + 4, \
          (((imm8) >> 6) & 3) + 4) }); }))
#endif
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_shuffle_ps(a, b, imm8) simde_mm_shuffle_ps((a), (b), imm8)
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_sqrt_ps (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_sqrt_ps(a);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
      r_.neon_f32 = vsqrtq_f32(a_.neon_f32);
    #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32x4_t est = vrsqrteq_f32(a_.neon_f32);
      for (int i = 0 ; i <= SIMDE_ACCURACY_PREFERENCE ; i++) {
        est = vmulq_f32(vrsqrtsq_f32(vmulq_f32(a_.neon_f32, est), est), est);
      }
      r_.neon_f32 = vmulq_f32(a_.neon_f32, est);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_sqrt(a_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) || defined(SIMDE_ZARCH_ZVECTOR_14_NATIVE)
      r_.altivec_f32 = vec_sqrt(a_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_f32 = __lsx_vfsqrt_s(a_.lsx_f32);
    #elif defined(simde_math_sqrt)
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < sizeof(r_.f32) / sizeof(r_.f32[0]) ; i++) {
        r_.f32[i] = simde_math_sqrtf(a_.f32[i]);
      }
    #else
      HEDLEY_UNREACHABLE();
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_sqrt_ps(a) simde_mm_sqrt_ps((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_sqrt_ss (simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_sqrt_ss(a);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_sqrt_ps(a));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_sqrt_ps(simde_x_mm_broadcastlow_ps(a)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32_t value =
            vgetq_lane_f32(simde__m128_to_private(simde_mm_sqrt_ps(a)).neon_f32, 0);
      r_.neon_f32 = vsetq_lane_f32(value, a_.neon_f32, 0);
    #elif defined(simde_math_sqrtf)
      r_.f32[0] = simde_math_sqrtf(a_.f32[0]);
      r_.f32[1] = a_.f32[1];
      r_.f32[2] = a_.f32[2];
      r_.f32[3] = a_.f32[3];
    #else
      HEDLEY_UNREACHABLE();
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_sqrt_ss(a) simde_mm_sqrt_ss((a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_store_ps (simde_float32 mem_addr[4], simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_store_ps(mem_addr, a);
  #else
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      vst1q_f32(mem_addr, a_.neon_f32);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      vec_st(a_.altivec_f32, 0, mem_addr);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      wasm_v128_store(mem_addr, a_.wasm_v128);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      __lsx_vst(a_.lsx_f32, mem_addr, 0);
    #else
      simde_memcpy(mem_addr, &a_, sizeof(a));
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_store_ps(mem_addr, a) simde_mm_store_ps(SIMDE_CHECKED_REINTERPRET_CAST(float*, simde_float32*, mem_addr), (a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_store1_ps (simde_float32 mem_addr[4], simde__m128 a) {
  simde_float32* mem_addr_ = SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m128);

  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_store_ps1(mem_addr_, a);
  #else
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      vst1q_f32(mem_addr_, vdupq_lane_f32(vget_low_f32(a_.neon_f32), 0));
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      wasm_v128_store(mem_addr_, wasm_i32x4_shuffle(a_.wasm_v128, a_.wasm_v128, 0, 0, 0, 0));
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      vec_st(vec_splat(a_.altivec_f32, 0), 0, mem_addr_);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      __lsx_vst(__lsx_vreplvei_w(a_.lsx_f32, 0), mem_addr_, 0);
    #elif defined(SIMDE_SHUFFLE_VECTOR_)
      simde__m128_private tmp_;
      tmp_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, a_.f32, 0, 0, 0, 0);
      simde_mm_store_ps(mem_addr_, tmp_.f32);
    #else
      SIMDE_VECTORIZE_ALIGNED(mem_addr_:16)
      for (size_t i = 0 ; i < sizeof(a_.f32) / sizeof(a_.f32[0]) ; i++) {
        mem_addr_[i] = a_.f32[0];
      }
    #endif
  #endif
}
#define simde_mm_store_ps1(mem_addr, a) simde_mm_store1_ps(mem_addr, a)
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_store_ps1(mem_addr, a) simde_mm_store1_ps(SIMDE_CHECKED_REINTERPRET_CAST(float*, simde_float32*, mem_addr), (a))
#  define _mm_store1_ps(mem_addr, a) simde_mm_store1_ps(SIMDE_CHECKED_REINTERPRET_CAST(float*, simde_float32*, mem_addr), (a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_store_ss (simde_float32* mem_addr, simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_store_ss(mem_addr, a);
  #else
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      vst1q_lane_f32(mem_addr, a_.neon_f32, 0);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      __lsx_vstelm_w(a_.lsx_f32, mem_addr, 0, 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      wasm_v128_store32_lane(HEDLEY_REINTERPRET_CAST(void*, mem_addr), a_.wasm_v128, 0);
    #else
      *mem_addr = a_.f32[0];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_store_ss(mem_addr, a) simde_mm_store_ss(SIMDE_CHECKED_REINTERPRET_CAST(float*, simde_float32*, mem_addr), (a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_storeh_pi (simde__m64* mem_addr, simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_storeh_pi(HEDLEY_REINTERPRET_CAST(__m64*, mem_addr), a);
  #else
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      vst1_f32(HEDLEY_REINTERPRET_CAST(float32_t*, mem_addr), vget_high_f32(a_.neon_f32));
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      wasm_v128_store64_lane(HEDLEY_REINTERPRET_CAST(void*, mem_addr), a_.wasm_v128, 1);
    #else
      simde_memcpy(mem_addr, &(a_.m64[1]), sizeof(a_.m64[1]));
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_storeh_pi(mem_addr, a) simde_mm_storeh_pi(mem_addr, (a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_storel_pi (simde__m64* mem_addr, simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_storel_pi(HEDLEY_REINTERPRET_CAST(__m64*, mem_addr), a);
  #elif defined(SIMDE_WASM_SIMD128_NATIVE)
    simde__m128_private a_ = simde__m128_to_private(a);
    wasm_v128_store64_lane(HEDLEY_REINTERPRET_CAST(void*, mem_addr), a_.wasm_v128, 0);
  #else
    simde__m64_private* dest_ = HEDLEY_REINTERPRET_CAST(simde__m64_private*, mem_addr);
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      dest_->neon_f32 = vget_low_f32(a_.neon_f32);
    #else
      dest_->f32[0] = a_.f32[0];
      dest_->f32[1] = a_.f32[1];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_storel_pi(mem_addr, a) simde_mm_storel_pi(mem_addr, (a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_storer_ps (simde_float32 mem_addr[4], simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_storer_ps(mem_addr, a);
  #else
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      vec_st(vec_reve(a_.altivec_f32), 0, mem_addr);
    #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32x4_t tmp = vrev64q_f32(a_.neon_f32);
      vst1q_f32(mem_addr, vextq_f32(tmp, tmp, 2));
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      __lsx_vst(__lsx_vshuf4i_w(a_.lsx_f32, 0x1b), mem_addr, 0);
    #elif defined(SIMDE_SHUFFLE_VECTOR_)
      a_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, a_.f32, 3, 2, 1, 0);
      simde_mm_store_ps(mem_addr, simde__m128_from_private(a_));
    #else
      SIMDE_VECTORIZE_ALIGNED(mem_addr:16)
      for (size_t i = 0 ; i < sizeof(a_.f32) / sizeof(a_.f32[0]) ; i++) {
        mem_addr[i] = a_.f32[((sizeof(a_.f32) / sizeof(a_.f32[0])) - 1) - i];
      }
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_storer_ps(mem_addr, a) simde_mm_storer_ps(SIMDE_CHECKED_REINTERPRET_CAST(float*, simde_float32*, mem_addr), (a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_storeu_ps (simde_float32 mem_addr[4], simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_storeu_ps(mem_addr, a);
  #else
    simde__m128_private a_ = simde__m128_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      vst1q_f32(mem_addr, a_.neon_f32);
    #elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
      vec_vsx_st(a_.altivec_f32, 0, mem_addr);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      __lsx_vst(a_.lsx_f32, mem_addr, 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      wasm_v128_store(mem_addr, a_.wasm_v128);
    #else
      simde_memcpy(mem_addr, &a_, sizeof(a_));
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_storeu_ps(mem_addr, a) simde_mm_storeu_ps(SIMDE_CHECKED_REINTERPRET_CAST(float*, simde_float32*, mem_addr), (a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_sub_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_sub_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      r_.neon_f32 = vsubq_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_f32x4_sub(a_.wasm_v128, b_.wasm_v128);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = vec_sub(a_.altivec_f32, b_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_f32 = __lsx_vfsub_s(a_.lsx_f32, b_.lsx_f32);
    #elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
      r_.f32 = a_.f32 - b_.f32;
    #else
      SIMDE_VECTORIZE
      for (size_t i = 0 ; i < (sizeof(r_.f32) / sizeof(r_.f32[0])) ; i++) {
        r_.f32[i] = a_.f32[i] - b_.f32[i];
      }
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_sub_ps(a, b) simde_mm_sub_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_sub_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_sub_ss(a, b);
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0) && defined(SIMDE_FAST_EXCEPTIONS)
    return simde_mm_move_ss(a, simde_mm_sub_ps(a, b));
  #elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
    return simde_mm_move_ss(a, simde_mm_sub_ps(simde_x_mm_broadcastlow_ps(a), simde_x_mm_broadcastlow_ps(b)));
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    r_.f32[0] = a_.f32[0] - b_.f32[0];
    r_.f32[1] = a_.f32[1];
    r_.f32[2] = a_.f32[2];
    r_.f32[3] = a_.f32[3];

    return simde__m128_from_private(r_);
  #endif
}

#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_sub_ss(a, b) simde_mm_sub_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_ucomieq_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_ucomieq_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);
    int r;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
      uint32x4_t a_eq_b = vceqq_f32(a_.neon_f32, b_.neon_f32);
      r = !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_eq_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r = wasm_f32x4_extract_lane(a_.wasm_v128, 0) == wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #elif defined(SIMDE_HAVE_FENV_H)
      fenv_t envp;
      int x = feholdexcept(&envp);
      r = a_.f32[0] == b_.f32[0];
      if (HEDLEY_LIKELY(x == 0))
        fesetenv(&envp);
    #else
      r = a_.f32[0] == b_.f32[0];
    #endif

    return r;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_ucomieq_ss(a, b) simde_mm_ucomieq_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_ucomige_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_ucomige_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);
    int r;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
      uint32x4_t a_ge_b = vcgeq_f32(a_.neon_f32, b_.neon_f32);
      r = !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_ge_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r = wasm_f32x4_extract_lane(a_.wasm_v128, 0) >= wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #elif defined(SIMDE_HAVE_FENV_H)
      fenv_t envp;
      int x = feholdexcept(&envp);
      r = a_.f32[0] >= b_.f32[0];
      if (HEDLEY_LIKELY(x == 0))
        fesetenv(&envp);
    #else
      r = a_.f32[0] >= b_.f32[0];
    #endif

    return r;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_ucomige_ss(a, b) simde_mm_ucomige_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_ucomigt_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_ucomigt_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);
    int r;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
      uint32x4_t a_gt_b = vcgtq_f32(a_.neon_f32, b_.neon_f32);
      r = !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_gt_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r = wasm_f32x4_extract_lane(a_.wasm_v128, 0) > wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #elif defined(SIMDE_HAVE_FENV_H)
      fenv_t envp;
      int x = feholdexcept(&envp);
      r = a_.f32[0] > b_.f32[0];
      if (HEDLEY_LIKELY(x == 0))
        fesetenv(&envp);
    #else
      r = a_.f32[0] > b_.f32[0];
    #endif

    return r;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_ucomigt_ss(a, b) simde_mm_ucomigt_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_ucomile_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_ucomile_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);
    int r;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
      uint32x4_t a_le_b = vcleq_f32(a_.neon_f32, b_.neon_f32);
      r = !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_le_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r = wasm_f32x4_extract_lane(a_.wasm_v128, 0) <= wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #elif defined(SIMDE_HAVE_FENV_H)
      fenv_t envp;
      int x = feholdexcept(&envp);
      r = a_.f32[0] <= b_.f32[0];
      if (HEDLEY_LIKELY(x == 0))
        fesetenv(&envp);
    #else
      r = a_.f32[0] <= b_.f32[0];
    #endif

    return r;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_ucomile_ss(a, b) simde_mm_ucomile_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_ucomilt_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_ucomilt_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);
    int r;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
      uint32x4_t a_lt_b = vcltq_f32(a_.neon_f32, b_.neon_f32);
      r = !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_lt_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r = wasm_f32x4_extract_lane(a_.wasm_v128, 0) < wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #elif defined(SIMDE_HAVE_FENV_H)
      fenv_t envp;
      int x = feholdexcept(&envp);
      r = a_.f32[0] < b_.f32[0];
      if (HEDLEY_LIKELY(x == 0))
        fesetenv(&envp);
    #else
      r = a_.f32[0] < b_.f32[0];
    #endif

    return r;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_ucomilt_ss(a, b) simde_mm_ucomilt_ss((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
int
simde_mm_ucomineq_ss (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_ucomineq_ss(a, b);
  #else
    simde__m128_private
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);
    int r;

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
      uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
      uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
      uint32x4_t a_neq_b = vmvnq_u32(vceqq_f32(a_.neon_f32, b_.neon_f32));
      r = !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_neq_b), 0) != 0);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r = wasm_f32x4_extract_lane(a_.wasm_v128, 0) != wasm_f32x4_extract_lane(b_.wasm_v128, 0);
    #elif defined(SIMDE_HAVE_FENV_H)
      fenv_t envp;
      int x = feholdexcept(&envp);
      r = a_.f32[0] != b_.f32[0];
      if (HEDLEY_LIKELY(x == 0))
        fesetenv(&envp);
    #else
      r = a_.f32[0] != b_.f32[0];
    #endif

    return r;
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_ucomineq_ss(a, b) simde_mm_ucomineq_ss((a), (b))
#endif

#if defined(SIMDE_X86_SSE_NATIVE)
#  if defined(__has_builtin)
#    if __has_builtin(__builtin_ia32_undef128)
#      define SIMDE_HAVE_UNDEFINED128
#    endif
#  elif !defined(__PGI) && !defined(SIMDE_BUG_GCC_REV_208793) && !defined(_MSC_VER)
#    define SIMDE_HAVE_UNDEFINED128
#  endif
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_unpackhi_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_unpackhi_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
      r_.neon_f32 = vzip2q_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32x2_t a1 = vget_high_f32(a_.neon_f32);
      float32x2_t b1 = vget_high_f32(b_.neon_f32);
      float32x2x2_t result = vzip_f32(a1, b1);
      r_.neon_f32 = vcombine_f32(result.val[0], result.val[1]);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vilvh_w(b_.lsx_i64, a_.lsx_i64);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_i32x4_shuffle(a_.wasm_v128, b_.wasm_v128, 2, 6, 3, 7);
    #elif defined(SIMDE_SHUFFLE_VECTOR_)
      r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 2, 6, 3, 7);
    #else
      r_.f32[0] = a_.f32[2];
      r_.f32[1] = b_.f32[2];
      r_.f32[2] = a_.f32[3];
      r_.f32[3] = b_.f32[3];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_unpackhi_ps(a, b) simde_mm_unpackhi_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
simde__m128
simde_mm_unpacklo_ps (simde__m128 a, simde__m128 b) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    return _mm_unpacklo_ps(a, b);
  #else
    simde__m128_private
      r_,
      a_ = simde__m128_to_private(a),
      b_ = simde__m128_to_private(b);

    #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
      r_.neon_f32 = vzip1q_f32(a_.neon_f32, b_.neon_f32);
    #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
      r_.altivec_f32 = vec_mergeh(a_.altivec_f32, b_.altivec_f32);
    #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
      r_.lsx_i64 = __lsx_vilvl_w(b_.lsx_i64, a_.lsx_i64);
    #elif defined(SIMDE_WASM_SIMD128_NATIVE)
      r_.wasm_v128 = wasm_i32x4_shuffle(a_.wasm_v128, b_.wasm_v128, 0, 4, 1, 5);
    #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      float32x2_t a1 = vget_low_f32(a_.neon_f32);
      float32x2_t b1 = vget_low_f32(b_.neon_f32);
      float32x2x2_t result = vzip_f32(a1, b1);
      r_.neon_f32 = vcombine_f32(result.val[0], result.val[1]);
    #elif defined(SIMDE_SHUFFLE_VECTOR_)
      r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 0, 4, 1, 5);
    #else
      r_.f32[0] = a_.f32[0];
      r_.f32[1] = b_.f32[0];
      r_.f32[2] = a_.f32[1];
      r_.f32[3] = b_.f32[1];
    #endif

    return simde__m128_from_private(r_);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_unpacklo_ps(a, b) simde_mm_unpacklo_ps((a), (b))
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_stream_pi (simde__m64* mem_addr, simde__m64 a) {
  #if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
    _mm_stream_pi(HEDLEY_REINTERPRET_CAST(__m64*, mem_addr), a);
  #else
    simde__m64_private*
      dest = HEDLEY_REINTERPRET_CAST(simde__m64_private*, mem_addr),
      a_ = simde__m64_to_private(a);

    #if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
      dest->i64[0] = vget_lane_s64(a_.neon_i64, 0);
    #else
      dest->i64[0] = a_.i64[0];
    #endif
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_stream_pi(mem_addr, a) simde_mm_stream_pi(mem_addr, (a))
#endif

SIMDE_FUNCTION_ATTRIBUTES
void
simde_mm_stream_ps (simde_float32 mem_addr[4], simde__m128 a) {
  #if defined(SIMDE_X86_SSE_NATIVE)
    _mm_stream_ps(mem_addr, a);
  #elif HEDLEY_HAS_BUILTIN(__builtin_nontemporal_store) && defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
    simde__m128_private a_ = simde__m128_to_private(a);
    __builtin_nontemporal_store(a_.f32, SIMDE_ALIGN_CAST(__typeof__(a_.f32)*, mem_addr));
  #else
    simde_mm_store_ps(mem_addr, a);
  #endif
}
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _mm_stream_ps(mem_addr, a) simde_mm_stream_ps(SIMDE_CHECKED_REINTERPRET_CAST(float*, simde_float32*, mem_addr), (a))
#endif

#if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && !defined(SIMDE_ARM_NEON_A64V8_NATIVE)
  #define SIMDE_MM_TRANSPOSE4_PS(row0, row1, row2, row3) \
    do { \
          float32x4x2_t SIMDE_MM_TRANSPOSE4_PS_ROW01 = vtrnq_f32(row0, row1); \
          float32x4x2_t SIMDE_MM_TRANSPOSE4_PS_ROW23 = vtrnq_f32(row2, row3); \
          row0 = vcombine_f32(vget_low_f32(SIMDE_MM_TRANSPOSE4_PS_ROW01.val[0]), \
                              vget_low_f32(SIMDE_MM_TRANSPOSE4_PS_ROW23.val[0])); \
          row1 = vcombine_f32(vget_low_f32(SIMDE_MM_TRANSPOSE4_PS_ROW01.val[1]), \
                              vget_low_f32(SIMDE_MM_TRANSPOSE4_PS_ROW23.val[1])); \
          row2 = vcombine_f32(vget_high_f32(SIMDE_MM_TRANSPOSE4_PS_ROW01.val[0]), \
                              vget_high_f32(SIMDE_MM_TRANSPOSE4_PS_ROW23.val[0])); \
          row3 = vcombine_f32(vget_high_f32(SIMDE_MM_TRANSPOSE4_PS_ROW01.val[1]), \
                              vget_high_f32(SIMDE_MM_TRANSPOSE4_PS_ROW23.val[1])); \
      } while (0)
#else
  #define SIMDE_MM_TRANSPOSE4_PS(row0, row1, row2, row3) \
    do { \
      simde__m128 SIMDE_MM_TRANSPOSE4_PS_tmp3, SIMDE_MM_TRANSPOSE4_PS_tmp2, SIMDE_MM_TRANSPOSE4_PS_tmp1, SIMDE_MM_TRANSPOSE4_PS_tmp0; \
      SIMDE_MM_TRANSPOSE4_PS_tmp0 = simde_mm_unpacklo_ps((row0), (row1)); \
      SIMDE_MM_TRANSPOSE4_PS_tmp2 = simde_mm_unpacklo_ps((row2), (row3)); \
      SIMDE_MM_TRANSPOSE4_PS_tmp1 = simde_mm_unpackhi_ps((row0), (row1)); \
      SIMDE_MM_TRANSPOSE4_PS_tmp3 = simde_mm_unpackhi_ps((row2), (row3)); \
      row0 = simde_mm_movelh_ps(SIMDE_MM_TRANSPOSE4_PS_tmp0, SIMDE_MM_TRANSPOSE4_PS_tmp2); \
      row1 = simde_mm_movehl_ps(SIMDE_MM_TRANSPOSE4_PS_tmp2, SIMDE_MM_TRANSPOSE4_PS_tmp0); \
      row2 = simde_mm_movelh_ps(SIMDE_MM_TRANSPOSE4_PS_tmp1, SIMDE_MM_TRANSPOSE4_PS_tmp3); \
      row3 = simde_mm_movehl_ps(SIMDE_MM_TRANSPOSE4_PS_tmp3, SIMDE_MM_TRANSPOSE4_PS_tmp1); \
    } while (0)
#endif
#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
#  define _MM_TRANSPOSE4_PS(row0, row1, row2, row3) SIMDE_MM_TRANSPOSE4_PS(row0, row1, row2, row3)
#endif

SIMDE_END_DECLS_

HEDLEY_DIAGNOSTIC_POP

#endif /* !defined(SIMDE_X86_SSE_H) */