llvm-project/clang/lib/Headers/smmintrin.h

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

2352 lines
96 KiB
C
Raw Normal View History

/*===---- smmintrin.h - SSE4 intrinsics ------------------------------------===
*
* Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
* See https://llvm.org/LICENSE.txt for license information.
* SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
*
*===-----------------------------------------------------------------------===
*/
#ifndef __SMMINTRIN_H
#define __SMMINTRIN_H
#if !defined(__i386__) && !defined(__x86_64__)
#error "This header is only meant to be used on x86 and x64 architecture"
#endif
#include <tmmintrin.h>
/* Define the default attributes for the functions in this file. */
[Builtins][Attributes][X86] Tag all X86 builtins with their required vector width. Add a min_vector_width function attribute and tag all x86 instrinsics with it This is part of an ongoing attempt at making 512 bit vectors illegal in the X86 backend type legalizer due to CPU frequency penalties associated with wide vectors on Skylake Server CPUs. We want the loop vectorizer to be able to emit IR containing wide vectors as intermediate operations in vectorized code and allow these wide vectors to be legalized to 256 bits by the X86 backend even though we are targetting a CPU that supports 512 bit vectors. This is similar to what happens with an AVX2 CPU, the vectorizer can emit wide vectors and the backend will split them. We want this splitting behavior, but still be able to use new Skylake instructions that work on 256-bit vectors and support things like masking and gather/scatter. Of course if the user uses explicit vector code in their source code we need to not split those operations. Especially if they have used any of the 512-bit vector intrinsics from immintrin.h. And we need to make it so that merely using the intrinsics produces the expected code in order to be backwards compatible. To support this goal, this patch adds a new IR function attribute "min-legal-vector-width" that can indicate the need for a minimum vector width to be legal in the backend. We need to ensure this attribute is set to the largest vector width needed by any intrinsics from immintrin.h that the function uses. The inliner will be reponsible for merging this attribute when a function is inlined. We may also need a way to limit inlining in the future as well, but we can discuss that in the future. To make things more complicated, there are two different ways intrinsics are implemented in immintrin.h. Either as an always_inline function containing calls to builtins(can be target specific or target independent) or vector extension code. Or as a macro wrapper around a taget specific builtin. I believe I've removed all cases where the macro was around a target independent builtin. To support the always_inline function case this patch adds attribute((min_vector_width(128))) that can be used to tag these functions with their vector width. All x86 intrinsic functions that operate on vectors have been tagged with this attribute. To support the macro case, all x86 specific builtins have also been tagged with the vector width that they require. Use of any builtin with this property will implicitly increase the min_vector_width of the function that calls it. I've done this as a new property in the attribute string for the builtin rather than basing it on the type string so that we can opt into it on a per builtin basis and avoid any impact to target independent builtins. There will be future work to support vectors passed as function arguments and supporting inline assembly. And whatever else we can find that isn't covered by this patch. Special thanks to Chandler who suggested this direction and reviewed a preview version of this patch. And thanks to Eric Christopher who has had many conversations with me about this issue. Differential Revision: https://reviews.llvm.org/D48617 llvm-svn: 336583
2018-07-10 03:00:16 +08:00
#define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("sse4.1"), __min_vector_width__(128)))
/* SSE4 Rounding macros. */
#define _MM_FROUND_TO_NEAREST_INT 0x00
#define _MM_FROUND_TO_NEG_INF 0x01
#define _MM_FROUND_TO_POS_INF 0x02
#define _MM_FROUND_TO_ZERO 0x03
#define _MM_FROUND_CUR_DIRECTION 0x04
#define _MM_FROUND_RAISE_EXC 0x00
#define _MM_FROUND_NO_EXC 0x08
#define _MM_FROUND_NINT (_MM_FROUND_RAISE_EXC | _MM_FROUND_TO_NEAREST_INT)
#define _MM_FROUND_FLOOR (_MM_FROUND_RAISE_EXC | _MM_FROUND_TO_NEG_INF)
#define _MM_FROUND_CEIL (_MM_FROUND_RAISE_EXC | _MM_FROUND_TO_POS_INF)
#define _MM_FROUND_TRUNC (_MM_FROUND_RAISE_EXC | _MM_FROUND_TO_ZERO)
#define _MM_FROUND_RINT (_MM_FROUND_RAISE_EXC | _MM_FROUND_CUR_DIRECTION)
#define _MM_FROUND_NEARBYINT (_MM_FROUND_NO_EXC | _MM_FROUND_CUR_DIRECTION)
/// Rounds up each element of the 128-bit vector of [4 x float] to an
/// integer and returns the rounded values in a 128-bit vector of
/// [4 x float].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_ceil_ps(__m128 X);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDPS / ROUNDPS </c> instruction.
///
/// \param X
/// A 128-bit vector of [4 x float] values to be rounded up.
/// \returns A 128-bit vector of [4 x float] containing the rounded values.
#define _mm_ceil_ps(X) _mm_round_ps((X), _MM_FROUND_CEIL)
/// Rounds up each element of the 128-bit vector of [2 x double] to an
/// integer and returns the rounded values in a 128-bit vector of
/// [2 x double].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128d _mm_ceil_pd(__m128d X);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDPD / ROUNDPD </c> instruction.
///
/// \param X
/// A 128-bit vector of [2 x double] values to be rounded up.
/// \returns A 128-bit vector of [2 x double] containing the rounded values.
#define _mm_ceil_pd(X) _mm_round_pd((X), _MM_FROUND_CEIL)
/// Copies three upper elements of the first 128-bit vector operand to
/// the corresponding three upper elements of the 128-bit result vector of
/// [4 x float]. Rounds up the lowest element of the second 128-bit vector
/// operand to an integer and copies it to the lowest element of the 128-bit
/// result vector of [4 x float].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_ceil_ss(__m128 X, __m128 Y);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDSS / ROUNDSS </c> instruction.
///
/// \param X
/// A 128-bit vector of [4 x float]. The values stored in bits [127:32] are
/// copied to the corresponding bits of the result.
/// \param Y
/// A 128-bit vector of [4 x float]. The value stored in bits [31:0] is
/// rounded up to the nearest integer and copied to the corresponding bits
/// of the result.
/// \returns A 128-bit vector of [4 x float] containing the copied and rounded
/// values.
#define _mm_ceil_ss(X, Y) _mm_round_ss((X), (Y), _MM_FROUND_CEIL)
/// Copies the upper element of the first 128-bit vector operand to the
/// corresponding upper element of the 128-bit result vector of [2 x double].
/// Rounds up the lower element of the second 128-bit vector operand to an
/// integer and copies it to the lower element of the 128-bit result vector
/// of [2 x double].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128d _mm_ceil_sd(__m128d X, __m128d Y);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDSD / ROUNDSD </c> instruction.
///
/// \param X
/// A 128-bit vector of [2 x double]. The value stored in bits [127:64] is
/// copied to the corresponding bits of the result.
/// \param Y
/// A 128-bit vector of [2 x double]. The value stored in bits [63:0] is
/// rounded up to the nearest integer and copied to the corresponding bits
/// of the result.
/// \returns A 128-bit vector of [2 x double] containing the copied and rounded
/// values.
#define _mm_ceil_sd(X, Y) _mm_round_sd((X), (Y), _MM_FROUND_CEIL)
/// Rounds down each element of the 128-bit vector of [4 x float] to an
/// an integer and returns the rounded values in a 128-bit vector of
/// [4 x float].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_floor_ps(__m128 X);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDPS / ROUNDPS </c> instruction.
///
/// \param X
/// A 128-bit vector of [4 x float] values to be rounded down.
/// \returns A 128-bit vector of [4 x float] containing the rounded values.
#define _mm_floor_ps(X) _mm_round_ps((X), _MM_FROUND_FLOOR)
/// Rounds down each element of the 128-bit vector of [2 x double] to an
/// integer and returns the rounded values in a 128-bit vector of
/// [2 x double].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128d _mm_floor_pd(__m128d X);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDPD / ROUNDPD </c> instruction.
///
/// \param X
/// A 128-bit vector of [2 x double].
/// \returns A 128-bit vector of [2 x double] containing the rounded values.
#define _mm_floor_pd(X) _mm_round_pd((X), _MM_FROUND_FLOOR)
/// Copies three upper elements of the first 128-bit vector operand to
/// the corresponding three upper elements of the 128-bit result vector of
/// [4 x float]. Rounds down the lowest element of the second 128-bit vector
/// operand to an integer and copies it to the lowest element of the 128-bit
/// result vector of [4 x float].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_floor_ss(__m128 X, __m128 Y);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDSS / ROUNDSS </c> instruction.
///
/// \param X
/// A 128-bit vector of [4 x float]. The values stored in bits [127:32] are
/// copied to the corresponding bits of the result.
/// \param Y
/// A 128-bit vector of [4 x float]. The value stored in bits [31:0] is
/// rounded down to the nearest integer and copied to the corresponding bits
/// of the result.
/// \returns A 128-bit vector of [4 x float] containing the copied and rounded
/// values.
#define _mm_floor_ss(X, Y) _mm_round_ss((X), (Y), _MM_FROUND_FLOOR)
/// Copies the upper element of the first 128-bit vector operand to the
/// corresponding upper element of the 128-bit result vector of [2 x double].
/// Rounds down the lower element of the second 128-bit vector operand to an
/// integer and copies it to the lower element of the 128-bit result vector
/// of [2 x double].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128d _mm_floor_sd(__m128d X, __m128d Y);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDSD / ROUNDSD </c> instruction.
///
/// \param X
/// A 128-bit vector of [2 x double]. The value stored in bits [127:64] is
/// copied to the corresponding bits of the result.
/// \param Y
/// A 128-bit vector of [2 x double]. The value stored in bits [63:0] is
/// rounded down to the nearest integer and copied to the corresponding bits
/// of the result.
/// \returns A 128-bit vector of [2 x double] containing the copied and rounded
/// values.
#define _mm_floor_sd(X, Y) _mm_round_sd((X), (Y), _MM_FROUND_FLOOR)
/// Rounds each element of the 128-bit vector of [4 x float] to an
/// integer value according to the rounding control specified by the second
/// argument and returns the rounded values in a 128-bit vector of
/// [4 x float].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_round_ps(__m128 X, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDPS / ROUNDPS </c> instruction.
///
/// \param X
/// A 128-bit vector of [4 x float].
/// \param M
/// An integer value that specifies the rounding operation. \n
/// Bits [7:4] are reserved. \n
/// Bit [3] is a precision exception value: \n
/// 0: A normal PE exception is used \n
/// 1: The PE field is not updated \n
/// Bit [2] is the rounding control source: \n
/// 0: Use bits [1:0] of \a M \n
/// 1: Use the current MXCSR setting \n
/// Bits [1:0] contain the rounding control definition: \n
/// 00: Nearest \n
/// 01: Downward (toward negative infinity) \n
/// 10: Upward (toward positive infinity) \n
/// 11: Truncated
/// \returns A 128-bit vector of [4 x float] containing the rounded values.
#define _mm_round_ps(X, M) \
((__m128)__builtin_ia32_roundps((__v4sf)(__m128)(X), (M)))
/// Copies three upper elements of the first 128-bit vector operand to
/// the corresponding three upper elements of the 128-bit result vector of
/// [4 x float]. Rounds the lowest element of the second 128-bit vector
/// operand to an integer value according to the rounding control specified
/// by the third argument and copies it to the lowest element of the 128-bit
/// result vector of [4 x float].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_round_ss(__m128 X, __m128 Y, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDSS / ROUNDSS </c> instruction.
///
/// \param X
/// A 128-bit vector of [4 x float]. The values stored in bits [127:32] are
/// copied to the corresponding bits of the result.
/// \param Y
/// A 128-bit vector of [4 x float]. The value stored in bits [31:0] is
/// rounded to the nearest integer using the specified rounding control and
/// copied to the corresponding bits of the result.
/// \param M
/// An integer value that specifies the rounding operation. \n
/// Bits [7:4] are reserved. \n
/// Bit [3] is a precision exception value: \n
/// 0: A normal PE exception is used \n
/// 1: The PE field is not updated \n
/// Bit [2] is the rounding control source: \n
/// 0: Use bits [1:0] of \a M \n
/// 1: Use the current MXCSR setting \n
/// Bits [1:0] contain the rounding control definition: \n
/// 00: Nearest \n
/// 01: Downward (toward negative infinity) \n
/// 10: Upward (toward positive infinity) \n
/// 11: Truncated
/// \returns A 128-bit vector of [4 x float] containing the copied and rounded
/// values.
#define _mm_round_ss(X, Y, M) \
((__m128)__builtin_ia32_roundss((__v4sf)(__m128)(X), \
(__v4sf)(__m128)(Y), (M)))
/// Rounds each element of the 128-bit vector of [2 x double] to an
/// integer value according to the rounding control specified by the second
/// argument and returns the rounded values in a 128-bit vector of
/// [2 x double].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128d _mm_round_pd(__m128d X, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDPD / ROUNDPD </c> instruction.
///
/// \param X
/// A 128-bit vector of [2 x double].
/// \param M
/// An integer value that specifies the rounding operation. \n
/// Bits [7:4] are reserved. \n
/// Bit [3] is a precision exception value: \n
/// 0: A normal PE exception is used \n
/// 1: The PE field is not updated \n
/// Bit [2] is the rounding control source: \n
/// 0: Use bits [1:0] of \a M \n
/// 1: Use the current MXCSR setting \n
/// Bits [1:0] contain the rounding control definition: \n
/// 00: Nearest \n
/// 01: Downward (toward negative infinity) \n
/// 10: Upward (toward positive infinity) \n
/// 11: Truncated
/// \returns A 128-bit vector of [2 x double] containing the rounded values.
#define _mm_round_pd(X, M) \
((__m128d)__builtin_ia32_roundpd((__v2df)(__m128d)(X), (M)))
/// Copies the upper element of the first 128-bit vector operand to the
/// corresponding upper element of the 128-bit result vector of [2 x double].
/// Rounds the lower element of the second 128-bit vector operand to an
/// integer value according to the rounding control specified by the third
/// argument and copies it to the lower element of the 128-bit result vector
/// of [2 x double].
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128d _mm_round_sd(__m128d X, __m128d Y, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VROUNDSD / ROUNDSD </c> instruction.
///
/// \param X
/// A 128-bit vector of [2 x double]. The value stored in bits [127:64] is
/// copied to the corresponding bits of the result.
/// \param Y
/// A 128-bit vector of [2 x double]. The value stored in bits [63:0] is
/// rounded to the nearest integer using the specified rounding control and
/// copied to the corresponding bits of the result.
/// \param M
/// An integer value that specifies the rounding operation. \n
/// Bits [7:4] are reserved. \n
/// Bit [3] is a precision exception value: \n
/// 0: A normal PE exception is used \n
/// 1: The PE field is not updated \n
/// Bit [2] is the rounding control source: \n
/// 0: Use bits [1:0] of \a M \n
/// 1: Use the current MXCSR setting \n
/// Bits [1:0] contain the rounding control definition: \n
/// 00: Nearest \n
/// 01: Downward (toward negative infinity) \n
/// 10: Upward (toward positive infinity) \n
/// 11: Truncated
/// \returns A 128-bit vector of [2 x double] containing the copied and rounded
/// values.
#define _mm_round_sd(X, Y, M) \
((__m128d)__builtin_ia32_roundsd((__v2df)(__m128d)(X), \
(__v2df)(__m128d)(Y), (M)))
/* SSE4 Packed Blending Intrinsics. */
/// Returns a 128-bit vector of [2 x double] where the values are
/// selected from either the first or second operand as specified by the
/// third operand, the control mask.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128d _mm_blend_pd(__m128d V1, __m128d V2, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VBLENDPD / BLENDPD </c> instruction.
///
/// \param V1
/// A 128-bit vector of [2 x double].
/// \param V2
/// A 128-bit vector of [2 x double].
/// \param M
/// An immediate integer operand, with mask bits [1:0] specifying how the
/// values are to be copied. The position of the mask bit corresponds to the
/// index of a copied value. When a mask bit is 0, the corresponding 64-bit
/// element in operand \a V1 is copied to the same position in the result.
/// When a mask bit is 1, the corresponding 64-bit element in operand \a V2
/// is copied to the same position in the result.
/// \returns A 128-bit vector of [2 x double] containing the copied values.
#define _mm_blend_pd(V1, V2, M) \
((__m128d) __builtin_ia32_blendpd ((__v2df)(__m128d)(V1), \
(__v2df)(__m128d)(V2), (int)(M)))
/// Returns a 128-bit vector of [4 x float] where the values are selected
/// from either the first or second operand as specified by the third
/// operand, the control mask.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_blend_ps(__m128 V1, __m128 V2, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VBLENDPS / BLENDPS </c> instruction.
///
/// \param V1
/// A 128-bit vector of [4 x float].
/// \param V2
/// A 128-bit vector of [4 x float].
/// \param M
/// An immediate integer operand, with mask bits [3:0] specifying how the
/// values are to be copied. The position of the mask bit corresponds to the
/// index of a copied value. When a mask bit is 0, the corresponding 32-bit
/// element in operand \a V1 is copied to the same position in the result.
/// When a mask bit is 1, the corresponding 32-bit element in operand \a V2
/// is copied to the same position in the result.
/// \returns A 128-bit vector of [4 x float] containing the copied values.
#define _mm_blend_ps(V1, V2, M) \
((__m128) __builtin_ia32_blendps ((__v4sf)(__m128)(V1), \
(__v4sf)(__m128)(V2), (int)(M)))
/// Returns a 128-bit vector of [2 x double] where the values are
/// selected from either the first or second operand as specified by the
/// third operand, the control mask.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VBLENDVPD / BLENDVPD </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [2 x double].
/// \param __V2
/// A 128-bit vector of [2 x double].
/// \param __M
/// A 128-bit vector operand, with mask bits 127 and 63 specifying how the
/// values are to be copied. The position of the mask bit corresponds to the
/// most significant bit of a copied value. When a mask bit is 0, the
/// corresponding 64-bit element in operand \a __V1 is copied to the same
/// position in the result. When a mask bit is 1, the corresponding 64-bit
/// element in operand \a __V2 is copied to the same position in the result.
/// \returns A 128-bit vector of [2 x double] containing the copied values.
static __inline__ __m128d __DEFAULT_FN_ATTRS
_mm_blendv_pd (__m128d __V1, __m128d __V2, __m128d __M)
{
return (__m128d) __builtin_ia32_blendvpd ((__v2df)__V1, (__v2df)__V2,
(__v2df)__M);
}
/// Returns a 128-bit vector of [4 x float] where the values are
/// selected from either the first or second operand as specified by the
/// third operand, the control mask.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VBLENDVPS / BLENDVPS </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [4 x float].
/// \param __V2
/// A 128-bit vector of [4 x float].
/// \param __M
/// A 128-bit vector operand, with mask bits 127, 95, 63, and 31 specifying
/// how the values are to be copied. The position of the mask bit corresponds
/// to the most significant bit of a copied value. When a mask bit is 0, the
/// corresponding 32-bit element in operand \a __V1 is copied to the same
/// position in the result. When a mask bit is 1, the corresponding 32-bit
/// element in operand \a __V2 is copied to the same position in the result.
/// \returns A 128-bit vector of [4 x float] containing the copied values.
static __inline__ __m128 __DEFAULT_FN_ATTRS
_mm_blendv_ps (__m128 __V1, __m128 __V2, __m128 __M)
{
return (__m128) __builtin_ia32_blendvps ((__v4sf)__V1, (__v4sf)__V2,
(__v4sf)__M);
}
/// Returns a 128-bit vector of [16 x i8] where the values are selected
/// from either of the first or second operand as specified by the third
/// operand, the control mask.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPBLENDVB / PBLENDVB </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [16 x i8].
/// \param __V2
/// A 128-bit vector of [16 x i8].
/// \param __M
/// A 128-bit vector operand, with mask bits 127, 119, 111...7 specifying
/// how the values are to be copied. The position of the mask bit corresponds
/// to the most significant bit of a copied value. When a mask bit is 0, the
/// corresponding 8-bit element in operand \a __V1 is copied to the same
/// position in the result. When a mask bit is 1, the corresponding 8-bit
/// element in operand \a __V2 is copied to the same position in the result.
/// \returns A 128-bit vector of [16 x i8] containing the copied values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_blendv_epi8 (__m128i __V1, __m128i __V2, __m128i __M)
{
return (__m128i) __builtin_ia32_pblendvb128 ((__v16qi)__V1, (__v16qi)__V2,
(__v16qi)__M);
}
/// Returns a 128-bit vector of [8 x i16] where the values are selected
/// from either of the first or second operand as specified by the third
/// operand, the control mask.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128i _mm_blend_epi16(__m128i V1, __m128i V2, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPBLENDW / PBLENDW </c> instruction.
///
/// \param V1
/// A 128-bit vector of [8 x i16].
/// \param V2
/// A 128-bit vector of [8 x i16].
/// \param M
/// An immediate integer operand, with mask bits [7:0] specifying how the
/// values are to be copied. The position of the mask bit corresponds to the
/// index of a copied value. When a mask bit is 0, the corresponding 16-bit
/// element in operand \a V1 is copied to the same position in the result.
/// When a mask bit is 1, the corresponding 16-bit element in operand \a V2
/// is copied to the same position in the result.
/// \returns A 128-bit vector of [8 x i16] containing the copied values.
#define _mm_blend_epi16(V1, V2, M) \
((__m128i) __builtin_ia32_pblendw128 ((__v8hi)(__m128i)(V1), \
(__v8hi)(__m128i)(V2), (int)(M)))
/* SSE4 Dword Multiply Instructions. */
/// Multiples corresponding elements of two 128-bit vectors of [4 x i32]
/// and returns the lower 32 bits of the each product in a 128-bit vector of
/// [4 x i32].
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMULLD / PMULLD </c> instruction.
///
/// \param __V1
/// A 128-bit integer vector.
/// \param __V2
/// A 128-bit integer vector.
/// \returns A 128-bit integer vector containing the products of both operands.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_mullo_epi32 (__m128i __V1, __m128i __V2)
{
return (__m128i) ((__v4su)__V1 * (__v4su)__V2);
}
/// Multiplies corresponding even-indexed elements of two 128-bit
/// vectors of [4 x i32] and returns a 128-bit vector of [2 x i64]
/// containing the products.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMULDQ / PMULDQ </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [4 x i32].
/// \param __V2
/// A 128-bit vector of [4 x i32].
/// \returns A 128-bit vector of [2 x i64] containing the products of both
/// operands.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_mul_epi32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_pmuldq128 ((__v4si)__V1, (__v4si)__V2);
}
/* SSE4 Floating Point Dot Product Instructions. */
/// Computes the dot product of the two 128-bit vectors of [4 x float]
/// and returns it in the elements of the 128-bit result vector of
/// [4 x float].
///
/// The immediate integer operand controls which input elements
/// will contribute to the dot product, and where the final results are
/// returned.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_dp_ps(__m128 X, __m128 Y, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VDPPS / DPPS </c> instruction.
///
/// \param X
/// A 128-bit vector of [4 x float].
/// \param Y
/// A 128-bit vector of [4 x float].
/// \param M
/// An immediate integer operand. Mask bits [7:4] determine which elements
/// of the input vectors are used, with bit [4] corresponding to the lowest
/// element and bit [7] corresponding to the highest element of each [4 x
/// float] vector. If a bit is set, the corresponding elements from the two
/// input vectors are used as an input for dot product; otherwise that input
/// is treated as zero. Bits [3:0] determine which elements of the result
/// will receive a copy of the final dot product, with bit [0] corresponding
/// to the lowest element and bit [3] corresponding to the highest element of
/// each [4 x float] subvector. If a bit is set, the dot product is returned
/// in the corresponding element; otherwise that element is set to zero.
/// \returns A 128-bit vector of [4 x float] containing the dot product.
#define _mm_dp_ps(X, Y, M) \
((__m128) __builtin_ia32_dpps((__v4sf)(__m128)(X), \
(__v4sf)(__m128)(Y), (M)))
/// Computes the dot product of the two 128-bit vectors of [2 x double]
/// and returns it in the elements of the 128-bit result vector of
/// [2 x double].
///
/// The immediate integer operand controls which input
/// elements will contribute to the dot product, and where the final results
/// are returned.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128d _mm_dp_pd(__m128d X, __m128d Y, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VDPPD / DPPD </c> instruction.
///
/// \param X
/// A 128-bit vector of [2 x double].
/// \param Y
/// A 128-bit vector of [2 x double].
/// \param M
/// An immediate integer operand. Mask bits [5:4] determine which elements
/// of the input vectors are used, with bit [4] corresponding to the lowest
/// element and bit [5] corresponding to the highest element of each of [2 x
/// double] vector. If a bit is set, the corresponding elements from the two
/// input vectors are used as an input for dot product; otherwise that input
/// is treated as zero. Bits [1:0] determine which elements of the result
/// will receive a copy of the final dot product, with bit [0] corresponding
/// to the lowest element and bit [1] corresponding to the highest element of
/// each [2 x double] vector. If a bit is set, the dot product is returned in
/// the corresponding element; otherwise that element is set to zero.
#define _mm_dp_pd(X, Y, M) \
((__m128d) __builtin_ia32_dppd((__v2df)(__m128d)(X), \
(__v2df)(__m128d)(Y), (M)))
/* SSE4 Streaming Load Hint Instruction. */
/// Loads integer values from a 128-bit aligned memory location to a
/// 128-bit integer vector.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VMOVNTDQA / MOVNTDQA </c> instruction.
///
/// \param __V
/// A pointer to a 128-bit aligned memory location that contains the integer
/// values.
/// \returns A 128-bit integer vector containing the data stored at the
/// specified memory location.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_stream_load_si128 (__m128i const *__V)
{
return (__m128i) __builtin_nontemporal_load ((const __v2di *) __V);
}
/* SSE4 Packed Integer Min/Max Instructions. */
/// Compares the corresponding elements of two 128-bit vectors of
/// [16 x i8] and returns a 128-bit vector of [16 x i8] containing the lesser
/// of the two values.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMINSB / PMINSB </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [16 x i8].
/// \param __V2
/// A 128-bit vector of [16 x i8]
/// \returns A 128-bit vector of [16 x i8] containing the lesser values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_min_epi8 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_elementwise_min((__v16qs) __V1, (__v16qs) __V2);
}
/// Compares the corresponding elements of two 128-bit vectors of
/// [16 x i8] and returns a 128-bit vector of [16 x i8] containing the
/// greater value of the two.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMAXSB / PMAXSB </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [16 x i8].
/// \param __V2
/// A 128-bit vector of [16 x i8].
/// \returns A 128-bit vector of [16 x i8] containing the greater values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_max_epi8 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_elementwise_max((__v16qs) __V1, (__v16qs) __V2);
}
/// Compares the corresponding elements of two 128-bit vectors of
/// [8 x u16] and returns a 128-bit vector of [8 x u16] containing the lesser
/// value of the two.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMINUW / PMINUW </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [8 x u16].
/// \param __V2
/// A 128-bit vector of [8 x u16].
/// \returns A 128-bit vector of [8 x u16] containing the lesser values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_min_epu16 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_elementwise_min((__v8hu) __V1, (__v8hu) __V2);
}
/// Compares the corresponding elements of two 128-bit vectors of
/// [8 x u16] and returns a 128-bit vector of [8 x u16] containing the
/// greater value of the two.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMAXUW / PMAXUW </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [8 x u16].
/// \param __V2
/// A 128-bit vector of [8 x u16].
/// \returns A 128-bit vector of [8 x u16] containing the greater values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_max_epu16 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_elementwise_max((__v8hu) __V1, (__v8hu) __V2);
}
/// Compares the corresponding elements of two 128-bit vectors of
/// [4 x i32] and returns a 128-bit vector of [4 x i32] containing the lesser
/// value of the two.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMINSD / PMINSD </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [4 x i32].
/// \param __V2
/// A 128-bit vector of [4 x i32].
/// \returns A 128-bit vector of [4 x i32] containing the lesser values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_min_epi32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_elementwise_min((__v4si) __V1, (__v4si) __V2);
}
/// Compares the corresponding elements of two 128-bit vectors of
/// [4 x i32] and returns a 128-bit vector of [4 x i32] containing the
/// greater value of the two.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMAXSD / PMAXSD </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [4 x i32].
/// \param __V2
/// A 128-bit vector of [4 x i32].
/// \returns A 128-bit vector of [4 x i32] containing the greater values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_max_epi32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_elementwise_max((__v4si) __V1, (__v4si) __V2);
}
/// Compares the corresponding elements of two 128-bit vectors of
/// [4 x u32] and returns a 128-bit vector of [4 x u32] containing the lesser
/// value of the two.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMINUD / PMINUD </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [4 x u32].
/// \param __V2
/// A 128-bit vector of [4 x u32].
/// \returns A 128-bit vector of [4 x u32] containing the lesser values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_min_epu32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_elementwise_min((__v4su) __V1, (__v4su) __V2);
}
/// Compares the corresponding elements of two 128-bit vectors of
/// [4 x u32] and returns a 128-bit vector of [4 x u32] containing the
/// greater value of the two.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMAXUD / PMAXUD </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [4 x u32].
/// \param __V2
/// A 128-bit vector of [4 x u32].
/// \returns A 128-bit vector of [4 x u32] containing the greater values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_max_epu32 (__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_elementwise_max((__v4su) __V1, (__v4su) __V2);
}
/* SSE4 Insertion and Extraction from XMM Register Instructions. */
/// Takes the first argument \a X and inserts an element from the second
/// argument \a Y as selected by the third argument \a N. That result then
/// has elements zeroed out also as selected by the third argument \a N. The
/// resulting 128-bit vector of [4 x float] is then returned.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128 _mm_insert_ps(__m128 X, __m128 Y, const int N);
/// \endcode
///
/// This intrinsic corresponds to the <c> VINSERTPS </c> instruction.
///
/// \param X
/// A 128-bit vector source operand of [4 x float]. With the exception of
/// those bits in the result copied from parameter \a Y and zeroed by bits
/// [3:0] of \a N, all bits from this parameter are copied to the result.
/// \param Y
/// A 128-bit vector source operand of [4 x float]. One single-precision
/// floating-point element from this source, as determined by the immediate
/// parameter, is copied to the result.
/// \param N
/// Specifies which bits from operand \a Y will be copied, which bits in the
/// result they will be be copied to, and which bits in the result will be
/// cleared. The following assignments are made: \n
/// Bits [7:6] specify the bits to copy from operand \a Y: \n
/// 00: Selects bits [31:0] from operand \a Y. \n
/// 01: Selects bits [63:32] from operand \a Y. \n
/// 10: Selects bits [95:64] from operand \a Y. \n
/// 11: Selects bits [127:96] from operand \a Y. \n
/// Bits [5:4] specify the bits in the result to which the selected bits
/// from operand \a Y are copied: \n
/// 00: Copies the selected bits from \a Y to result bits [31:0]. \n
/// 01: Copies the selected bits from \a Y to result bits [63:32]. \n
/// 10: Copies the selected bits from \a Y to result bits [95:64]. \n
/// 11: Copies the selected bits from \a Y to result bits [127:96]. \n
/// Bits[3:0]: If any of these bits are set, the corresponding result
/// element is cleared.
/// \returns A 128-bit vector of [4 x float] containing the copied
/// single-precision floating point elements from the operands.
#define _mm_insert_ps(X, Y, N) __builtin_ia32_insertps128((X), (Y), (N))
/// Extracts a 32-bit integer from a 128-bit vector of [4 x float] and
/// returns it, using the immediate value parameter \a N as a selector.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_extract_ps(__m128 X, const int N);
/// \endcode
///
/// This intrinsic corresponds to the <c> VEXTRACTPS / EXTRACTPS </c>
/// instruction.
///
/// \param X
/// A 128-bit vector of [4 x float].
/// \param N
/// An immediate value. Bits [1:0] determines which bits from the argument
/// \a X are extracted and returned: \n
/// 00: Bits [31:0] of parameter \a X are returned. \n
/// 01: Bits [63:32] of parameter \a X are returned. \n
/// 10: Bits [95:64] of parameter \a X are returned. \n
/// 11: Bits [127:96] of parameter \a X are returned.
/// \returns A 32-bit integer containing the extracted 32 bits of float data.
#define _mm_extract_ps(X, N) \
__builtin_bit_cast(int, __builtin_ia32_vec_ext_v4sf((__v4sf)(__m128)(X), (int)(N)))
/* Miscellaneous insert and extract macros. */
/* Extract a single-precision float from X at index N into D. */
#define _MM_EXTRACT_FLOAT(D, X, N) \
do { (D) = __builtin_ia32_vec_ext_v4sf((__v4sf)(__m128)(X), (int)(N)); } while (0)
/* Or together 2 sets of indexes (X and Y) with the zeroing bits (Z) to create
an index suitable for _mm_insert_ps. */
#define _MM_MK_INSERTPS_NDX(X, Y, Z) (((X) << 6) | ((Y) << 4) | (Z))
/* Extract a float from X at index N into the first index of the return. */
#define _MM_PICK_OUT_PS(X, N) _mm_insert_ps (_mm_setzero_ps(), (X), \
_MM_MK_INSERTPS_NDX((N), 0, 0x0e))
/* Insert int into packed integer array at index. */
/// Constructs a 128-bit vector of [16 x i8] by first making a copy of
/// the 128-bit integer vector parameter, and then inserting the lower 8 bits
/// of an integer parameter \a I into an offset specified by the immediate
/// value parameter \a N.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128i _mm_insert_epi8(__m128i X, int I, const int N);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPINSRB / PINSRB </c> instruction.
///
/// \param X
/// A 128-bit integer vector of [16 x i8]. This vector is copied to the
/// result and then one of the sixteen elements in the result vector is
/// replaced by the lower 8 bits of \a I.
/// \param I
/// An integer. The lower 8 bits of this operand are written to the result
/// beginning at the offset specified by \a N.
/// \param N
/// An immediate value. Bits [3:0] specify the bit offset in the result at
/// which the lower 8 bits of \a I are written. \n
/// 0000: Bits [7:0] of the result are used for insertion. \n
/// 0001: Bits [15:8] of the result are used for insertion. \n
/// 0010: Bits [23:16] of the result are used for insertion. \n
/// 0011: Bits [31:24] of the result are used for insertion. \n
/// 0100: Bits [39:32] of the result are used for insertion. \n
/// 0101: Bits [47:40] of the result are used for insertion. \n
/// 0110: Bits [55:48] of the result are used for insertion. \n
/// 0111: Bits [63:56] of the result are used for insertion. \n
/// 1000: Bits [71:64] of the result are used for insertion. \n
/// 1001: Bits [79:72] of the result are used for insertion. \n
/// 1010: Bits [87:80] of the result are used for insertion. \n
/// 1011: Bits [95:88] of the result are used for insertion. \n
/// 1100: Bits [103:96] of the result are used for insertion. \n
/// 1101: Bits [111:104] of the result are used for insertion. \n
/// 1110: Bits [119:112] of the result are used for insertion. \n
/// 1111: Bits [127:120] of the result are used for insertion.
/// \returns A 128-bit integer vector containing the constructed values.
#define _mm_insert_epi8(X, I, N) \
((__m128i)__builtin_ia32_vec_set_v16qi((__v16qi)(__m128i)(X), \
(int)(I), (int)(N)))
/// Constructs a 128-bit vector of [4 x i32] by first making a copy of
/// the 128-bit integer vector parameter, and then inserting the 32-bit
/// integer parameter \a I at the offset specified by the immediate value
/// parameter \a N.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128i _mm_insert_epi32(__m128i X, int I, const int N);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPINSRD / PINSRD </c> instruction.
///
/// \param X
/// A 128-bit integer vector of [4 x i32]. This vector is copied to the
/// result and then one of the four elements in the result vector is
/// replaced by \a I.
/// \param I
/// A 32-bit integer that is written to the result beginning at the offset
/// specified by \a N.
/// \param N
/// An immediate value. Bits [1:0] specify the bit offset in the result at
/// which the integer \a I is written. \n
/// 00: Bits [31:0] of the result are used for insertion. \n
/// 01: Bits [63:32] of the result are used for insertion. \n
/// 10: Bits [95:64] of the result are used for insertion. \n
/// 11: Bits [127:96] of the result are used for insertion.
/// \returns A 128-bit integer vector containing the constructed values.
#define _mm_insert_epi32(X, I, N) \
((__m128i)__builtin_ia32_vec_set_v4si((__v4si)(__m128i)(X), \
(int)(I), (int)(N)))
#ifdef __x86_64__
/// Constructs a 128-bit vector of [2 x i64] by first making a copy of
/// the 128-bit integer vector parameter, and then inserting the 64-bit
/// integer parameter \a I, using the immediate value parameter \a N as an
/// insertion location selector.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128i _mm_insert_epi64(__m128i X, long long I, const int N);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPINSRQ / PINSRQ </c> instruction.
///
/// \param X
/// A 128-bit integer vector of [2 x i64]. This vector is copied to the
/// result and then one of the two elements in the result vector is replaced
/// by \a I.
/// \param I
/// A 64-bit integer that is written to the result beginning at the offset
/// specified by \a N.
/// \param N
/// An immediate value. Bit [0] specifies the bit offset in the result at
/// which the integer \a I is written. \n
/// 0: Bits [63:0] of the result are used for insertion. \n
/// 1: Bits [127:64] of the result are used for insertion. \n
/// \returns A 128-bit integer vector containing the constructed values.
#define _mm_insert_epi64(X, I, N) \
((__m128i)__builtin_ia32_vec_set_v2di((__v2di)(__m128i)(X), \
(long long)(I), (int)(N)))
#endif /* __x86_64__ */
/* Extract int from packed integer array at index. This returns the element
* as a zero extended value, so it is unsigned.
*/
/// Extracts an 8-bit element from the 128-bit integer vector of
/// [16 x i8], using the immediate value parameter \a N as a selector.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_extract_epi8(__m128i X, const int N);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPEXTRB / PEXTRB </c> instruction.
///
/// \param X
/// A 128-bit integer vector.
/// \param N
/// An immediate value. Bits [3:0] specify which 8-bit vector element from
/// the argument \a X to extract and copy to the result. \n
/// 0000: Bits [7:0] of parameter \a X are extracted. \n
/// 0001: Bits [15:8] of the parameter \a X are extracted. \n
/// 0010: Bits [23:16] of the parameter \a X are extracted. \n
/// 0011: Bits [31:24] of the parameter \a X are extracted. \n
/// 0100: Bits [39:32] of the parameter \a X are extracted. \n
/// 0101: Bits [47:40] of the parameter \a X are extracted. \n
/// 0110: Bits [55:48] of the parameter \a X are extracted. \n
/// 0111: Bits [63:56] of the parameter \a X are extracted. \n
/// 1000: Bits [71:64] of the parameter \a X are extracted. \n
/// 1001: Bits [79:72] of the parameter \a X are extracted. \n
/// 1010: Bits [87:80] of the parameter \a X are extracted. \n
/// 1011: Bits [95:88] of the parameter \a X are extracted. \n
/// 1100: Bits [103:96] of the parameter \a X are extracted. \n
/// 1101: Bits [111:104] of the parameter \a X are extracted. \n
/// 1110: Bits [119:112] of the parameter \a X are extracted. \n
/// 1111: Bits [127:120] of the parameter \a X are extracted.
/// \returns An unsigned integer, whose lower 8 bits are selected from the
/// 128-bit integer vector parameter and the remaining bits are assigned
/// zeros.
#define _mm_extract_epi8(X, N) \
((int)(unsigned char)__builtin_ia32_vec_ext_v16qi((__v16qi)(__m128i)(X), \
(int)(N)))
/// Extracts a 32-bit element from the 128-bit integer vector of
/// [4 x i32], using the immediate value parameter \a N as a selector.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_extract_epi32(__m128i X, const int N);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPEXTRD / PEXTRD </c> instruction.
///
/// \param X
/// A 128-bit integer vector.
/// \param N
/// An immediate value. Bits [1:0] specify which 32-bit vector element from
/// the argument \a X to extract and copy to the result. \n
/// 00: Bits [31:0] of the parameter \a X are extracted. \n
/// 01: Bits [63:32] of the parameter \a X are extracted. \n
/// 10: Bits [95:64] of the parameter \a X are extracted. \n
/// 11: Bits [127:96] of the parameter \a X are exracted.
/// \returns An integer, whose lower 32 bits are selected from the 128-bit
/// integer vector parameter and the remaining bits are assigned zeros.
#define _mm_extract_epi32(X, N) \
((int)__builtin_ia32_vec_ext_v4si((__v4si)(__m128i)(X), (int)(N)))
#ifdef __x86_64__
/// Extracts a 64-bit element from the 128-bit integer vector of
/// [2 x i64], using the immediate value parameter \a N as a selector.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// long long _mm_extract_epi64(__m128i X, const int N);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPEXTRQ / PEXTRQ </c> instruction.
///
/// \param X
/// A 128-bit integer vector.
/// \param N
/// An immediate value. Bit [0] specifies which 64-bit vector element from
/// the argument \a X to return. \n
/// 0: Bits [63:0] are returned. \n
/// 1: Bits [127:64] are returned. \n
/// \returns A 64-bit integer.
#define _mm_extract_epi64(X, N) \
((long long)__builtin_ia32_vec_ext_v2di((__v2di)(__m128i)(X), (int)(N)))
#endif /* __x86_64 */
/* SSE4 128-bit Packed Integer Comparisons. */
/// Tests whether the specified bits in a 128-bit integer vector are all
/// zeros.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPTEST / PTEST </c> instruction.
///
/// \param __M
/// A 128-bit integer vector containing the bits to be tested.
/// \param __V
/// A 128-bit integer vector selecting which bits to test in operand \a __M.
/// \returns TRUE if the specified bits are all zeros; FALSE otherwise.
static __inline__ int __DEFAULT_FN_ATTRS
_mm_testz_si128(__m128i __M, __m128i __V)
{
return __builtin_ia32_ptestz128((__v2di)__M, (__v2di)__V);
}
/// Tests whether the specified bits in a 128-bit integer vector are all
/// ones.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPTEST / PTEST </c> instruction.
///
/// \param __M
/// A 128-bit integer vector containing the bits to be tested.
/// \param __V
/// A 128-bit integer vector selecting which bits to test in operand \a __M.
/// \returns TRUE if the specified bits are all ones; FALSE otherwise.
static __inline__ int __DEFAULT_FN_ATTRS
_mm_testc_si128(__m128i __M, __m128i __V)
{
return __builtin_ia32_ptestc128((__v2di)__M, (__v2di)__V);
}
/// Tests whether the specified bits in a 128-bit integer vector are
/// neither all zeros nor all ones.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPTEST / PTEST </c> instruction.
///
/// \param __M
/// A 128-bit integer vector containing the bits to be tested.
/// \param __V
/// A 128-bit integer vector selecting which bits to test in operand \a __M.
/// \returns TRUE if the specified bits are neither all zeros nor all ones;
/// FALSE otherwise.
static __inline__ int __DEFAULT_FN_ATTRS
_mm_testnzc_si128(__m128i __M, __m128i __V)
{
return __builtin_ia32_ptestnzc128((__v2di)__M, (__v2di)__V);
}
/// Tests whether the specified bits in a 128-bit integer vector are all
/// ones.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_test_all_ones(__m128i V);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPTEST / PTEST </c> instruction.
///
/// \param V
/// A 128-bit integer vector containing the bits to be tested.
/// \returns TRUE if the bits specified in the operand are all set to 1; FALSE
/// otherwise.
#define _mm_test_all_ones(V) _mm_testc_si128((V), _mm_cmpeq_epi32((V), (V)))
/// Tests whether the specified bits in a 128-bit integer vector are
/// neither all zeros nor all ones.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_test_mix_ones_zeros(__m128i M, __m128i V);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPTEST / PTEST </c> instruction.
///
/// \param M
/// A 128-bit integer vector containing the bits to be tested.
/// \param V
/// A 128-bit integer vector selecting which bits to test in operand \a M.
/// \returns TRUE if the specified bits are neither all zeros nor all ones;
/// FALSE otherwise.
#define _mm_test_mix_ones_zeros(M, V) _mm_testnzc_si128((M), (V))
/// Tests whether the specified bits in a 128-bit integer vector are all
/// zeros.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_test_all_zeros(__m128i M, __m128i V);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPTEST / PTEST </c> instruction.
///
/// \param M
/// A 128-bit integer vector containing the bits to be tested.
/// \param V
/// A 128-bit integer vector selecting which bits to test in operand \a M.
/// \returns TRUE if the specified bits are all zeros; FALSE otherwise.
#define _mm_test_all_zeros(M, V) _mm_testz_si128 ((M), (V))
/* SSE4 64-bit Packed Integer Comparisons. */
/// Compares each of the corresponding 64-bit values of the 128-bit
/// integer vectors for equality.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPCMPEQQ / PCMPEQQ </c> instruction.
///
/// \param __V1
/// A 128-bit integer vector.
/// \param __V2
/// A 128-bit integer vector.
/// \returns A 128-bit integer vector containing the comparison results.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cmpeq_epi64(__m128i __V1, __m128i __V2)
{
return (__m128i)((__v2di)__V1 == (__v2di)__V2);
}
/* SSE4 Packed Integer Sign-Extension. */
/// Sign-extends each of the lower eight 8-bit integer elements of a
/// 128-bit vector of [16 x i8] to 16-bit values and returns them in a
/// 128-bit vector of [8 x i16]. The upper eight elements of the input vector
/// are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVSXBW / PMOVSXBW </c> instruction.
///
/// \param __V
/// A 128-bit vector of [16 x i8]. The lower eight 8-bit elements are sign-
/// extended to 16-bit values.
/// \returns A 128-bit vector of [8 x i16] containing the sign-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepi8_epi16(__m128i __V)
{
Fix the SSE4 byte sign extension in a cleaner way, and more thoroughly test that our intrinsics behave the same under -fsigned-char and -funsigned-char. This further testing uncovered that AVX-2 has a broken cmpgt for 8-bit elements, and has for a long time. This is fixed in the same way as SSE4 handles the case. The other ISA extensions currently work correctly because they use specific instruction intrinsics. As soon as they are rewritten in terms of generic IR, they will need to add these special casts. I've added the necessary testing to catch this however, so we shouldn't have to chase it down again. I considered changing the core typedef to be signed, but that seems like a bad idea. Notably, it would be an ABI break if anyone is reaching into the innards of the intrinsic headers and passing __v16qi on an API boundary. I can't be completely confident that this wouldn't happen due to a macro expanding in a lambda, etc., so it seems much better to leave it alone. It also matches GCC's behavior exactly. A fun side note is that for both GCC and Clang, -funsigned-char really does change the semantics of __v16qi. To observe this, consider: % cat x.cc #include <smmintrin.h> #include <iostream> int main() { __v16qi a = { 1, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}; __v16qi b = _mm_set1_epi8(-1); std::cout << (int)(a / b)[0] << ", " << (int)(a / b)[1] << '\n'; } % clang++ -o x x.cc && ./x -1, 1 % clang++ -funsigned-char -o x x.cc && ./x 0, 1 However, while this may be surprising, both Clang and GCC agree. Differential Revision: http://reviews.llvm.org/D13324 llvm-svn: 249097
2015-10-02 07:40:12 +08:00
/* This function always performs a signed extension, but __v16qi is a char
which may be signed or unsigned, so use __v16qs. */
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v16qs)__V, (__v16qs)__V, 0, 1, 2, 3, 4, 5, 6, 7), __v8hi);
}
/// Sign-extends each of the lower four 8-bit integer elements of a
/// 128-bit vector of [16 x i8] to 32-bit values and returns them in a
/// 128-bit vector of [4 x i32]. The upper twelve elements of the input
/// vector are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVSXBD / PMOVSXBD </c> instruction.
///
/// \param __V
/// A 128-bit vector of [16 x i8]. The lower four 8-bit elements are
/// sign-extended to 32-bit values.
/// \returns A 128-bit vector of [4 x i32] containing the sign-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepi8_epi32(__m128i __V)
{
Fix the SSE4 byte sign extension in a cleaner way, and more thoroughly test that our intrinsics behave the same under -fsigned-char and -funsigned-char. This further testing uncovered that AVX-2 has a broken cmpgt for 8-bit elements, and has for a long time. This is fixed in the same way as SSE4 handles the case. The other ISA extensions currently work correctly because they use specific instruction intrinsics. As soon as they are rewritten in terms of generic IR, they will need to add these special casts. I've added the necessary testing to catch this however, so we shouldn't have to chase it down again. I considered changing the core typedef to be signed, but that seems like a bad idea. Notably, it would be an ABI break if anyone is reaching into the innards of the intrinsic headers and passing __v16qi on an API boundary. I can't be completely confident that this wouldn't happen due to a macro expanding in a lambda, etc., so it seems much better to leave it alone. It also matches GCC's behavior exactly. A fun side note is that for both GCC and Clang, -funsigned-char really does change the semantics of __v16qi. To observe this, consider: % cat x.cc #include <smmintrin.h> #include <iostream> int main() { __v16qi a = { 1, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}; __v16qi b = _mm_set1_epi8(-1); std::cout << (int)(a / b)[0] << ", " << (int)(a / b)[1] << '\n'; } % clang++ -o x x.cc && ./x -1, 1 % clang++ -funsigned-char -o x x.cc && ./x 0, 1 However, while this may be surprising, both Clang and GCC agree. Differential Revision: http://reviews.llvm.org/D13324 llvm-svn: 249097
2015-10-02 07:40:12 +08:00
/* This function always performs a signed extension, but __v16qi is a char
which may be signed or unsigned, so use __v16qs. */
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v16qs)__V, (__v16qs)__V, 0, 1, 2, 3), __v4si);
}
/// Sign-extends each of the lower two 8-bit integer elements of a
/// 128-bit integer vector of [16 x i8] to 64-bit values and returns them in
/// a 128-bit vector of [2 x i64]. The upper fourteen elements of the input
/// vector are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVSXBQ / PMOVSXBQ </c> instruction.
///
/// \param __V
/// A 128-bit vector of [16 x i8]. The lower two 8-bit elements are
/// sign-extended to 64-bit values.
/// \returns A 128-bit vector of [2 x i64] containing the sign-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepi8_epi64(__m128i __V)
{
Fix the SSE4 byte sign extension in a cleaner way, and more thoroughly test that our intrinsics behave the same under -fsigned-char and -funsigned-char. This further testing uncovered that AVX-2 has a broken cmpgt for 8-bit elements, and has for a long time. This is fixed in the same way as SSE4 handles the case. The other ISA extensions currently work correctly because they use specific instruction intrinsics. As soon as they are rewritten in terms of generic IR, they will need to add these special casts. I've added the necessary testing to catch this however, so we shouldn't have to chase it down again. I considered changing the core typedef to be signed, but that seems like a bad idea. Notably, it would be an ABI break if anyone is reaching into the innards of the intrinsic headers and passing __v16qi on an API boundary. I can't be completely confident that this wouldn't happen due to a macro expanding in a lambda, etc., so it seems much better to leave it alone. It also matches GCC's behavior exactly. A fun side note is that for both GCC and Clang, -funsigned-char really does change the semantics of __v16qi. To observe this, consider: % cat x.cc #include <smmintrin.h> #include <iostream> int main() { __v16qi a = { 1, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}; __v16qi b = _mm_set1_epi8(-1); std::cout << (int)(a / b)[0] << ", " << (int)(a / b)[1] << '\n'; } % clang++ -o x x.cc && ./x -1, 1 % clang++ -funsigned-char -o x x.cc && ./x 0, 1 However, while this may be surprising, both Clang and GCC agree. Differential Revision: http://reviews.llvm.org/D13324 llvm-svn: 249097
2015-10-02 07:40:12 +08:00
/* This function always performs a signed extension, but __v16qi is a char
which may be signed or unsigned, so use __v16qs. */
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v16qs)__V, (__v16qs)__V, 0, 1), __v2di);
}
/// Sign-extends each of the lower four 16-bit integer elements of a
/// 128-bit integer vector of [8 x i16] to 32-bit values and returns them in
/// a 128-bit vector of [4 x i32]. The upper four elements of the input
/// vector are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVSXWD / PMOVSXWD </c> instruction.
///
/// \param __V
/// A 128-bit vector of [8 x i16]. The lower four 16-bit elements are
/// sign-extended to 32-bit values.
/// \returns A 128-bit vector of [4 x i32] containing the sign-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepi16_epi32(__m128i __V)
{
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v8hi)__V, (__v8hi)__V, 0, 1, 2, 3), __v4si);
}
/// Sign-extends each of the lower two 16-bit integer elements of a
/// 128-bit integer vector of [8 x i16] to 64-bit values and returns them in
/// a 128-bit vector of [2 x i64]. The upper six elements of the input
/// vector are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVSXWQ / PMOVSXWQ </c> instruction.
///
/// \param __V
/// A 128-bit vector of [8 x i16]. The lower two 16-bit elements are
/// sign-extended to 64-bit values.
/// \returns A 128-bit vector of [2 x i64] containing the sign-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepi16_epi64(__m128i __V)
{
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v8hi)__V, (__v8hi)__V, 0, 1), __v2di);
}
/// Sign-extends each of the lower two 32-bit integer elements of a
/// 128-bit integer vector of [4 x i32] to 64-bit values and returns them in
/// a 128-bit vector of [2 x i64]. The upper two elements of the input vector
/// are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVSXDQ / PMOVSXDQ </c> instruction.
///
/// \param __V
/// A 128-bit vector of [4 x i32]. The lower two 32-bit elements are
/// sign-extended to 64-bit values.
/// \returns A 128-bit vector of [2 x i64] containing the sign-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepi32_epi64(__m128i __V)
{
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v4si)__V, (__v4si)__V, 0, 1), __v2di);
}
/* SSE4 Packed Integer Zero-Extension. */
/// Zero-extends each of the lower eight 8-bit integer elements of a
/// 128-bit vector of [16 x i8] to 16-bit values and returns them in a
/// 128-bit vector of [8 x i16]. The upper eight elements of the input vector
/// are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVZXBW / PMOVZXBW </c> instruction.
///
/// \param __V
/// A 128-bit vector of [16 x i8]. The lower eight 8-bit elements are
/// zero-extended to 16-bit values.
/// \returns A 128-bit vector of [8 x i16] containing the zero-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepu8_epi16(__m128i __V)
{
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v16qu)__V, (__v16qu)__V, 0, 1, 2, 3, 4, 5, 6, 7), __v8hi);
}
/// Zero-extends each of the lower four 8-bit integer elements of a
/// 128-bit vector of [16 x i8] to 32-bit values and returns them in a
/// 128-bit vector of [4 x i32]. The upper twelve elements of the input
/// vector are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVZXBD / PMOVZXBD </c> instruction.
///
/// \param __V
/// A 128-bit vector of [16 x i8]. The lower four 8-bit elements are
/// zero-extended to 32-bit values.
/// \returns A 128-bit vector of [4 x i32] containing the zero-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepu8_epi32(__m128i __V)
{
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v16qu)__V, (__v16qu)__V, 0, 1, 2, 3), __v4si);
}
/// Zero-extends each of the lower two 8-bit integer elements of a
/// 128-bit integer vector of [16 x i8] to 64-bit values and returns them in
/// a 128-bit vector of [2 x i64]. The upper fourteen elements of the input
/// vector are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVZXBQ / PMOVZXBQ </c> instruction.
///
/// \param __V
/// A 128-bit vector of [16 x i8]. The lower two 8-bit elements are
/// zero-extended to 64-bit values.
/// \returns A 128-bit vector of [2 x i64] containing the zero-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepu8_epi64(__m128i __V)
{
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v16qu)__V, (__v16qu)__V, 0, 1), __v2di);
}
/// Zero-extends each of the lower four 16-bit integer elements of a
/// 128-bit integer vector of [8 x i16] to 32-bit values and returns them in
/// a 128-bit vector of [4 x i32]. The upper four elements of the input
/// vector are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVZXWD / PMOVZXWD </c> instruction.
///
/// \param __V
/// A 128-bit vector of [8 x i16]. The lower four 16-bit elements are
/// zero-extended to 32-bit values.
/// \returns A 128-bit vector of [4 x i32] containing the zero-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepu16_epi32(__m128i __V)
{
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v8hu)__V, (__v8hu)__V, 0, 1, 2, 3), __v4si);
}
/// Zero-extends each of the lower two 16-bit integer elements of a
/// 128-bit integer vector of [8 x i16] to 64-bit values and returns them in
/// a 128-bit vector of [2 x i64]. The upper six elements of the input vector
/// are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVZXWQ / PMOVZXWQ </c> instruction.
///
/// \param __V
/// A 128-bit vector of [8 x i16]. The lower two 16-bit elements are
/// zero-extended to 64-bit values.
/// \returns A 128-bit vector of [2 x i64] containing the zero-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepu16_epi64(__m128i __V)
{
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v8hu)__V, (__v8hu)__V, 0, 1), __v2di);
}
/// Zero-extends each of the lower two 32-bit integer elements of a
/// 128-bit integer vector of [4 x i32] to 64-bit values and returns them in
/// a 128-bit vector of [2 x i64]. The upper two elements of the input vector
/// are unused.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPMOVZXDQ / PMOVZXDQ </c> instruction.
///
/// \param __V
/// A 128-bit vector of [4 x i32]. The lower two 32-bit elements are
/// zero-extended to 64-bit values.
/// \returns A 128-bit vector of [2 x i64] containing the zero-extended values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cvtepu32_epi64(__m128i __V)
{
return (__m128i)__builtin_convertvector(__builtin_shufflevector((__v4su)__V, (__v4su)__V, 0, 1), __v2di);
}
/* SSE4 Pack with Unsigned Saturation. */
/// Converts 32-bit signed integers from both 128-bit integer vector
/// operands into 16-bit unsigned integers, and returns the packed result.
/// Values greater than 0xFFFF are saturated to 0xFFFF. Values less than
/// 0x0000 are saturated to 0x0000.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPACKUSDW / PACKUSDW </c> instruction.
///
/// \param __V1
/// A 128-bit vector of [4 x i32]. Each 32-bit element is treated as a
/// signed integer and is converted to a 16-bit unsigned integer with
/// saturation. Values greater than 0xFFFF are saturated to 0xFFFF. Values
/// less than 0x0000 are saturated to 0x0000. The converted [4 x i16] values
/// are written to the lower 64 bits of the result.
/// \param __V2
/// A 128-bit vector of [4 x i32]. Each 32-bit element is treated as a
/// signed integer and is converted to a 16-bit unsigned integer with
/// saturation. Values greater than 0xFFFF are saturated to 0xFFFF. Values
/// less than 0x0000 are saturated to 0x0000. The converted [4 x i16] values
/// are written to the higher 64 bits of the result.
/// \returns A 128-bit vector of [8 x i16] containing the converted values.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_packus_epi32(__m128i __V1, __m128i __V2)
{
return (__m128i) __builtin_ia32_packusdw128((__v4si)__V1, (__v4si)__V2);
}
/* SSE4 Multiple Packed Sums of Absolute Difference. */
/// Subtracts 8-bit unsigned integer values and computes the absolute
/// values of the differences to the corresponding bits in the destination.
/// Then sums of the absolute differences are returned according to the bit
/// fields in the immediate operand.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128i _mm_mpsadbw_epu8(__m128i X, __m128i Y, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VMPSADBW / MPSADBW </c> instruction.
///
/// \param X
/// A 128-bit vector of [16 x i8].
/// \param Y
/// A 128-bit vector of [16 x i8].
/// \param M
/// An 8-bit immediate operand specifying how the absolute differences are to
/// be calculated, according to the following algorithm:
/// \code
/// // M2 represents bit 2 of the immediate operand
/// // M10 represents bits [1:0] of the immediate operand
/// i = M2 * 4;
/// j = M10 * 4;
/// for (k = 0; k < 8; k = k + 1) {
/// d0 = abs(X[i + k + 0] - Y[j + 0]);
/// d1 = abs(X[i + k + 1] - Y[j + 1]);
/// d2 = abs(X[i + k + 2] - Y[j + 2]);
/// d3 = abs(X[i + k + 3] - Y[j + 3]);
/// r[k] = d0 + d1 + d2 + d3;
/// }
/// \endcode
/// \returns A 128-bit integer vector containing the sums of the sets of
/// absolute differences between both operands.
#define _mm_mpsadbw_epu8(X, Y, M) \
((__m128i) __builtin_ia32_mpsadbw128((__v16qi)(__m128i)(X), \
(__v16qi)(__m128i)(Y), (M)))
/// Finds the minimum unsigned 16-bit element in the input 128-bit
/// vector of [8 x u16] and returns it and along with its index.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPHMINPOSUW / PHMINPOSUW </c>
/// instruction.
///
/// \param __V
/// A 128-bit vector of [8 x u16].
/// \returns A 128-bit value where bits [15:0] contain the minimum value found
/// in parameter \a __V, bits [18:16] contain the index of the minimum value
/// and the remaining bits are set to 0.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_minpos_epu16(__m128i __V)
{
return (__m128i) __builtin_ia32_phminposuw128((__v8hi)__V);
}
/* Handle the sse4.2 definitions here. */
/* These definitions are normally in nmmintrin.h, but gcc puts them in here
so we'll do the same. */
#undef __DEFAULT_FN_ATTRS
#define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("sse4.2")))
/* These specify the type of data that we're comparing. */
#define _SIDD_UBYTE_OPS 0x00
#define _SIDD_UWORD_OPS 0x01
#define _SIDD_SBYTE_OPS 0x02
#define _SIDD_SWORD_OPS 0x03
/* These specify the type of comparison operation. */
#define _SIDD_CMP_EQUAL_ANY 0x00
#define _SIDD_CMP_RANGES 0x04
#define _SIDD_CMP_EQUAL_EACH 0x08
#define _SIDD_CMP_EQUAL_ORDERED 0x0c
/* These macros specify the polarity of the operation. */
#define _SIDD_POSITIVE_POLARITY 0x00
#define _SIDD_NEGATIVE_POLARITY 0x10
#define _SIDD_MASKED_POSITIVE_POLARITY 0x20
#define _SIDD_MASKED_NEGATIVE_POLARITY 0x30
/* These macros are used in _mm_cmpXstri() to specify the return. */
#define _SIDD_LEAST_SIGNIFICANT 0x00
#define _SIDD_MOST_SIGNIFICANT 0x40
/* These macros are used in _mm_cmpXstri() to specify the return. */
#define _SIDD_BIT_MASK 0x00
#define _SIDD_UNIT_MASK 0x40
/* SSE4.2 Packed Comparison Intrinsics. */
/// Uses the immediate operand \a M to perform a comparison of string
/// data with implicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns a 128-bit integer vector representing the result
/// mask of the comparison.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128i _mm_cmpistrm(__m128i A, __m128i B, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPISTRM / PCMPISTRM </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words, the type of comparison to perform, and the format of the return
/// value. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B. \n
/// Bit [6]: Determines whether the result is zero-extended or expanded to 16
/// bytes. \n
/// 0: The result is zero-extended to 16 bytes. \n
/// 1: The result is expanded to 16 bytes (this expansion is performed by
/// repeating each bit 8 or 16 times).
/// \returns Returns a 128-bit integer vector representing the result mask of
/// the comparison.
#define _mm_cmpistrm(A, B, M) \
((__m128i)__builtin_ia32_pcmpistrm128((__v16qi)(__m128i)(A), \
(__v16qi)(__m128i)(B), (int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with implicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns an integer representing the result index of the
/// comparison.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpistri(__m128i A, __m128i B, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPISTRI / PCMPISTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words, the type of comparison to perform, and the format of the return
/// value. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B. \n
/// Bit [6]: Determines whether the index of the lowest set bit or the
/// highest set bit is returned. \n
/// 0: The index of the least significant set bit. \n
/// 1: The index of the most significant set bit. \n
/// \returns Returns an integer representing the result index of the comparison.
#define _mm_cmpistri(A, B, M) \
((int)__builtin_ia32_pcmpistri128((__v16qi)(__m128i)(A), \
(__v16qi)(__m128i)(B), (int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with explicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns a 128-bit integer vector representing the result
/// mask of the comparison.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// __m128i _mm_cmpestrm(__m128i A, int LA, __m128i B, int LB, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPESTRM / PCMPESTRM </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LA
/// An integer that specifies the length of the string in \a A.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LB
/// An integer that specifies the length of the string in \a B.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words, the type of comparison to perform, and the format of the return
/// value. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B. \n
/// Bit [6]: Determines whether the result is zero-extended or expanded to 16
/// bytes. \n
/// 0: The result is zero-extended to 16 bytes. \n
/// 1: The result is expanded to 16 bytes (this expansion is performed by
/// repeating each bit 8 or 16 times). \n
/// \returns Returns a 128-bit integer vector representing the result mask of
/// the comparison.
#define _mm_cmpestrm(A, LA, B, LB, M) \
((__m128i)__builtin_ia32_pcmpestrm128((__v16qi)(__m128i)(A), (int)(LA), \
(__v16qi)(__m128i)(B), (int)(LB), \
(int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with explicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns an integer representing the result index of the
/// comparison.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpestri(__m128i A, int LA, __m128i B, int LB, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPESTRI / PCMPESTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LA
/// An integer that specifies the length of the string in \a A.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LB
/// An integer that specifies the length of the string in \a B.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words, the type of comparison to perform, and the format of the return
/// value. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B. \n
/// Bit [6]: Determines whether the index of the lowest set bit or the
/// highest set bit is returned. \n
/// 0: The index of the least significant set bit. \n
/// 1: The index of the most significant set bit. \n
/// \returns Returns an integer representing the result index of the comparison.
#define _mm_cmpestri(A, LA, B, LB, M) \
((int)__builtin_ia32_pcmpestri128((__v16qi)(__m128i)(A), (int)(LA), \
(__v16qi)(__m128i)(B), (int)(LB), \
(int)(M)))
/* SSE4.2 Packed Comparison Intrinsics and EFlag Reading. */
/// Uses the immediate operand \a M to perform a comparison of string
/// data with implicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns 1 if the bit mask is zero and the length of the
/// string in \a B is the maximum, otherwise, returns 0.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpistra(__m128i A, __m128i B, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPISTRI / PCMPISTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B. \n
/// \returns Returns 1 if the bit mask is zero and the length of the string in
/// \a B is the maximum; otherwise, returns 0.
#define _mm_cmpistra(A, B, M) \
((int)__builtin_ia32_pcmpistria128((__v16qi)(__m128i)(A), \
(__v16qi)(__m128i)(B), (int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with implicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns 1 if the bit mask is non-zero, otherwise, returns
/// 0.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpistrc(__m128i A, __m128i B, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPISTRI / PCMPISTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B.
/// \returns Returns 1 if the bit mask is non-zero, otherwise, returns 0.
#define _mm_cmpistrc(A, B, M) \
((int)__builtin_ia32_pcmpistric128((__v16qi)(__m128i)(A), \
(__v16qi)(__m128i)(B), (int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with implicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns bit 0 of the resulting bit mask.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpistro(__m128i A, __m128i B, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPISTRI / PCMPISTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B. \n
/// \returns Returns bit 0 of the resulting bit mask.
#define _mm_cmpistro(A, B, M) \
((int)__builtin_ia32_pcmpistrio128((__v16qi)(__m128i)(A), \
(__v16qi)(__m128i)(B), (int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with implicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns 1 if the length of the string in \a A is less than
/// the maximum, otherwise, returns 0.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpistrs(__m128i A, __m128i B, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPISTRI / PCMPISTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B. \n
/// \returns Returns 1 if the length of the string in \a A is less than the
/// maximum, otherwise, returns 0.
#define _mm_cmpistrs(A, B, M) \
((int)__builtin_ia32_pcmpistris128((__v16qi)(__m128i)(A), \
(__v16qi)(__m128i)(B), (int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with implicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns 1 if the length of the string in \a B is less than
/// the maximum, otherwise, returns 0.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpistrz(__m128i A, __m128i B, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPISTRI / PCMPISTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B.
/// \returns Returns 1 if the length of the string in \a B is less than the
/// maximum, otherwise, returns 0.
#define _mm_cmpistrz(A, B, M) \
((int)__builtin_ia32_pcmpistriz128((__v16qi)(__m128i)(A), \
(__v16qi)(__m128i)(B), (int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with explicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns 1 if the bit mask is zero and the length of the
/// string in \a B is the maximum, otherwise, returns 0.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpestra(__m128i A, int LA, __m128i B, int LB, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPESTRI / PCMPESTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LA
/// An integer that specifies the length of the string in \a A.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LB
/// An integer that specifies the length of the string in \a B.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B.
/// \returns Returns 1 if the bit mask is zero and the length of the string in
/// \a B is the maximum, otherwise, returns 0.
#define _mm_cmpestra(A, LA, B, LB, M) \
((int)__builtin_ia32_pcmpestria128((__v16qi)(__m128i)(A), (int)(LA), \
(__v16qi)(__m128i)(B), (int)(LB), \
(int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with explicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns 1 if the resulting mask is non-zero, otherwise,
/// returns 0.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpestrc(__m128i A, int LA, __m128i B, int LB, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPESTRI / PCMPESTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LA
/// An integer that specifies the length of the string in \a A.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LB
/// An integer that specifies the length of the string in \a B.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B. \n
/// \returns Returns 1 if the resulting mask is non-zero, otherwise, returns 0.
#define _mm_cmpestrc(A, LA, B, LB, M) \
((int)__builtin_ia32_pcmpestric128((__v16qi)(__m128i)(A), (int)(LA), \
(__v16qi)(__m128i)(B), (int)(LB), \
(int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with explicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns bit 0 of the resulting bit mask.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpestro(__m128i A, int LA, __m128i B, int LB, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPESTRI / PCMPESTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LA
/// An integer that specifies the length of the string in \a A.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LB
/// An integer that specifies the length of the string in \a B.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B.
/// \returns Returns bit 0 of the resulting bit mask.
#define _mm_cmpestro(A, LA, B, LB, M) \
((int)__builtin_ia32_pcmpestrio128((__v16qi)(__m128i)(A), (int)(LA), \
(__v16qi)(__m128i)(B), (int)(LB), \
(int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with explicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns 1 if the length of the string in \a A is less than
/// the maximum, otherwise, returns 0.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpestrs(__m128i A, int LA, __m128i B, int LB, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPESTRI / PCMPESTRI </c>
/// instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LA
/// An integer that specifies the length of the string in \a A.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LB
/// An integer that specifies the length of the string in \a B.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement in the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B. \n
/// \returns Returns 1 if the length of the string in \a A is less than the
/// maximum, otherwise, returns 0.
#define _mm_cmpestrs(A, LA, B, LB, M) \
((int)__builtin_ia32_pcmpestris128((__v16qi)(__m128i)(A), (int)(LA), \
(__v16qi)(__m128i)(B), (int)(LB), \
(int)(M)))
/// Uses the immediate operand \a M to perform a comparison of string
/// data with explicitly defined lengths that is contained in source operands
/// \a A and \a B. Returns 1 if the length of the string in \a B is less than
/// the maximum, otherwise, returns 0.
///
/// \headerfile <x86intrin.h>
///
/// \code
/// int _mm_cmpestrz(__m128i A, int LA, __m128i B, int LB, const int M);
/// \endcode
///
/// This intrinsic corresponds to the <c> VPCMPESTRI </c> instruction.
///
/// \param A
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LA
/// An integer that specifies the length of the string in \a A.
/// \param B
/// A 128-bit integer vector containing one of the source operands to be
/// compared.
/// \param LB
/// An integer that specifies the length of the string in \a B.
/// \param M
/// An 8-bit immediate operand specifying whether the characters are bytes or
/// words and the type of comparison to perform. \n
/// Bits [1:0]: Determine source data format. \n
/// 00: 16 unsigned bytes \n
/// 01: 8 unsigned words \n
/// 10: 16 signed bytes \n
/// 11: 8 signed words \n
/// Bits [3:2]: Determine comparison type and aggregation method. \n
/// 00: Subset: Each character in \a B is compared for equality with all
/// the characters in \a A. \n
/// 01: Ranges: Each character in \a B is compared to \a A. The comparison
/// basis is greater than or equal for even-indexed elements in \a A,
/// and less than or equal for odd-indexed elements in \a A. \n
/// 10: Match: Compare each pair of corresponding characters in \a A and
/// \a B for equality. \n
/// 11: Substring: Search \a B for substring matches of \a A. \n
/// Bits [5:4]: Determine whether to perform a one's complement on the bit
/// mask of the comparison results. \n
/// 00: No effect. \n
/// 01: Negate the bit mask. \n
/// 10: No effect. \n
/// 11: Negate the bit mask only for bits with an index less than or equal
/// to the size of \a A or \a B.
/// \returns Returns 1 if the length of the string in \a B is less than the
/// maximum, otherwise, returns 0.
#define _mm_cmpestrz(A, LA, B, LB, M) \
((int)__builtin_ia32_pcmpestriz128((__v16qi)(__m128i)(A), (int)(LA), \
(__v16qi)(__m128i)(B), (int)(LB), \
(int)(M)))
/* SSE4.2 Compare Packed Data -- Greater Than. */
/// Compares each of the corresponding 64-bit values of the 128-bit
/// integer vectors to determine if the values in the first operand are
/// greater than those in the second operand.
///
/// \headerfile <x86intrin.h>
///
/// This intrinsic corresponds to the <c> VPCMPGTQ / PCMPGTQ </c> instruction.
///
/// \param __V1
/// A 128-bit integer vector.
/// \param __V2
/// A 128-bit integer vector.
/// \returns A 128-bit integer vector containing the comparison results.
static __inline__ __m128i __DEFAULT_FN_ATTRS
_mm_cmpgt_epi64(__m128i __V1, __m128i __V2)
{
return (__m128i)((__v2di)__V1 > (__v2di)__V2);
}
#undef __DEFAULT_FN_ATTRS
#include <popcntintrin.h>
#include <crc32intrin.h>
#endif /* __SMMINTRIN_H */