| Server IP : 109.234.164.53 / Your IP : 216.73.216.110 Web Server : Apache System : Linux cervelle.o2switch.net 4.18.0-553.32.1.lve.el8.x86_64 #1 SMP Thu Dec 19 13:14:03 UTC 2024 x86_64 User : computer3 ( 1098) PHP Version : 7.1.33 Disable Function : NONE MySQL : OFF | cURL : ON | WGET : ON | Perl : ON | Python : ON | Sudo : OFF | Pkexec : OFF Directory : /proc/845366/root/usr/lib/clang/21/include/ |
Upload File : |
/*===----------- avxvnniint16intrin.h - AVXVNNIINT16 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 __IMMINTRIN_H
#error \
"Never use <avxvnniint16intrin.h> directly; include <immintrin.h> instead."
#endif // __IMMINTRIN_H
#ifndef __AVXVNNIINT16INTRIN_H
#define __AVXVNNIINT16INTRIN_H
/// Multiply groups of 2 adjacent pairs of signed 16-bit integers in \a __A with
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m128i _mm_dpwsud_epi32(__m128i __W, __m128i __A, __m128i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWSUD instruction.
///
/// \param __W
/// A 128-bit vector of [4 x int].
/// \param __A
/// A 128-bit vector of [8 x short].
/// \param __B
/// A 128-bit vector of [8 x unsigned short].
/// \returns
/// A 128-bit vector of [4 x int].
///
/// \code{.operation}
/// FOR j := 0 to 3
/// tmp1.dword := SignExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
/// tmp2.dword := SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
/// ENDFOR
/// dst[MAX:128] := 0
/// \endcode
#define _mm_dpwsud_epi32(__W, __A, __B) \
((__m128i)__builtin_ia32_vpdpwsud128((__v4si)(__W), (__v4si)(__A), \
(__v4si)(__B)))
/// Multiply groups of 2 adjacent pairs of signed 16-bit integers in \a __A with
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m256i _mm256_dpwsud_epi32(__m256i __W, __m256i __A, __m256i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWSUD instruction.
///
/// \param __W
/// A 256-bit vector of [8 x int].
/// \param __A
/// A 256-bit vector of [16 x short].
/// \param __B
/// A 256-bit vector of [16 x unsigned short].
/// \returns
/// A 256-bit vector of [8 x int].
///
/// \code{.operation}
/// FOR j := 0 to 7
/// tmp1.dword := SignExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
/// tmp2.dword := SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
/// ENDFOR
/// dst[MAX:256] := 0
/// \endcode
#define _mm256_dpwsud_epi32(__W, __A, __B) \
((__m256i)__builtin_ia32_vpdpwsud256((__v8si)(__W), (__v8si)(__A), \
(__v8si)(__B)))
/// Multiply groups of 2 adjacent pairs of signed 16-bit integers in \a __A with
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W with signed saturation, and store the packed
/// 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m128i _mm_dpwsuds_epi32(__m128i __W, __m128i __A, __m128i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWSUDS instruction.
///
/// \param __W
/// A 128-bit vector of [4 x int].
/// \param __A
/// A 128-bit vector of [8 x short].
/// \param __B
/// A 128-bit vector of [8 x unsigned short].
/// \returns
/// A 128-bit vector of [4 x int].
///
/// \code{.operation}
/// FOR j := 0 to 3
/// tmp1.dword := SignExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
/// tmp2.dword := SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
/// dst.dword[j] := SIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
/// ENDFOR
/// dst[MAX:128] := 0
/// \endcode
#define _mm_dpwsuds_epi32(__W, __A, __B) \
((__m128i)__builtin_ia32_vpdpwsuds128((__v4si)(__W), (__v4si)(__A), \
(__v4si)(__B)))
/// Multiply groups of 2 adjacent pairs of signed 16-bit integers in \a __A with
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W with signed saturation, and store the packed
/// 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m256i _mm256_dpwsuds_epi32(__m256i __W, __m256i __A, __m256i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWSUDS instruction.
///
/// \param __W
/// A 256-bit vector of [8 x int].
/// \param __A
/// A 256-bit vector of [16 x short].
/// \param __B
/// A 256-bit vector of [16 x unsigned short].
/// \returns
/// A 256-bit vector of [8 x int].
///
/// \code{.operation}
/// FOR j := 0 to 7
/// tmp1.dword := SignExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
/// tmp2.dword := SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
/// dst.dword[j] := SIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
/// ENDFOR
/// dst[MAX:256] := 0
/// \endcode
#define _mm256_dpwsuds_epi32(__W, __A, __B) \
((__m256i)__builtin_ia32_vpdpwsuds256((__v8si)(__W), (__v8si)(__A), \
(__v8si)(__B)))
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
/// corresponding signed 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m128i _mm_dpbusd_epi32(__m128i __W, __m128i __A, __m128i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWUSD instruction.
///
/// \param __W
/// A 128-bit vector of [4 x int].
/// \param __A
/// A 128-bit vector of [8 x unsigned short].
/// \param __B
/// A 128-bit vector of [8 x short].
/// \returns
/// A 128-bit vector of [4 x int].
///
/// \code{.operation}
/// FOR j := 0 to 3
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * SignExtend32(__B.word[2*j])
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
/// ENDFOR
/// dst[MAX:128] := 0
/// \endcode
#define _mm_dpwusd_epi32(__W, __A, __B) \
((__m128i)__builtin_ia32_vpdpwusd128((__v4si)(__W), (__v4si)(__A), \
(__v4si)(__B)))
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
/// corresponding signed 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m256i _mm256_dpwusd_epi32(__m256i __W, __m256i __A, __m256i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWUSD instruction.
///
/// \param __W
/// A 256-bit vector of [8 x int].
/// \param __A
/// A 256-bit vector of [16 x unsigned short].
/// \param __B
/// A 256-bit vector of [16 x short].
/// \returns
/// A 256-bit vector of [8 x int].
///
/// \code{.operation}
/// FOR j := 0 to 7
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * SignExtend32(__B.word[2*j])
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
/// ENDFOR
/// dst[MAX:256] := 0
/// \endcode
#define _mm256_dpwusd_epi32(__W, __A, __B) \
((__m256i)__builtin_ia32_vpdpwusd256((__v8si)(__W), (__v8si)(__A), \
(__v8si)(__B)))
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
/// corresponding signed 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W with signed saturation, and store the packed
/// 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m128i _mm_dpwusds_epi32(__m128i __W, __m128i __A, __m128i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWSUDS instruction.
///
/// \param __W
/// A 128-bit vector of [4 x int].
/// \param __A
/// A 128-bit vector of [8 x unsigned short].
/// \param __B
/// A 128-bit vector of [8 x short].
/// \returns
/// A 128-bit vector of [4 x int].
///
/// \code{.operation}
/// FOR j := 0 to 3
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * SignExtend32(__B.word[2*j])
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
/// dst.dword[j] := SIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
/// ENDFOR
/// dst[MAX:128] := 0
/// \endcode
#define _mm_dpwusds_epi32(__W, __A, __B) \
((__m128i)__builtin_ia32_vpdpwusds128((__v4si)(__W), (__v4si)(__A), \
(__v4si)(__B)))
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
/// corresponding signed 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W with signed saturation, and store the packed
/// 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m256i _mm256_dpwsuds_epi32(__m256i __W, __m256i __A, __m256i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWSUDS instruction.
///
/// \param __W
/// A 256-bit vector of [8 x int].
/// \param __A
/// A 256-bit vector of [16 x unsigned short].
/// \param __B
/// A 256-bit vector of [16 x short].
/// \returns
/// A 256-bit vector of [8 x int].
///
/// \code{.operation}
/// FOR j := 0 to 7
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * SignExtend32(__B.word[2*j])
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
/// dst.dword[j] := SIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
/// ENDFOR
/// dst[MAX:256] := 0
/// \endcode
#define _mm256_dpwusds_epi32(__W, __A, __B) \
((__m256i)__builtin_ia32_vpdpwusds256((__v8si)(__W), (__v8si)(__A), \
(__v8si)(__B)))
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m128i _mm_dpwuud_epi32(__m128i __W, __m128i __A, __m128i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWUUD instruction.
///
/// \param __W
/// A 128-bit vector of [4 x unsigned int].
/// \param __A
/// A 128-bit vector of [8 x unsigned short].
/// \param __B
/// A 128-bit vector of [8 x unsigned short].
/// \returns
/// A 128-bit vector of [4 x unsigned int].
///
/// \code{.operation}
/// FOR j := 0 to 3
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
/// ENDFOR
/// dst[MAX:128] := 0
/// \endcode
#define _mm_dpwuud_epi32(__W, __A, __B) \
((__m128i)__builtin_ia32_vpdpwuud128((__v4si)(__W), (__v4si)(__A), \
(__v4si)(__B)))
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m256i _mm256_dpwuud_epi32(__m256i __W, __m256i __A, __m256i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWUUD instruction.
///
/// \param __W
/// A 256-bit vector of [8 x unsigned int].
/// \param __A
/// A 256-bit vector of [16 x unsigned short].
/// \param __B
/// A 256-bit vector of [16 x unsigned short].
/// \returns
/// A 256-bit vector of [8 x unsigned int].
///
/// \code{.operation}
/// FOR j := 0 to 7
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
/// ENDFOR
/// dst[MAX:256] := 0
/// \endcode
#define _mm256_dpwuud_epi32(__W, __A, __B) \
((__m256i)__builtin_ia32_vpdpwuud256((__v8si)(__W), (__v8si)(__A), \
(__v8si)(__B)))
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W with signed saturation, and store the packed
/// 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m128i _mm_dpwsuds_epi32(__m128i __W, __m128i __A, __m128i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWSUDS instruction.
///
/// \param __W
/// A 128-bit vector of [4 x unsigned int].
/// \param __A
/// A 128-bit vector of [8 x unsigned short].
/// \param __B
/// A 128-bit vector of [8 x unsigned short].
/// \returns
/// A 128-bit vector of [4 x unsigned int].
///
/// \code{.operation}
/// FOR j := 0 to 3
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
/// dst.dword[j] := UNSIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
/// ENDFOR
/// dst[MAX:128] := 0
/// \endcode
#define _mm_dpwuuds_epi32(__W, __A, __B) \
((__m128i)__builtin_ia32_vpdpwuuds128((__v4si)(__W), (__v4si)(__A), \
(__v4si)(__B)))
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
/// signed 16-bit results. Sum these 2 results with the corresponding
/// 32-bit integer in \a __W with signed saturation, and store the packed
/// 32-bit results in \a dst.
///
/// \headerfile <immintrin.h>
///
/// \code
/// __m256i _mm256_dpwuuds_epi32(__m256i __W, __m256i __A, __m256i __B)
/// \endcode
///
/// This intrinsic corresponds to the \c VPDPWSUDS instruction.
///
/// \param __W
/// A 256-bit vector of [8 x unsigned int].
/// \param __A
/// A 256-bit vector of [16 x unsigned short].
/// \param __B
/// A 256-bit vector of [16 x unsigned short].
/// \returns
/// A 256-bit vector of [8 x unsigned int].
///
/// \code{.operation}
/// FOR j := 0 to 7
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
/// dst.dword[j] := UNSIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
/// ENDFOR
/// dst[MAX:256] := 0
/// \endcode
#define _mm256_dpwuuds_epi32(__W, __A, __B) \
((__m256i)__builtin_ia32_vpdpwuuds256((__v8si)(__W), (__v8si)(__A), \
(__v8si)(__B)))
#endif // __AVXVNNIINT16INTRIN_H