On Fri, Oct 14, 2022 at 3:58 PM Haochen Jiang via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
> From: Kong Lingling <lingling.k...@intel.com>
>
> gcc/ChangeLog:
>
> * common/config/i386/i386-common.cc
> (OPTION_MASK_ISA2_AVXNECONVERT_SET,
> OPTION_MASK_ISA2_AVXNECONVERT_UNSET): New.
> (ix86_handle_option): Handle -mavxneconvert, unset
> avxneconvert when avx2 is disabled.
> * common/config/i386/i386-cpuinfo.h (processor_types): Add
> FEATURE_AVXNECONVERT.
> * common/config/i386/i386-isas.h: Add ISA_NAME_TABLE_ENTRY for
> avxneconvert.
> * common/config/i386/cpuinfo.h (get_available_features):
> Detect avxneconvert.
> * config.gcc: Add avxneconvertintrin.h
> * config/i386/avxneconvertintrin.h: New.
> * config/i386/cpuid.h (bit_AVXNECONVERT): New.
> * config/i386/i386-builtin-types.def: Add
> DEF_POINTER_TYPE (PCV8HF, V8HF, CONST),
> DEF_POINTER_TYPE (PCV16HF, V16HF, CONST),
> DEF_FUNCTION_TYPE (V4SF, PCSHORT),
> DEF_FUNCTION_TYPE (V8SF, PCSHORT),
> DEF_FUNCTION_TYPE (V4SF, PCV8BF),
> DEF_FUNCTION_TYPE (V4SF, PCV8BF),
> DEF_FUNCTION_TYPE (V8SF, PCV16HF),
> DEF_FUNCTION_TYPE (V8SF, PCV16BF).
> * config/i386/i386-builtin.def: Add new builtins.
> * config/i386/i386-c.cc (ix86_target_macros_internal): Define
> __AVXNECONVERT__.
> * config/i386/i386-expand.cc (ix86_expand_special_args_builtin):
> Handle V4SF_FTYPE_PCSHORT,V8SF_FTYPE_PCSHORT,V4SF_FTYPE_PCV8BF,
> V4SF_FTYPE_PCV8HF,V8SF_FTYPE_PCV16BF,V8SF_FTYPE_PCV16HF.
> * config/i386/i386-isa.def : Add DEF_PTA(AVXNECONVERT) New.
> * config/i386/i386-options.cc (isa2_opts): Add -mavxneconvert.
> (ix86_valid_target_attribute_inner_p): Handle avxneconvert.
> * config/i386/i386.opt: Add option -mavxneconvert.
> * config/i386/immintrin.h: Inculde avxneconvertintrin.h.
> * config/i386/sse.md: (avx_vbcstne<vbcstnetype>2ps_<mode>),
> (avx_vcvtne<vcvtnebf16type>2ps_<mode>),
> (avx_vcvtne<vcvtnephtype>2ps_<mode>),
> (avx_vcvtneps2bf16_<mode>): New define_insn
> (avx512f_cvtneps2bf16_<mode>):Ditto.
> (avx512f_cvtneps2bf16_<mode>_mask):Ditto.
> * doc/invoke.texi: Document -mavxneconvert.
> * doc/extend.texi: Document avxneconvert.
> * doc/sourcebuild.texi: Document target avxneconvert.
>
> gcc/testsuite/ChangeLog:
>
> * gcc.target/i386/avx-check.h: Add avxneconvert check.
> * gcc.target/i386/funcspec-56.inc: Add new target attribute.
> * gcc.target/i386/sse-12.c: Add -mavxneconvert.
> * gcc.target/i386/sse-13.c: Ditto.
> * gcc.target/i386/sse-14.c: Ditto.
> * gcc.target/i386/sse-22.c: Ditto.
> * gcc.target/i386/sse-23.c: Ditto.
> * g++.dg/other/i386-2.C: Ditto.
> * g++.dg/other/i386-3.C: Ditto.
> * lib/target-supports.exp:add check_effective_target_avxneconvert.
> * gcc.target/i386/avx-ne-convert-1.c: New test.
> * gcc.target/i386/avx-ne-convert-vbcstnebf162ps-2.c: Ditto.
> * gcc.target/i386/avx-ne-convert-vbcstnesh2ps-2.c: Ditto.
> * gcc.target/i386/avx-ne-convert-vcvtneebf162ps-2.c: Ditto.
> * gcc.target/i386/avx-ne-convert-vcvtneeph2ps-2.c: Ditto.
> * gcc.target/i386/avx-ne-convert-vcvtneobf162ps-2.c: Ditto.
> * gcc.target/i386/avx-ne-convert-vcvtneoph2ps-2.c: Ditto.
> * gcc.target/i386/avx-ne-convert-vcvtneps2bf16-2.c: Ditto.
> * gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1.c: Rename..
> * gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1a.c: To this.
> * gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1b.c: New test.
> ---
> gcc/common/config/i386/cpuinfo.h | 2 +
> gcc/common/config/i386/i386-common.cc | 21 ++-
> gcc/common/config/i386/i386-cpuinfo.h | 1 +
> gcc/common/config/i386/i386-isas.h | 2 +
> gcc/config.gcc | 2 +-
> gcc/config/i386/avxneconvertintrin.h | 140 ++++++++++++++++++
> gcc/config/i386/cpuid.h | 1 +
> gcc/config/i386/i386-builtin-types.def | 17 +++
> gcc/config/i386/i386-builtin.def | 18 +++
> gcc/config/i386/i386-c.cc | 2 +
> gcc/config/i386/i386-expand.cc | 8 +
> gcc/config/i386/i386-isa.def | 1 +
> gcc/config/i386/i386-options.cc | 4 +-
> gcc/config/i386/i386.opt | 5 +
> gcc/config/i386/immintrin.h | 4 +
> gcc/config/i386/sse.md | 100 ++++++++++++-
> gcc/doc/extend.texi | 5 +
> gcc/doc/invoke.texi | 9 +-
> gcc/doc/sourcebuild.texi | 3 +
> gcc/testsuite/g++.dg/other/i386-2.C | 2 +-
> gcc/testsuite/g++.dg/other/i386-3.C | 2 +-
> gcc/testsuite/gcc.target/i386/avx-check.h | 3 +
> .../gcc.target/i386/avx-ne-convert-1.c | 45 ++++++
> .../i386/avx-ne-convert-vbcstnebf162ps-2.c | 54 +++++++
> .../i386/avx-ne-convert-vbcstnesh2ps-2.c | 42 ++++++
> .../i386/avx-ne-convert-vcvtneebf162ps-2.c | 73 +++++++++
> .../i386/avx-ne-convert-vcvtneeph2ps-2.c | 66 +++++++++
> .../i386/avx-ne-convert-vcvtneobf162ps-2.c | 75 ++++++++++
> .../i386/avx-ne-convert-vcvtneoph2ps-2.c | 66 +++++++++
> .../i386/avx-ne-convert-vcvtneps2bf16-2.c | 58 ++++++++
> ...16-1.c => avx512bf16vl-vcvtneps2bf16-1a.c} | 0
> .../i386/avx512bf16vl-vcvtneps2bf16-1b.c | 27 ++++
> gcc/testsuite/gcc.target/i386/funcspec-56.inc | 2 +
> gcc/testsuite/gcc.target/i386/sse-12.c | 2 +-
> gcc/testsuite/gcc.target/i386/sse-13.c | 2 +-
> gcc/testsuite/gcc.target/i386/sse-14.c | 2 +-
> gcc/testsuite/gcc.target/i386/sse-22.c | 4 +-
> gcc/testsuite/gcc.target/i386/sse-23.c | 2 +-
> gcc/testsuite/lib/target-supports.exp | 12 ++
> 39 files changed, 868 insertions(+), 16 deletions(-)
> create mode 100644 gcc/config/i386/avxneconvertintrin.h
> create mode 100644 gcc/testsuite/gcc.target/i386/avx-ne-convert-1.c
> create mode 100644
> gcc/testsuite/gcc.target/i386/avx-ne-convert-vbcstnebf162ps-2.c
> create mode 100644
> gcc/testsuite/gcc.target/i386/avx-ne-convert-vbcstnesh2ps-2.c
> create mode 100644
> gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneebf162ps-2.c
> create mode 100644
> gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneeph2ps-2.c
> create mode 100644
> gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneobf162ps-2.c
> create mode 100644
> gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneoph2ps-2.c
> create mode 100644
> gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneps2bf16-2.c
> rename gcc/testsuite/gcc.target/i386/{avx512bf16vl-vcvtneps2bf16-1.c =>
> avx512bf16vl-vcvtneps2bf16-1a.c} (100%)
> create mode 100644
> gcc/testsuite/gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1b.c
>
> diff --git a/gcc/common/config/i386/cpuinfo.h
> b/gcc/common/config/i386/cpuinfo.h
> index bed88003f8e..e9fd586704d 100644
> --- a/gcc/common/config/i386/cpuinfo.h
> +++ b/gcc/common/config/i386/cpuinfo.h
> @@ -797,6 +797,8 @@ get_available_features (struct __processor_model
> *cpu_model,
> set_feature (FEATURE_AVXIFMA);
> if (edx & bit_AVXVNNIINT8)
> set_feature (FEATURE_AVXVNNIINT8);
> + if (edx & bit_AVXNECONVERT)
> + set_feature (FEATURE_AVXNECONVERT);
> }
> if (avx512_usable)
> {
> diff --git a/gcc/common/config/i386/i386-common.cc
> b/gcc/common/config/i386/i386-common.cc
> index 6a2a7e3d25a..f9c906f75cf 100644
> --- a/gcc/common/config/i386/i386-common.cc
> +++ b/gcc/common/config/i386/i386-common.cc
> @@ -109,6 +109,7 @@ along with GCC; see the file COPYING3. If not see
> #define OPTION_MASK_ISA2_AMX_INT8_SET OPTION_MASK_ISA2_AMX_INT8
> #define OPTION_MASK_ISA2_AMX_BF16_SET OPTION_MASK_ISA2_AMX_BF16
> #define OPTION_MASK_ISA2_AVXVNNIINT8_SET OPTION_MASK_ISA2_AVXVNNIINT8
> +#define OPTION_MASK_ISA2_AVXNECONVERT_SET OPTION_MASK_ISA2_AVXNECONVERT
>
> /* SSE4 includes both SSE4.1 and SSE4.2. -msse4 should be the same
> as -msse4.2. */
> @@ -215,7 +216,8 @@ along with GCC; see the file COPYING3. If not see
> (OPTION_MASK_ISA_AVX2 | OPTION_MASK_ISA_AVX512F_UNSET)
> #define OPTION_MASK_ISA2_AVX2_UNSET \
> (OPTION_MASK_ISA2_AVXIFMA_UNSET | OPTION_MASK_ISA2_AVXVNNI_UNSET \
> - | OPTION_MASK_ISA2_AVXVNNIINT8_UNSET | OPTION_MASK_ISA2_AVX512F_UNSET)
> + | OPTION_MASK_ISA2_AVXVNNIINT8_UNSET |
> OPTION_MASK_ISA2_AVXNECONVERT_UNSET \
> + | OPTION_MASK_ISA2_AVX512F_UNSET)
> #define OPTION_MASK_ISA_AVX512F_UNSET \
> (OPTION_MASK_ISA_AVX512F | OPTION_MASK_ISA_AVX512CD_UNSET \
> | OPTION_MASK_ISA_AVX512PF_UNSET | OPTION_MASK_ISA_AVX512ER_UNSET \
> @@ -280,6 +282,7 @@ along with GCC; see the file COPYING3. If not see
> (OPTION_MASK_ISA2_KL | OPTION_MASK_ISA2_WIDEKL_UNSET)
> #define OPTION_MASK_ISA2_WIDEKL_UNSET OPTION_MASK_ISA2_WIDEKL
> #define OPTION_MASK_ISA2_AVXVNNIINT8_UNSET OPTION_MASK_ISA2_AVXVNNIINT8
> +#define OPTION_MASK_ISA2_AVXNECONVERT_UNSET OPTION_MASK_ISA2_AVXNECONVERT
>
> /* SSE4 includes both SSE4.1 and SSE4.2. -mno-sse4 should the same
> as -mno-sse4.1. */
> @@ -1162,6 +1165,22 @@ ix86_handle_option (struct gcc_options *opts,
> }
> return true;
>
> + case OPT_mavxneconvert:
> + if (value)
> + {
> + opts->x_ix86_isa_flags2 |= OPTION_MASK_ISA2_AVXNECONVERT_SET;
> + opts->x_ix86_isa_flags2_explicit |=
> OPTION_MASK_ISA2_AVXNECONVERT_SET;
> + opts->x_ix86_isa_flags |= OPTION_MASK_ISA_AVX2_SET;
> + opts->x_ix86_isa_flags_explicit |= OPTION_MASK_ISA_AVX2_SET;
> + }
> + else
> + {
> + opts->x_ix86_isa_flags2 &= ~OPTION_MASK_ISA2_AVXNECONVERT_UNSET;
> + opts->x_ix86_isa_flags2_explicit
> + |= OPTION_MASK_ISA2_AVXNECONVERT_UNSET;
> + }
> + return true;
> +
> case OPT_mfma:
> if (value)
> {
> diff --git a/gcc/common/config/i386/i386-cpuinfo.h
> b/gcc/common/config/i386/i386-cpuinfo.h
> index 9a6b92fab79..2d3fbfc817a 100644
> --- a/gcc/common/config/i386/i386-cpuinfo.h
> +++ b/gcc/common/config/i386/i386-cpuinfo.h
> @@ -242,6 +242,7 @@ enum processor_features
> FEATURE_X86_64_V4,
> FEATURE_AVXIFMA,
> FEATURE_AVXVNNIINT8,
> + FEATURE_AVXNECONVERT,
> CPU_FEATURE_MAX
> };
>
> diff --git a/gcc/common/config/i386/i386-isas.h
> b/gcc/common/config/i386/i386-isas.h
> index 8c1f351056c..bceaee589ee 100644
> --- a/gcc/common/config/i386/i386-isas.h
> +++ b/gcc/common/config/i386/i386-isas.h
> @@ -178,4 +178,6 @@ ISA_NAMES_TABLE_START
> ISA_NAMES_TABLE_ENTRY("avxifma", FEATURE_AVXIFMA, P_NONE, "-mavxifma")
> ISA_NAMES_TABLE_ENTRY("avxvnniint8", FEATURE_AVXVNNIINT8,
> P_NONE, "-mavxvnniint8")
> + ISA_NAMES_TABLE_ENTRY("avxneconvert", FEATURE_AVXNECONVERT,
> + P_NONE, "-mavxneconvert")
> ISA_NAMES_TABLE_END
> diff --git a/gcc/config.gcc b/gcc/config.gcc
> index 4df78238910..840b62aee61 100644
> --- a/gcc/config.gcc
> +++ b/gcc/config.gcc
> @@ -422,7 +422,7 @@ i[34567]86-*-* | x86_64-*-*)
> amxbf16intrin.h x86gprintrin.h uintrintrin.h
> hresetintrin.h keylockerintrin.h avxvnniintrin.h
> mwaitintrin.h avx512fp16intrin.h avx512fp16vlintrin.h
> - avxifmaintrin.h avxvnniint8intrin.h"
> + avxifmaintrin.h avxvnniint8intrin.h
> avxneconvertintrin.h"
> ;;
> ia64-*-*)
> extra_headers=ia64intrin.h
> diff --git a/gcc/config/i386/avxneconvertintrin.h
> b/gcc/config/i386/avxneconvertintrin.h
> new file mode 100644
> index 00000000000..30199384725
> --- /dev/null
> +++ b/gcc/config/i386/avxneconvertintrin.h
> @@ -0,0 +1,140 @@
> +/* Copyright (C) 2021 Free Software Foundation, Inc.
> +
> + This file is part of GCC.
> +
> + GCC is free software; you can redistribute it and/or modify
> + it under the terms of the GNU General Public License as published by
> + the Free Software Foundation; either version 3, or (at your option)
> + any later version.
> +
> + GCC is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + GNU General Public License for more details.
> +
> + Under Section 7 of GPL version 3, you are granted additional
> + permissions described in the GCC Runtime Library Exception, version
> + 3.1, as published by the Free Software Foundation.
> +
> + You should have received a copy of the GNU General Public License and
> + a copy of the GCC Runtime Library Exception along with this program;
> + see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
> + <http://www.gnu.org/licenses/>. */
> +
> +#ifndef _IMMINTRIN_H_INCLUDED
> +#error "Never use <avxneconvertintrin.h> directly; include <immintrin.h>
> instead."
> +#endif
> +
> +#ifndef _AVXNECONVERTINTRIN_H_INCLUDED
> +#define _AVXNECONVERTINTRIN_H_INCLUDED
> +
> +#ifndef __AVXNECONVERT__
> +#pragma GCC push_options
> +#pragma GCC target ("avxneconvert")
> +#define __DISABLE_AVXNECONVERT__
> +#endif /* __AVXNECONVERT__ */
> +
> +extern __inline __m128
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm_bcstnebf16_ps (const void *__P)
> +{
> + return (__m128) __builtin_ia32_vbcstnebf162ps128 ((const short *) __P);
> +}
> +
> +extern __inline __m256
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm256_bcstnebf16_ps (const void *__P)
> +{
> + return (__m256) __builtin_ia32_vbcstnebf162ps256 ((const short *) __P);
> +}
> +
> +extern __inline __m128
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm_bcstnesh_ps (const void *__P)
> +{
> + return (__m128) __builtin_ia32_vbcstnesh2ps128 ((const short *) __P);
> +}
> +
> +extern __inline __m256
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm256_bcstnesh_ps (const void *__P)
> +{
> + return (__m256) __builtin_ia32_vbcstnesh2ps256 ((const short *) __P);
> +}
> +
> +extern __inline __m128
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm_cvtneebf16_ps (const __m128bf16 *__A)
> +{
> + return (__m128) __builtin_ia32_vcvtneebf162ps128 ((const __v8bf *) __A);
> +}
> +
> +extern __inline __m256
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm256_cvtneebf16_ps (const __m256bf16 *__A)
> +{
> + return (__m256) __builtin_ia32_vcvtneebf162ps256 ((const __v16bf *) __A);
> +}
> +
> +extern __inline __m128
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm_cvtneeph_ps (const __m128h *__A)
> +{
> + return (__m128) __builtin_ia32_vcvtneeph2ps128 ((const __v8hf *) __A);
> +}
> +
> +extern __inline __m256
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm256_cvtneeph_ps (const __m256h *__A)
> +{
> + return (__m256) __builtin_ia32_vcvtneeph2ps256 ((const __v16hf *) __A);
> +}
> +
> +extern __inline __m128
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm_cvtneobf16_ps (const __m128bf16 *__A)
> +{
> + return (__m128) __builtin_ia32_vcvtneobf162ps128 ((const __v8bf *) __A);
> +}
> +
> +extern __inline __m256
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm256_cvtneobf16_ps (const __m256bf16 *__A)
> +{
> + return (__m256) __builtin_ia32_vcvtneobf162ps256 ((const __v16bf *) __A);
> +}
> +
> +extern __inline __m128
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm_cvtneoph_ps (const __m128h *__A)
> +{
> + return (__m128) __builtin_ia32_vcvtneoph2ps128 ((const __v8hf *) __A);
> +}
> +
> +extern __inline __m256
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm256_cvtneoph_ps (const __m256h *__A)
> +{
> + return (__m256) __builtin_ia32_vcvtneoph2ps256 ((const __v16hf *) __A);
> +}
> +
> +extern __inline __m128bf16
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm_cvtneps_avx_pbh (__m128 __A)
> +{
> + return (__m128bf16) __builtin_ia32_vcvtneps2bf16128 (__A);
> +}
> +
> +extern __inline __m128bf16
> +__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
> +_mm256_cvtneps_avx_pbh (__m256 __A)
> +{
> + return (__m128bf16) __builtin_ia32_vcvtneps2bf16256 (__A);
> +}
> +
> +#ifdef __DISABLE_AVXNECONVERT__
> +#undef __DISABLE_AVXNECONVERT__
> +#pragma GCC pop_options
> +#endif /* __DISABLE_AVXNECONVERT__ */
> +
> +#endif /* _AVXNECONVERTINTRIN_H_INCLUDED */
> diff --git a/gcc/config/i386/cpuid.h b/gcc/config/i386/cpuid.h
> index f5fad22149a..18bbc0cb7be 100644
> --- a/gcc/config/i386/cpuid.h
> +++ b/gcc/config/i386/cpuid.h
> @@ -50,6 +50,7 @@
>
> /* %edx */
> #define bit_AVXVNNIINT8 (1 << 4)
> +#define bit_AVXNECONVERT (1 << 5)
> #define bit_CMPXCHG8B (1 << 8)
> #define bit_CMOV (1 << 15)
> #define bit_MMX (1 << 23)
> diff --git a/gcc/config/i386/i386-builtin-types.def
> b/gcc/config/i386/i386-builtin-types.def
> index 63a360b0f8b..ebf6e5b4ad8 100644
> --- a/gcc/config/i386/i386-builtin-types.def
> +++ b/gcc/config/i386/i386-builtin-types.def
> @@ -87,6 +87,7 @@ DEF_VECTOR_TYPE (V8QI, QI)
> DEF_VECTOR_TYPE (V2DF, DOUBLE)
> DEF_VECTOR_TYPE (V4SF, FLOAT)
> DEF_VECTOR_TYPE (V8HF, FLOAT16)
> +DEF_VECTOR_TYPE (V8BF, BFLOAT16)
> DEF_VECTOR_TYPE (V2DI, DI)
> DEF_VECTOR_TYPE (V4SI, SI)
> DEF_VECTOR_TYPE (V8HI, HI)
> @@ -100,6 +101,7 @@ DEF_VECTOR_TYPE (V16UQI, UQI, V16QI)
> DEF_VECTOR_TYPE (V4DF, DOUBLE)
> DEF_VECTOR_TYPE (V8SF, FLOAT)
> DEF_VECTOR_TYPE (V16HF, FLOAT16)
> +DEF_VECTOR_TYPE (V16BF, BFLOAT16)
> DEF_VECTOR_TYPE (V4DI, DI)
> DEF_VECTOR_TYPE (V8SI, SI)
> DEF_VECTOR_TYPE (V16HI, HI)
> @@ -111,6 +113,7 @@ DEF_VECTOR_TYPE (V16UHI, UHI, V16HI)
> # AVX512F vectors
> DEF_VECTOR_TYPE (V32SF, FLOAT)
> DEF_VECTOR_TYPE (V32HF, FLOAT16)
> +DEF_VECTOR_TYPE (V32BF, BFLOAT16)
> DEF_VECTOR_TYPE (V16SF, FLOAT)
> DEF_VECTOR_TYPE (V8DF, DOUBLE)
> DEF_VECTOR_TYPE (V8DI, DI)
> @@ -179,6 +182,10 @@ DEF_POINTER_TYPE (PCV4DF, V4DF, CONST)
> DEF_POINTER_TYPE (PCV4SF, V4SF, CONST)
> DEF_POINTER_TYPE (PCV8DF, V8DF, CONST)
> DEF_POINTER_TYPE (PCV8SF, V8SF, CONST)
> +DEF_POINTER_TYPE (PCV8HF, V8HF, CONST)
> +DEF_POINTER_TYPE (PCV8BF, V8BF, CONST)
> +DEF_POINTER_TYPE (PCV16HF, V16HF, CONST)
> +DEF_POINTER_TYPE (PCV16BF, V16BF, CONST)
> DEF_POINTER_TYPE (PCV16SF, V16SF, CONST)
>
> DEF_POINTER_TYPE (PCV2DI, V2DI, CONST)
> @@ -254,12 +261,14 @@ DEF_FUNCTION_TYPE (V4DF, V4SI)
> DEF_FUNCTION_TYPE (V8DF, V8DF)
> DEF_FUNCTION_TYPE (V4HI, V4HI)
> DEF_FUNCTION_TYPE (V4SF, PCFLOAT)
> +DEF_FUNCTION_TYPE (V4SF, PCSHORT)
> DEF_FUNCTION_TYPE (V4SF, V2DF)
> DEF_FUNCTION_TYPE (V4SF, V2DF, V4SF, UQI)
> DEF_FUNCTION_TYPE (V4SF, V4DF)
> DEF_FUNCTION_TYPE (V4SF, V4DF, V4SF, UQI)
> DEF_FUNCTION_TYPE (V4SF, V4SF)
> DEF_FUNCTION_TYPE (V4SF, PCV4SF)
> +DEF_FUNCTION_TYPE (V4SF, PCV8HF)
> DEF_FUNCTION_TYPE (V4SF, V4SI)
> DEF_FUNCTION_TYPE (V4SF, V8SF)
> DEF_FUNCTION_TYPE (V4SF, V8HI)
> @@ -275,8 +284,10 @@ DEF_FUNCTION_TYPE (V8HI, V16QI)
> DEF_FUNCTION_TYPE (V8HI, V8HI)
> DEF_FUNCTION_TYPE (V8QI, V8QI)
> DEF_FUNCTION_TYPE (V8SF, PCFLOAT)
> +DEF_FUNCTION_TYPE (V8SF, PCSHORT)
> DEF_FUNCTION_TYPE (V8SF, PCV4SF)
> DEF_FUNCTION_TYPE (V8SF, PCV8SF)
> +DEF_FUNCTION_TYPE (V8SF, PCV16HF)
> DEF_FUNCTION_TYPE (V8SF, V4SF)
> DEF_FUNCTION_TYPE (V8SF, V8SF)
> DEF_FUNCTION_TYPE (V8SF, V8SI)
> @@ -1389,3 +1400,9 @@ DEF_FUNCTION_TYPE (V32HF, V32HF)
> DEF_FUNCTION_TYPE_ALIAS (V8HF_FTYPE_V8HF, ROUND)
> DEF_FUNCTION_TYPE_ALIAS (V16HF_FTYPE_V16HF, ROUND)
> DEF_FUNCTION_TYPE_ALIAS (V32HF_FTYPE_V32HF, ROUND)
> +
> +# AVXNECONVERT builtins
> +DEF_FUNCTION_TYPE (V8BF, V8SF)
> +DEF_FUNCTION_TYPE (V8BF, V4SF)
> +DEF_FUNCTION_TYPE (V4SF, PCV8BF)
> +DEF_FUNCTION_TYPE (V8SF, PCV16BF)
> diff --git a/gcc/config/i386/i386-builtin.def
> b/gcc/config/i386/i386-builtin.def
> index e6edae5728b..a429577180c 100644
> --- a/gcc/config/i386/i386-builtin.def
> +++ b/gcc/config/i386/i386-builtin.def
> @@ -274,6 +274,20 @@ BDESC (OPTION_MASK_ISA_RTM, 0, CODE_FOR_xbegin,
> "__builtin_ia32_xbegin", IX86_BU
> BDESC (OPTION_MASK_ISA_RTM, 0, CODE_FOR_xend, "__builtin_ia32_xend",
> IX86_BUILTIN_XEND, UNKNOWN, (int) VOID_FTYPE_VOID)
> BDESC (OPTION_MASK_ISA_RTM, 0, CODE_FOR_xtest, "__builtin_ia32_xtest",
> IX86_BUILTIN_XTEST, UNKNOWN, (int) INT_FTYPE_VOID)
>
> +/* AVX-NE-CONVERT */
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vbcstnebf162ps_v4sf,
> "__builtin_ia32_vbcstnebf162ps128", IX86_BUILTIN_VBCSTNEBF162PS128, UNKNOWN,
> (int) V4SF_FTYPE_PCSHORT)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vbcstnebf162ps_v8sf,
> "__builtin_ia32_vbcstnebf162ps256", IX86_BUILTIN_VBCSTNEBF162PS256, UNKNOWN,
> (int) V8SF_FTYPE_PCSHORT)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vbcstnesh2ps_v4sf,
> "__builtin_ia32_vbcstnesh2ps128", IX86_BUILTIN_VBCSTNESH2PS128, UNKNOWN,
> (int) V4SF_FTYPE_PCSHORT)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vbcstnesh2ps_v8sf,
> "__builtin_ia32_vbcstnesh2ps256", IX86_BUILTIN_VBCSTNESH2PS256, UNKNOWN,
> (int) V8SF_FTYPE_PCSHORT)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vcvtneebf162ps_v4sf,
> "__builtin_ia32_vcvtneebf162ps128", IX86_BUILTIN_VCVTNEEBF162PS128, UNKNOWN,
> (int) V4SF_FTYPE_PCV8BF)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vcvtneebf162ps_v8sf,
> "__builtin_ia32_vcvtneebf162ps256", IX86_BUILTIN_VCVTNEEBF162PS256, UNKNOWN,
> (int) V8SF_FTYPE_PCV16BF)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vcvtneeph2ps_v4sf,
> "__builtin_ia32_vcvtneeph2ps128", IX86_BUILTIN_VCVTNEEPH2PS128, UNKNOWN,
> (int) V4SF_FTYPE_PCV8HF)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vcvtneeph2ps_v8sf,
> "__builtin_ia32_vcvtneeph2ps256", IX86_BUILTIN_VCVTNEEPH2PS256, UNKNOWN,
> (int) V8SF_FTYPE_PCV16HF)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vcvtneobf162ps_v4sf,
> "__builtin_ia32_vcvtneobf162ps128", IX86_BUILTIN_VCVTNEOBF162PS128, UNKNOWN,
> (int) V4SF_FTYPE_PCV8BF)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vcvtneobf162ps_v8sf,
> "__builtin_ia32_vcvtneobf162ps256", IX86_BUILTIN_VCVTNEOBF162PS256, UNKNOWN,
> (int) V8SF_FTYPE_PCV16BF)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vcvtneoph2ps_v4sf,
> "__builtin_ia32_vcvtneoph2ps128", IX86_BUILTIN_VCVTNEOPH2PS128, UNKNOWN,
> (int) V4SF_FTYPE_PCV8HF)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_vcvtneoph2ps_v8sf,
> "__builtin_ia32_vcvtneoph2ps256", IX86_BUILTIN_VCVTNEOPH2PS256, UNKNOWN,
> (int) V8SF_FTYPE_PCV16HF)
> +
> /* AVX512BW */
> BDESC (OPTION_MASK_ISA_AVX512BW, 0, CODE_FOR_avx512bw_loadv32hi_mask,
> "__builtin_ia32_loaddquhi512_mask", IX86_BUILTIN_LOADDQUHI512_MASK, UNKNOWN,
> (int) V32HI_FTYPE_PCSHORT_V32HI_USI)
> BDESC (OPTION_MASK_ISA_AVX512BW, 0, CODE_FOR_avx512bw_loadv64qi_mask,
> "__builtin_ia32_loaddquqi512_mask", IX86_BUILTIN_LOADDQUQI512_MASK, UNKNOWN,
> (int) V64QI_FTYPE_PCCHAR_V64QI_UDI)
> @@ -2809,6 +2823,10 @@ BDESC (0, OPTION_MASK_ISA2_AVX512BF16,
> CODE_FOR_avx512f_dpbf16ps_v4sf, "__builti
> BDESC (0, OPTION_MASK_ISA2_AVX512BF16, CODE_FOR_avx512f_dpbf16ps_v4sf_mask,
> "__builtin_ia32_dpbf16ps_v4sf_mask", IX86_BUILTIN_DPHI16PS_V4SF_MASK,
> UNKNOWN, (int) V4SF_FTYPE_V4SF_V8HI_V8HI_UQI)
> BDESC (0, OPTION_MASK_ISA2_AVX512BF16, CODE_FOR_avx512f_dpbf16ps_v4sf_maskz,
> "__builtin_ia32_dpbf16ps_v4sf_maskz", IX86_BUILTIN_DPHI16PS_V4SF_MASKZ,
> UNKNOWN, (int) V4SF_FTYPE_V4SF_V8HI_V8HI_UQI)
>
> +/* AVX-NE-CONVERT */
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_avx_vcvtneps2bf16_v4sf,
> "__builtin_ia32_vcvtneps2bf16128", IX86_BUILTIN_VCVTNEPS2BF16128, UNKNOWN,
> (int) V8BF_FTYPE_V4SF)
> +BDESC (0, OPTION_MASK_ISA2_AVXNECONVERT, CODE_FOR_avx_vcvtneps2bf16_v8sf,
> "__builtin_ia32_vcvtneps2bf16256", IX86_BUILTIN_VCVTNEPS2BF16256, UNKNOWN,
> (int) V8BF_FTYPE_V8SF)
> +
> /* AVX512FP16. */
> BDESC (OPTION_MASK_ISA_AVX512VL, OPTION_MASK_ISA2_AVX512FP16,
> CODE_FOR_addv8hf3_mask, "__builtin_ia32_addph128_mask",
> IX86_BUILTIN_ADDPH128_MASK, UNKNOWN, (int) V8HF_FTYPE_V8HF_V8HF_V8HF_UQI)
> BDESC (OPTION_MASK_ISA_AVX512VL, OPTION_MASK_ISA2_AVX512FP16,
> CODE_FOR_addv16hf3_mask, "__builtin_ia32_addph256_mask",
> IX86_BUILTIN_ADDPH256_MASK, UNKNOWN, (int) V16HF_FTYPE_V16HF_V16HF_V16HF_UHI)
> diff --git a/gcc/config/i386/i386-c.cc b/gcc/config/i386/i386-c.cc
> index a9a35c0a18a..48934df664c 100644
> --- a/gcc/config/i386/i386-c.cc
> +++ b/gcc/config/i386/i386-c.cc
> @@ -637,6 +637,8 @@ ix86_target_macros_internal (HOST_WIDE_INT isa_flag,
> def_or_undef (parse_in, "__AVXIFMA__");
> if (isa_flag2 & OPTION_MASK_ISA2_AVXVNNIINT8)
> def_or_undef (parse_in, "__AVXVNNIINT8__");
> + if (isa_flag2 & OPTION_MASK_ISA2_AVXNECONVERT)
> + def_or_undef (parse_in, "__AVXNECONVERT__");
> if (TARGET_IAMCU)
> {
> def_or_undef (parse_in, "__iamcu");
> diff --git a/gcc/config/i386/i386-expand.cc b/gcc/config/i386/i386-expand.cc
> index a0f8a98986e..1e29fe584af 100644
> --- a/gcc/config/i386/i386-expand.cc
> +++ b/gcc/config/i386/i386-expand.cc
> @@ -10427,7 +10427,9 @@ ix86_expand_args_builtin (const struct
> builtin_description *d,
> case V4DI_FTYPE_V4DI:
> case V16HI_FTYPE_V16SF:
> case V8HI_FTYPE_V8SF:
> + case V8BF_FTYPE_V8SF:
> case V8HI_FTYPE_V4SF:
> + case V8BF_FTYPE_V4SF:
> nargs = 1;
> break;
> case V4SF_FTYPE_V4SF_VEC_MERGE:
> @@ -11860,6 +11862,12 @@ ix86_expand_special_args_builtin (const struct
> builtin_description *d,
> case V8SF_FTYPE_PCV4SF:
> case V8SF_FTYPE_PCFLOAT:
> case V4SF_FTYPE_PCFLOAT:
> + case V4SF_FTYPE_PCSHORT:
> + case V4SF_FTYPE_PCV8BF:
> + case V4SF_FTYPE_PCV8HF:
> + case V8SF_FTYPE_PCSHORT:
> + case V8SF_FTYPE_PCV16BF:
> + case V8SF_FTYPE_PCV16HF:
> case V4DF_FTYPE_PCV2DF:
> case V4DF_FTYPE_PCDOUBLE:
> case V2DF_FTYPE_PCDOUBLE:
> diff --git a/gcc/config/i386/i386-isa.def b/gcc/config/i386/i386-isa.def
> index c95b917c6ce..4ea3f96f69f 100644
> --- a/gcc/config/i386/i386-isa.def
> +++ b/gcc/config/i386/i386-isa.def
> @@ -111,3 +111,4 @@ DEF_PTA(AVXVNNI)
> DEF_PTA(AVX512FP16)
> DEF_PTA(AVXIFMA)
> DEF_PTA(AVXVNNIINT8)
> +DEF_PTA(AVXNECONVERT)
> diff --git a/gcc/config/i386/i386-options.cc b/gcc/config/i386/i386-options.cc
> index 3e6d04433a6..e59e2d8aeaf 100644
> --- a/gcc/config/i386/i386-options.cc
> +++ b/gcc/config/i386/i386-options.cc
> @@ -228,7 +228,8 @@ static struct ix86_target_opts isa2_opts[] =
> { "-mavxvnni", OPTION_MASK_ISA2_AVXVNNI },
> { "-mavx512fp16", OPTION_MASK_ISA2_AVX512FP16 },
> { "-mavxifma", OPTION_MASK_ISA2_AVXIFMA },
> - { "-mavxvnniint8", OPTION_MASK_ISA2_AVXVNNIINT8 }
> + { "-mavxvnniint8", OPTION_MASK_ISA2_AVXVNNIINT8 },
> + { "-mavxneconvert", OPTION_MASK_ISA2_AVXNECONVERT }
> };
> static struct ix86_target_opts isa_opts[] =
> {
> @@ -1076,6 +1077,7 @@ ix86_valid_target_attribute_inner_p (tree fndecl, tree
> args, char *p_strings[],
> IX86_ATTR_ISA ("avx512fp16", OPT_mavx512fp16),
> IX86_ATTR_ISA ("avxifma", OPT_mavxifma),
> IX86_ATTR_ISA ("avxvnniint8", OPT_mavxvnniint8),
> + IX86_ATTR_ISA ("avxneconvert", OPT_mavxneconvert),
>
> /* enum options */
> IX86_ATTR_ENUM ("fpmath=", OPT_mfpmath_),
> diff --git a/gcc/config/i386/i386.opt b/gcc/config/i386/i386.opt
> index 53d534f6392..6e07b89ac4c 100644
> --- a/gcc/config/i386/i386.opt
> +++ b/gcc/config/i386/i386.opt
> @@ -1224,3 +1224,8 @@ mavxvnniint8
> Target Mask(ISA2_AVXVNNIINT8) Var(ix86_isa_flags2) Save
> Support MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, AVX, AVX2 and
> AVXVNNIINT8 built-in functions and code generation.
> +
> +mavxneconvert
> +Target Mask(ISA2_AVXNECONVERT) Var(ix86_isa_flags2) Save
> +Support MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, AVX, AVX2, and
> +AVXNECONVERT build-in functions and code generation.
> diff --git a/gcc/config/i386/immintrin.h b/gcc/config/i386/immintrin.h
> index c62d50f1951..d7433f639c8 100644
> --- a/gcc/config/i386/immintrin.h
> +++ b/gcc/config/i386/immintrin.h
> @@ -124,6 +124,10 @@
> #include <avx512bf16intrin.h>
> #endif
>
> +#ifdef __AVX2__
> +#include <avxneconvertintrin.h>
> +#endif
> +
> #include <amxtileintrin.h>
>
> #include <amxint8intrin.h>
> diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
> index 49490a213ea..bef4447de62 100644
> --- a/gcc/config/i386/sse.md
> +++ b/gcc/config/i386/sse.md
> @@ -171,6 +171,14 @@
> UNSPEC_VPMADDWDACCD
> UNSPEC_VPMADDWDACCSSD
>
> + ;; For AVXNECONVERT support
> + UNSPEC_VCVTNEBF16SF
> + UNSPEC_VCVTNESHSF
> + UNSPEC_VCVTNEEBF16SF
> + UNSPEC_VCVTNEEPHSF
> + UNSPEC_VCVTNEOBF16SF
> + UNSPEC_VCVTNEOPHSF
> +
> ;; For VAES support
> UNSPEC_VAESDEC
> UNSPEC_VAESDECLAST
> @@ -28930,9 +28938,69 @@
> ;; Converting from SF to BF
> (define_mode_attr sf_cvt_bf16
> [(V4SF "V8HI") (V8SF "V8HI") (V16SF "V16HI")])
> +(define_mode_attr sf_cvt_bfloat16
> + [(V4SF "V8BF") (V8SF "V8BF")])
> ;; Mapping from BF to SF
> (define_mode_attr sf_bf16
> [(V4SF "V8HI") (V8SF "V16HI") (V16SF "V32HI")])
> +(define_mode_attr sf_bfloat16
> + [(V4SF "V8BF") (V8SF "V16BF") (V16SF "V32BF")])
> +;; Mapping from PH to SF
> +(define_mode_attr ph_cvt_sf
> + [(V4SF "V8HF") (V8SF "V16HF")])
> +
> +(define_int_iterator VBCSTNE
> + [UNSPEC_VCVTNEBF16SF
> + UNSPEC_VCVTNESHSF])
> +
> +(define_int_attr vbcstnetype
> + [(UNSPEC_VCVTNEBF16SF "bf16") (UNSPEC_VCVTNESHSF "sh")])
> +
> +(define_insn "vbcstne<vbcstnetype>2ps_<mode>"
> + [(set (match_operand:VF1_128_256 0 "register_operand" "=x")
> + (vec_duplicate:VF1_128_256
> + (unspec:SF
> + [(match_operand:HI 1 "memory_operand" "m")]
> + VBCSTNE)))]
> + "TARGET_AVXNECONVERT"
> + "vbcstne<vbcstnetype>2ps\t{%1, %0|%0, %1}"
> + [(set_attr "prefix" "vex")
> + (set_attr "mode" "<sseinsnmode>")])
Since jakub has support bf16 software emulation, can we rewrite it
with general rtl ir without unspec?
Like (float_extend:SF (match_operand:BF "memory_operand" "m")
> +
> +(define_int_iterator VCVTNEBF16
> + [UNSPEC_VCVTNEEBF16SF
> + UNSPEC_VCVTNEOBF16SF])
> +
> +(define_int_attr vcvtnebf16type
> + [(UNSPEC_VCVTNEEBF16SF "ebf16")
> + (UNSPEC_VCVTNEOBF16SF "obf16")])
> +(define_insn "vcvtne<vcvtnebf16type>2ps_<mode>"
> + [(set (match_operand:VF1_128_256 0 "register_operand" "=x")
> + (unspec:VF1_128_256
> + [(match_operand:<sf_bfloat16> 1 "memory_operand" "m")]
> + VCVTNEBF16))]
> + "TARGET_AVXNECONVERT"
> + "vcvtne<vcvtnebf16type>2ps\t{%1, %0|%0, %1}"
> + [(set_attr "prefix" "vex")
> + (set_attr "mode" "<sseinsnmode>")])
Similar for this one and all those patterns below.
> +
> +(define_int_iterator VCVTNEPH
> + [UNSPEC_VCVTNEEPHSF
> + UNSPEC_VCVTNEOPHSF])
> +
> +(define_int_attr vcvtnephtype
> + [(UNSPEC_VCVTNEEPHSF "eph")
> + (UNSPEC_VCVTNEOPHSF "oph")])
> +
> +(define_insn "vcvtne<vcvtnephtype>2ps_<mode>"
> + [(set (match_operand:VF1_128_256 0 "register_operand" "=x")
> + (unspec:VF1_128_256
> + [(match_operand:<ph_cvt_sf> 1 "memory_operand" "m")]
> + VCVTNEPH))]
> + "TARGET_AVXNECONVERT"
> + "vcvtne<vcvtnephtype>2ps\t{%1, %0|%0, %1}"
> + [(set_attr "prefix" "vex")
> + (set_attr "mode" "<sseinsnmode>")])
>
> (define_expand "avx512f_cvtne2ps2bf16_<mode>_maskz"
> [(match_operand:BF16 0 "register_operand")
> @@ -28966,13 +29034,41 @@
> DONE;
> })
>
> -(define_insn "avx512f_cvtneps2bf16_<mode><mask_name>"
> +(define_insn "avx_vcvtneps2bf16_<mode>"
> + [(set (match_operand:<sf_cvt_bfloat16> 0 "register_operand" "=v")
> + (unspec:<sf_cvt_bfloat16>
> + [(match_operand:VF1_128_256 1 "register_operand" "v")]
> + UNSPEC_VCVTNEPS2BF16))]
> + "TARGET_AVXNECONVERT"
> + "%{vex%} vcvtneps2bf16\t{%1, %0|%0, %1}"
> + [(set_attr "prefix" "vex")])
> +
> +(define_insn "avx512f_cvtneps2bf16_<mode>"
> [(set (match_operand:<sf_cvt_bf16> 0 "register_operand" "=v")
> (unspec:<sf_cvt_bf16>
> [(match_operand:VF1_AVX512VL 1 "register_operand" "v")]
> UNSPEC_VCVTNEPS2BF16))]
> "TARGET_AVX512BF16"
> - "vcvtneps2bf16\t{%1, %0<mask_operand2>|%0<mask_operand2>, %1}")
> + {
> + if (<MODE_SIZE> <=32
> + && TARGET_AVXNECONVERT
> + && !EXT_REX_SSE_REG_P (operands[0])
> + && !EXT_REX_SSE_REG_P (operands[1]))
> + return "%{vex%} vcvtneps2bf16\t{%1, %0|%0, %1}";
> + else
> + return "vcvtneps2bf16\t{%1, %0|%0, %1}";
> + })
> +
> +(define_insn "avx512f_cvtneps2bf16_<mode>_mask"
> + [(set (match_operand:<sf_cvt_bf16> 0 "register_operand" "=v")
> + (vec_merge:<sf_cvt_bf16>
> + (unspec:<sf_cvt_bf16>
> + [(match_operand:VF1_AVX512VL 1 "register_operand" "v")]
> + UNSPEC_VCVTNEPS2BF16)
> + (match_operand:<sf_cvt_bf16> 2 "nonimm_or_0_operand" "0C")
> + (match_operand:<avx512fmaskmode> 3 "register_operand" "Yk")))]
> + "TARGET_AVX512BF16"
> + "vcvtneps2bf16\t{%1, %0%{%3%}%N2|%0%{%3%}%N2, %1}")
>
> (define_expand "avx512f_dpbf16ps_<mode>_maskz"
> [(match_operand:VF1_AVX512VL 0 "register_operand")
> diff --git a/gcc/doc/extend.texi b/gcc/doc/extend.texi
> index 9a8de9fc226..0a4396f92bb 100644
> --- a/gcc/doc/extend.texi
> +++ b/gcc/doc/extend.texi
> @@ -7070,6 +7070,11 @@ Enable/disable the generation of the AVXIFMA
> instructions.
> @cindex @code{target("avxvnniint8")} function attribute, x86
> Enable/disable the generation of the AVXVNNIINT8 instructions.
>
> +@item avxneconvert
> +@itemx no-avxneconvert
> +@cindex @code{target("avxneconvert")} function attribute, x86
> +Enable/disable the generation of the AVXNECONVERT instructions.
> +
> @item cld
> @itemx no-cld
> @cindex @code{target("cld")} function attribute, x86
> diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
> index d4ff7549bf3..307fb7fa441 100644
> --- a/gcc/doc/invoke.texi
> +++ b/gcc/doc/invoke.texi
> @@ -1436,7 +1436,7 @@ See RS/6000 and PowerPC Options.
> -mavx5124fmaps -mavx512vnni -mavx5124vnniw -mprfchw -mrdpid @gol
> -mrdseed -msgx -mavx512vp2intersect -mserialize -mtsxldtrk@gol
> -mamx-tile -mamx-int8 -mamx-bf16 -muintr -mhreset -mavxvnni@gol
> --mavx512fp16 -mavxifma -mavxvnniint8 @gol
> +-mavx512fp16 -mavxifma -mavxvnniint8 -mavxneconvert @gol
> -mcldemote -mms-bitfields -mno-align-stringops -minline-all-stringops @gol
> -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
> -mkl -mwidekl @gol
> @@ -32899,6 +32899,9 @@ preferred alignment to
> @option{-mpreferred-stack-boundary=2}.
> @need 200
> @itemx -mavxvnniint8
> @opindex mavxvnniint8
> +@need 200
> +@itemx -mavxneconvert
> +@opindex mavxneconvert
> These switches enable the use of instructions in the MMX, SSE,
> SSE2, SSE3, SSSE3, SSE4, SSE4A, SSE4.1, SSE4.2, AVX, AVX2, AVX512F, AVX512PF,
> AVX512ER, AVX512CD, AVX512VL, AVX512BW, AVX512DQ, AVX512IFMA, AVX512VBMI,
> SHA,
> @@ -32909,8 +32912,8 @@ XSAVEOPT, XSAVEC, XSAVES, RTM, HLE, TBM, MWAITX,
> CLZERO, PKU, AVX512VBMI2,
> GFNI, VAES, WAITPKG, VPCLMULQDQ, AVX512BITALG, MOVDIRI, MOVDIR64B,
> AVX512BF16,
> ENQCMD, AVX512VPOPCNTDQ, AVX5124FMAPS, AVX512VNNI, AVX5124VNNIW, SERIALIZE,
> UINTR, HRESET, AMXTILE, AMXINT8, AMXBF16, KL, WIDEKL, AVXVNNI, AVX512FP16,
> -AVXIFMA, AVXVNNIINT8 or CLDEMOTE extended instruction sets. Each has a
> -corresponding @option{-mno-} option to disable use of these instructions.
> +AVXIFMA, AVXVNNIINT8, AVXNECONVERT or CLDEMOTE extended instruction sets.
> Each
> +has a corresponding @option{-mno-} option to disable use of these
> instructions.
>
> These extensions are also available as built-in functions: see
> @ref{x86 Built-in Functions}, for details of the functions enabled and
> diff --git a/gcc/doc/sourcebuild.texi b/gcc/doc/sourcebuild.texi
> index e21a1d381e0..a12175b6498 100644
> --- a/gcc/doc/sourcebuild.texi
> +++ b/gcc/doc/sourcebuild.texi
> @@ -2493,6 +2493,9 @@ Target supports the execution of
> @code{avx512vp2intersect} instructions.
> @item avxifma
> Target supports the execution of @code{avxifma} instructions.
>
> +@item avxneconvert
> +Target supports the execution of @code{avxneconvert} instructions.
> +
> @item avxvnniint8
> Target supports the execution of @code{avxvnniint8} instructions.
>
> diff --git a/gcc/testsuite/g++.dg/other/i386-2.C
> b/gcc/testsuite/g++.dg/other/i386-2.C
> index ebd01fe47bc..dd3e71f25ed 100644
> --- a/gcc/testsuite/g++.dg/other/i386-2.C
> +++ b/gcc/testsuite/g++.dg/other/i386-2.C
> @@ -1,5 +1,5 @@
> /* { dg-do compile { target i?86-*-* x86_64-*-* } } */
> -/* { dg-options "-O -pedantic-errors -march=k8 -msse4a -m3dnow -mavx -mavx2
> -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp
> -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr
> -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha -mprefetchwt1
> -mxsavec -mxsaves -mclflushopt -mavx512dq -mavx512bw -mavx512vl -mavx512ifma
> -mavx512vbmi -mavx512vbmi2 -mavx5124fmaps -mavx5124vnniw -mavx512vpopcntdq
> -mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni -mavx512bitalg -mpconfig
> -mwbnoinvd -mavx512bf16 -menqcmd -mavx512vp2intersect -mserialize -mtsxldtrk
> -mamx-tile -mamx-int8 -mamx-bf16 -mkl -mwidekl -mavxvnni -mavx512fp16
> -mavxifma -mavxvnniint8" } */
> +/* { dg-options "-O -pedantic-errors -march=k8 -msse4a -m3dnow -mavx -mavx2
> -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm -mlwp
> -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr
> -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha -mprefetchwt1
> -mxsavec -mxsaves -mclflushopt -mavx512dq -mavx512bw -mavx512vl -mavx512ifma
> -mavx512vbmi -mavx512vbmi2 -mavx5124fmaps -mavx5124vnniw -mavx512vpopcntdq
> -mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni -mavx512bitalg -mpconfig
> -mwbnoinvd -mavx512bf16 -menqcmd -mavx512vp2intersect -mserialize -mtsxldtrk
> -mamx-tile -mamx-int8 -mamx-bf16 -mkl -mwidekl -mavxvnni -mavx512fp16
> -mavxifma -mavxvnniint8 -mavxneconvert" } */
>
> /* Test that {,x,e,p,t,s,w,a,b,i}mmintrin.h, mm3dnow.h, fma4intrin.h,
> xopintrin.h, abmintrin.h, bmiintrin.h, tbmintrin.h, lwpintrin.h,
> diff --git a/gcc/testsuite/g++.dg/other/i386-3.C
> b/gcc/testsuite/g++.dg/other/i386-3.C
> index b66498f1d4c..cd7045cc4e4 100644
> --- a/gcc/testsuite/g++.dg/other/i386-3.C
> +++ b/gcc/testsuite/g++.dg/other/i386-3.C
> @@ -1,5 +1,5 @@
> /* { dg-do compile { target i?86-*-* x86_64-*-* } } */
> -/* { dg-options "-O -fkeep-inline-functions -march=k8 -msse4a -m3dnow -mavx
> -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm
> -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr
> -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha -mprefetchwt1
> -mxsavec -mxsaves -mclflushopt -mavx512dq -mavx512bw -mavx512vl -mavx512ifma
> -mavx512vbmi -mavx512vbmi2 -mavx5124fmaps -mavx5124vnniw -mavx512vpopcntdq
> -mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni -mavx512bitalg -mpconfig
> -mwbnoinvd -mavx512bf16 -menqcmd -mavx512vp2intersect -mserialize -mtsxldtrk
> -mamx-tile -mamx-int8 -mamx-bf16 -mkl -mwidekl -mavxvnni -mavx512fp16
> -mavxifma -mavxvnniint8" } */
> +/* { dg-options "-O -fkeep-inline-functions -march=k8 -msse4a -m3dnow -mavx
> -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2 -mtbm
> -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx -mfxsr
> -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha -mprefetchwt1
> -mxsavec -mxsaves -mclflushopt -mavx512dq -mavx512bw -mavx512vl -mavx512ifma
> -mavx512vbmi -mavx512vbmi2 -mavx5124fmaps -mavx5124vnniw -mavx512vpopcntdq
> -mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni -mavx512bitalg -mpconfig
> -mwbnoinvd -mavx512bf16 -menqcmd -mavx512vp2intersect -mserialize -mtsxldtrk
> -mamx-tile -mamx-int8 -mamx-bf16 -mkl -mwidekl -mavxvnni -mavx512fp16
> -mavxifma -mavxvnniint8 -mavxneconvert" } */
>
> /* Test that {,x,e,p,t,s,w,a,b,i}mmintrin.h, mm3dnow.h, fma4intrin.h,
> xopintrin.h, abmintrin.h, bmiintrin.h, tbmintrin.h, lwpintrin.h,
> diff --git a/gcc/testsuite/gcc.target/i386/avx-check.h
> b/gcc/testsuite/gcc.target/i386/avx-check.h
> index 77507ca2edc..666eff50780 100644
> --- a/gcc/testsuite/gcc.target/i386/avx-check.h
> +++ b/gcc/testsuite/gcc.target/i386/avx-check.h
> @@ -28,6 +28,9 @@ main ()
> #endif
> #ifdef AVXVNNIINT8
> && __builtin_cpu_supports ("avxvnniint8")
> +#endif
> +#ifdef AVXNECONVERT
> + && __builtin_cpu_supports ("avxneconvert")
> #endif
> )
> {
> diff --git a/gcc/testsuite/gcc.target/i386/avx-ne-convert-1.c
> b/gcc/testsuite/gcc.target/i386/avx-ne-convert-1.c
> new file mode 100644
> index 00000000000..b1848037e81
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/avx-ne-convert-1.c
> @@ -0,0 +1,45 @@
> +/* { dg-do compile } */
> +/* { dg-options "-mavxneconvert -O2" } */
> +/* { dg-final { scan-assembler-times "vbcstnebf162ps\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vbcstnebf162ps\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vbcstnesh2ps\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vbcstnesh2ps\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneebf162ps\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneebf162ps\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneeph2ps\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneeph2ps\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneobf162ps\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneobf162ps\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneoph2ps\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneoph2ps\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "\{vex\} vcvtneps2bf16\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\r]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "\{vex\} vcvtneps2bf16\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\r]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +#include <immintrin.h>
> +
> +volatile __m128 x1;
> +volatile __m256 x2;
> +volatile __m128bf16 res1, res2;
> +const void *a;
> +__m128bf16 *b;
> +__m256bf16 *c;
> +__m128h *d;
> +__m256h *e;
> +
> +void extern
> +avx_ne_convert_test (void)
> +{
> + x1 = _mm_bcstnebf16_ps (a);
> + x2 = _mm256_bcstnebf16_ps (a);
> + x1 = _mm_bcstnesh_ps (a);
> + x2 = _mm256_bcstnesh_ps (a);
> + x1 = _mm_cvtneebf16_ps (b);
> + x2 = _mm256_cvtneebf16_ps (c);
> + x1 = _mm_cvtneeph_ps (d);
> + x2 = _mm256_cvtneeph_ps (e);
> + x1 = _mm_cvtneobf16_ps (b);
> + x2 = _mm256_cvtneobf16_ps (c);
> + x1 = _mm_cvtneoph_ps (d);
> + x2 = _mm256_cvtneoph_ps (e);
> + res1 = _mm_cvtneps_avx_pbh (x1);
> + res2 = _mm256_cvtneps_avx_pbh (x2);
> +}
> diff --git a/gcc/testsuite/gcc.target/i386/avx-ne-convert-vbcstnebf162ps-2.c
> b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vbcstnebf162ps-2.c
> new file mode 100644
> index 00000000000..2707c58f7cd
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vbcstnebf162ps-2.c
> @@ -0,0 +1,54 @@
> +/* { dg-do run } */
> +/* { dg-options "-mavxneconvert -O2" } */
> +/* { dg-require-effective-target avxneconvert } */
> +#define AVXNECONVERT
> +#include <stdint.h>
> +
> +#ifndef CHECK
> +#define CHECK "avx-check.h"
> +#endif
> +
> +#ifndef TEST
> +#define TEST avx_test
> +#endif
> +
> +#include CHECK
> +
> +typedef union
> +{
> + uint32_t int32;
> + float flt;
> +} float_int_t;
> +
> +static uint16_t convert_fp32_to_bf16 (float fp)
> +{
> + float_int_t fi;
> + fi.flt = fp;
> + return ((fi.int32 >> 16) & 0xffff);
> +}
> +
> +void TEST (void)
> +{
> + union128 dst_128;
> + union256 dst_256;
> + float res_ref_128[4], res_ref_256[8], fp32;
> + uint16_t var;
> + fp32 = (float) 3 * 2 + 5.5;
> + for (int i = 0; i < 4; i++)
> + {
> + res_ref_128[i] = fp32;
> + dst_128.a[i] = 117;
> + }
> + for (int i = 0; i < 8; i++)
> + {
> + res_ref_256[i] = fp32;
> + dst_256.a[i] = 117;
> + }
> + var = convert_fp32_to_bf16 (fp32);
> + dst_128.x = _mm_bcstnebf16_ps (&var);
> + dst_256.x = _mm256_bcstnebf16_ps (&var);
> + if (check_union128 (dst_128, res_ref_128))
> + abort();
> + if (check_union256 (dst_256, res_ref_256))
> + abort();
> +}
> diff --git a/gcc/testsuite/gcc.target/i386/avx-ne-convert-vbcstnesh2ps-2.c
> b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vbcstnesh2ps-2.c
> new file mode 100644
> index 00000000000..0e6f38334b8
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vbcstnesh2ps-2.c
> @@ -0,0 +1,42 @@
> +/* { dg-do run } */
> +/* { dg-options "-mavxneconvert -mf16c -O2" } */
> +/* { dg-require-effective-target avxneconvert } */
> +#define AVXNECONVERT
> +#include <stdint.h>
> +#include <immintrin.h>
> +
> +#ifndef CHECK
> +#define CHECK "avx-check.h"
> +#endif
> +
> +#ifndef TEST
> +#define TEST avx_test
> +#endif
> +
> +#include CHECK
> +
> +void TEST (void)
> +{
> + union128 dst_128;
> + union256 dst_256;
> + float res_ref_128[4], res_ref_256[8], fp32;
> + uint16_t var;
> + fp32 = (float) 3 * 2 + 8.5;
> + for (int i = 0; i < 4; i++)
> + {
> + res_ref_128[i] = fp32;
> + dst_128.a[i] = 117;
> + }
> + for (int i = 0; i < 8; i++)
> + {
> + res_ref_256[i] = fp32;
> + dst_256.a[i] = 117;
> + }
> + var = _cvtss_sh (fp32, 0);
> + dst_128.x = _mm_bcstnesh_ps (&var);
> + dst_256.x = _mm256_bcstnesh_ps (&var);
> + if (check_union128 (dst_128, res_ref_128))
> + abort();
> + if (check_union256 (dst_256, res_ref_256))
> + abort();
> +}
> diff --git a/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneebf162ps-2.c
> b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneebf162ps-2.c
> new file mode 100644
> index 00000000000..c80f3fdedec
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneebf162ps-2.c
> @@ -0,0 +1,73 @@
> +/* { dg-do run } */
> +/* { dg-options "-mavxneconvert -O2" } */
> +/* { dg-require-effective-target avxneconvert } */
> +#define AVXNECONVERT
> +#include <stdint.h>
> +
> +#ifndef CHECK
> +#define CHECK "avx-check.h"
> +#endif
> +
> +#ifndef TEST
> +#define TEST avx_test
> +#endif
> +
> +#include CHECK
> +
> +typedef union
> +{
> + uint32_t int32;
> + float flt;
> +} float_int_t;
> +
> +typedef union
> +{
> + __m128bf16 x;
> + uint32_t a[4];
> +} union128bf16_i;
> +
> +typedef union
> +{
> + __m256bf16 x;
> + uint32_t a[8];
> +} union256bf16_i;
> +
> +static uint16_t convert_fp32_to_bf16 (float fp)
> +{
> + float_int_t fi;
> + fi.flt = fp;
> + return ((fi.int32 >> 16) & 0xffff);
> +}
> +
> +void TEST (void)
> +{
> + union128 dst_128;
> + union256 dst_256;
> + float res_ref_128[4], res_ref_256[8], fp32;
> + uint16_t bf16;
> + union128bf16_i src_128bh;
> + union256bf16_i src_256bh;
> +
> + for (int i = 0; i < 4; i++)
> + {
> + fp32 = (float) 3 * i + 5 + i * 0.5;
> + bf16 = convert_fp32_to_bf16 (fp32);
> + src_128bh.a[i] = bf16; // store bf16 at the lower part of the dword
> + res_ref_128[i] = fp32;
> + dst_128.a[i] = 117;
> + }
> + for (int i = 0; i < 8; i++)
> + {
> + fp32 = (float) 3 * i + 5 + i * 0.5;
> + bf16 = convert_fp32_to_bf16 (fp32);
> + src_256bh.a[i] = bf16; // store bf16 at the lower part of the dword
> + res_ref_256[i] = fp32;
> + dst_256.a[i] = 117;
> + }
> + dst_128.x = _mm_cvtneebf16_ps (&src_128bh.x);
> + dst_256.x = _mm256_cvtneebf16_ps (&src_256bh.x);
> + if (check_union128 (dst_128, res_ref_128))
> + abort();
> + if (check_union256 (dst_256, res_ref_256))
> + abort();
> +}
> diff --git a/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneeph2ps-2.c
> b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneeph2ps-2.c
> new file mode 100644
> index 00000000000..a862894746d
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneeph2ps-2.c
> @@ -0,0 +1,66 @@
> +/* { dg-do run } */
> +/* { dg-options "-mavxneconvert -mf16c -O2" } */
> +/* { dg-require-effective-target avxneconvert } */
> +#define AVXNECONVERT
> +#include <stdint.h>
> +
> +#ifndef CHECK
> +#define CHECK "avx-check.h"
> +#endif
> +
> +#ifndef TEST
> +#define TEST avx_test
> +#endif
> +
> +#include CHECK
> +
> +typedef union
> +{
> + uint32_t int32;
> + float flt;
> +} float_int_t;
> +
> +typedef union
> +{
> + __m128h x;
> + uint32_t a[4];
> +} union128h;
> +
> +typedef union
> +{
> + __m256h x;
> + uint32_t a[8];
> +} union256h;
> +
> +void TEST (void)
> +{
> + union128 dst_128;
> + union256 dst_256;
> + float res_ref_128[4], res_ref_256[8], fp32;
> + uint16_t fp16;
> + union128h src_128h;
> + union256h src_256h;
> +
> + for (int i = 0; i < 4; i++)
> + {
> + fp32 = (float) 3 * i + 5 + i * 0.5;
> + fp16 = _cvtss_sh (fp32, 0);
> + src_128h.a[i] = fp16;
> + res_ref_128[i] = fp32;
> + dst_128.a[i] = 117;
> + }
> + for (int i = 0; i < 8; i++)
> + {
> + fp32 = (float) 3 * i + 5 + i * 0.5;
> + fp16 = _cvtss_sh (fp32, 0);
> + src_256h.a[i] = fp16;
> + res_ref_256[i] = fp32;
> + dst_256.a[i] = 117;
> + }
> + dst_128.x = _mm_cvtneeph_ps (&src_128h.x);
> + dst_256.x = _mm256_cvtneeph_ps (&src_256h.x);
> + if (check_union128 (dst_128, res_ref_128))
> + abort();
> + if (check_union256 (dst_256, res_ref_256))
> + abort();
> +}
> diff --git a/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneobf162ps-2.c
> b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneobf162ps-2.c
> new file mode 100644
> index 00000000000..d95aee067ae
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneobf162ps-2.c
> @@ -0,0 +1,75 @@
> +/* { dg-do run } */
> +/* { dg-options "-mavxneconvert -O2" } */
> +/* { dg-require-effective-target avxneconvert } */
> +#define AVXNECONVERT
> +#include <stdint.h>
> +
> +#ifndef CHECK
> +#define CHECK "avx-check.h"
> +#endif
> +
> +#ifndef TEST
> +#define TEST avx_test
> +#endif
> +
> +#include CHECK
> +
> +typedef union
> +{
> + uint32_t int32;
> + float flt;
> +} float_int_t;
> +
> +typedef union
> +{
> + __m128bf16 x;
> + uint32_t a[4];
> +} union128bf16_i;
> +
> +typedef union
> +{
> + __m256bf16 x;
> + uint32_t a[8];
> +} union256bf16_i;
> +
> +static uint16_t convert_fp32_to_bf16 (float fp)
> +{
> + float_int_t fi;
> + fi.flt = fp;
> + return ((fi.int32 >> 16) & 0xffff);
> +}
> +
> +void TEST (void)
> +{
> + union128 dst_128;
> + union256 dst_256;
> + float res_ref_128[4], res_ref_256[8], fp32;
> + uint16_t bf16;
> + union128bf16_i src_128bh;
> + union256bf16_i src_256bh;
> +
> + for (int i = 0; i < 4; i++)
> + {
> + fp32 = (float) 3 * i + 5 + i * 0.5;
> + bf16 = convert_fp32_to_bf16 (fp32);
> + // store bf16 at the upper part of the dword
> + src_128bh.a[i] = (bf16 << 16) & 0xffff0000;
> + res_ref_128[i] = fp32;
> + dst_128.a[i] = 117;
> + }
> + for (int i = 0; i < 8; i++)
> + {
> + fp32 = (float) 3 * i + 5 + i * 0.5;
> + bf16 = convert_fp32_to_bf16 (fp32);
> + // store bf16 at the upper part of the dword
> + src_256bh.a[i] = (bf16 << 16) & 0xffff0000;
> + res_ref_256[i] = fp32;
> + dst_256.a[i] = 117;
> + }
> + dst_128.x = _mm_cvtneobf16_ps (&src_128bh.x);
> + dst_256.x = _mm256_cvtneobf16_ps (&src_256bh.x);
> + if (check_union128 (dst_128, res_ref_128))
> + abort();
> + if (check_union256 (dst_256, res_ref_256))
> + abort();
> +}
> diff --git a/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneoph2ps-2.c
> b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneoph2ps-2.c
> new file mode 100644
> index 00000000000..95eb5d74765
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneoph2ps-2.c
> @@ -0,0 +1,66 @@
> +/* { dg-do run } */
> +/* { dg-options "-mavxneconvert -mf16c -O2" } */
> +/* { dg-require-effective-target avxneconvert } */
> +#define AVXNECONVERT
> +#include <stdint.h>
> +
> +#ifndef CHECK
> +#define CHECK "avx-check.h"
> +#endif
> +
> +#ifndef TEST
> +#define TEST avx_test
> +#endif
> +
> +#include CHECK
> +
> +typedef union
> +{
> + uint32_t int32;
> + float flt;
> +} float_int_t;
> +
> +typedef union
> +{
> + __m128h x;
> + uint32_t a[4];
> +} union128h;
> +
> +typedef union
> +{
> + __m256h x;
> + uint32_t a[8];
> +} union256h;
> +
> +void TEST (void)
> +{
> + union128 dst_128;
> + union256 dst_256;
> + float res_ref_128[4], res_ref_256[8], fp32;
> + uint16_t fp16;
> + union128h src_128h;
> + union256h src_256h;
> +
> + for (int i = 0; i < 4; i++)
> + {
> + fp32 = (float) 3 * i + 5 + i * 0.5;
> + fp16 = _cvtss_sh (fp32, 0);
> + src_128h.a[i] = fp16 << 16;
> + res_ref_128[i] = fp32;
> + dst_128.a[i] = 117;
> + }
> + for (int i = 0; i < 8; i++)
> + {
> + fp32 = (float) 3 * i + 5 + i * 0.5;
> + fp16 = _cvtss_sh (fp32, 0);
> + src_256h.a[i] = fp16 << 16;
> + res_ref_256[i] = fp32;
> + dst_256.a[i] = 117;
> + }
> + dst_128.x = _mm_cvtneoph_ps (&src_128h.x);
> + dst_256.x = _mm256_cvtneoph_ps (&src_256h.x);
> + if (check_union128 (dst_128, res_ref_128))
> + abort();
> + if (check_union256 (dst_256, res_ref_256))
> + abort();
> +}
> diff --git a/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneps2bf16-2.c
> b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneps2bf16-2.c
> new file mode 100644
> index 00000000000..0861521111a
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/avx-ne-convert-vcvtneps2bf16-2.c
> @@ -0,0 +1,58 @@
> +/* { dg-do run } */
> +/* { dg-options "-mavxneconvert -O2" } */
> +/* { dg-require-effective-target avxneconvert } */
> +#define AVXNECONVERT
> +#include <stdint.h>
> +
> +#ifndef CHECK
> +#define CHECK "avx-check.h"
> +#endif
> +
> +#ifndef TEST
> +#define TEST avx_test
> +#endif
> +
> +#include CHECK
> +
> +typedef union
> +{
> + uint32_t int32;
> + float flt;
> +} float_int_t;
> +
> +typedef union
> +{
> + __m128bf16 x;
> + unsigned short a[8];
> +} union128bf16;
> +
> +void TEST (void)
> +{
> + union128 src_128;
> + union256 src_256;
> + union128bf16 dst_128, dst_256;
> + uint16_t res_ref_128[8] = {0}, res_ref_256[8];
> + float_int_t fp32;
> + for (int i = 0; i < 4; i++)
> + {
> + fp32.flt = (float) 2 * i + 7 + i * 0.25;
> + src_128.a[i] = fp32.flt;
> + res_ref_128[i] = fp32.int32 >> 16;
> + dst_128.a[i] = 117;
> + }
> +
> + for (int i = 0; i < 8; i++)
> + {
> + fp32.flt = (float) 2 * i + 7 + i * 0.25;
> + src_256.a[i] = fp32.flt;
> + res_ref_256[i] = fp32.int32 >> 16;
> + dst_256.a[i] = 117;
> + }
> + dst_128.x = _mm_cvtneps_avx_pbh (src_128.x);
> + dst_256.x = _mm256_cvtneps_avx_pbh (src_256.x);
> +
> + if (checkVus (dst_128.a, res_ref_128, 8))
> + abort();
> + if (checkVus (dst_128.a, res_ref_128, 8))
> + abort();
> +}
> diff --git a/gcc/testsuite/gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1.c
> b/gcc/testsuite/gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1a.c
> similarity index 100%
> rename from gcc/testsuite/gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1.c
> rename to gcc/testsuite/gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1a.c
> diff --git a/gcc/testsuite/gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1b.c
> b/gcc/testsuite/gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1b.c
> new file mode 100644
> index 00000000000..8b5d6a644bc
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/avx512bf16vl-vcvtneps2bf16-1b.c
> @@ -0,0 +1,27 @@
> +/* { dg-do compile } */
> +/* { dg-options "-mavx512bf16 -mavx512vl -mavxneconvert -O2" } */
> +/* { dg-final { scan-assembler-times "\{vex\} vcvtneps2bf16\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\r]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneps2bf16\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\r]*%xmm\[0-9\]+\{%k\[0-9\]\}(?:\n|\[
> \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneps2bf16\[
> \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\r]*%xmm\[0-9\]+\{%k\[0-9\]\}\{z\}\[^\n\r]*(?:\n|\[
> \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "\{vex\} vcvtneps2bf16\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\r]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneps2bf16\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\r]*%xmm\[0-9\]+\{%k\[0-9\]\}(?:\n|\[
> \\t\]+#)" 1 } } */
> +/* { dg-final { scan-assembler-times "vcvtneps2bf16\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\r]*%xmm\[0-9\]+\{%k\[0-9\]\}\{z\}\[^\n\r]*(?:\n|\[
> \\t\]+#)" 1 } } */
> +
> +#include <immintrin.h>
> +
> +volatile __m128bh res1, res2;
> +volatile __m128 x1;
> +volatile __m256 x2;
> +volatile __mmask8 m8;
> +
> +void extern
> +avx512bf16_test (void)
> +{
> + res2 = _mm256_cvtneps_pbh (x2);
> + res2 = _mm256_mask_cvtneps_pbh (res2, m8, x2);
> + res2 = _mm256_maskz_cvtneps_pbh (m8, x2);
> +
> + res1 = _mm_cvtneps_pbh (x1);
> + res1 = _mm_mask_cvtneps_pbh (res1, m8, x1);
> + res1 = _mm_maskz_cvtneps_pbh (m8, x1);
> +}
> diff --git a/gcc/testsuite/gcc.target/i386/funcspec-56.inc
> b/gcc/testsuite/gcc.target/i386/funcspec-56.inc
> index a681bffe3e7..b3d33df7c9c 100644
> --- a/gcc/testsuite/gcc.target/i386/funcspec-56.inc
> +++ b/gcc/testsuite/gcc.target/i386/funcspec-56.inc
> @@ -82,6 +82,7 @@ extern void test_avxvnni (void)
> __attribute__((__target__("avxvnni")));
> extern void test_avx512fp16 (void)
> __attribute__((__target__("avx512fp16")));
> extern void test_avxifma (void)
> __attribute__((__target__("avxifma")));
> extern void test_avxvnniint8 (void)
> __attribute__((__target__("avxvnniint8")));
> +extern void test_avxneconvert (void)
> __attribute__((__target__("avxneconvert")));
>
> extern void test_no_sgx (void)
> __attribute__((__target__("no-sgx")));
> extern void test_no_avx5124fmaps(void)
> __attribute__((__target__("no-avx5124fmaps")));
> @@ -165,6 +166,7 @@ extern void test_no_avxvnni (void)
> __attribute__((__target__("no-avxvnni")));
> extern void test_no_avx512fp16 (void)
> __attribute__((__target__("no-avx512fp16")));
> extern void test_no_avxifma (void)
> __attribute__((__target__("no-avxifma")));
> extern void test_no_avxvnniint8 (void)
> __attribute__((__target__("no-avxvnniint8")));
> +extern void test_no_avxneconvert (void)
> __attribute__((__target__("no-avxneconvert")));
>
> extern void test_arch_nocona (void)
> __attribute__((__target__("arch=nocona")));
> extern void test_arch_core2 (void)
> __attribute__((__target__("arch=core2")));
> diff --git a/gcc/testsuite/gcc.target/i386/sse-12.c
> b/gcc/testsuite/gcc.target/i386/sse-12.c
> index ddde2df6657..3eabc49a6ab 100644
> --- a/gcc/testsuite/gcc.target/i386/sse-12.c
> +++ b/gcc/testsuite/gcc.target/i386/sse-12.c
> @@ -3,7 +3,7 @@
> popcntintrin.h gfniintrin.h and mm_malloc.h are usable
> with -O -std=c89 -pedantic-errors. */
> /* { dg-do compile } */
> -/* { dg-options "-O -std=c89 -pedantic-errors -march=k8 -msse4a -m3dnow
> -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2
> -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx
> -mfxsr -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha
> -mprefetchwt1 -mxsavec -mxsaves -mclflushopt -mavx512bw -mavx512dq -mavx512vl
> -mavx512vbmi -mavx512vbmi2 -mavx512ifma -mavx5124fmaps -mavx5124vnniw
> -mavx512vpopcntdq -mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni
> -mavx512bitalg -mpconfig -mwbnoinvd -mavx512bf16 -menqcmd
> -mavx512vp2intersect -mserialize -mtsxldtrk -mamx-tile -mamx-int8 -mamx-bf16
> -mkl -mwidekl -mavxvnni -mavxifma -mavxvnniint8" } */
> +/* { dg-options "-O -std=c89 -pedantic-errors -march=k8 -msse4a -m3dnow
> -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -mbmi2
> -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw -madx
> -mfxsr -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha
> -mprefetchwt1 -mxsavec -mxsaves -mclflushopt -mavx512bw -mavx512dq -mavx512vl
> -mavx512vbmi -mavx512vbmi2 -mavx512ifma -mavx5124fmaps -mavx5124vnniw
> -mavx512vpopcntdq -mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni
> -mavx512bitalg -mpconfig -mwbnoinvd -mavx512bf16 -menqcmd
> -mavx512vp2intersect -mserialize -mtsxldtrk -mamx-tile -mamx-int8 -mamx-bf16
> -mkl -mwidekl -mavxvnni -mavxifma -mavxvnniint8 -mavxneconvert" } */
>
> #include <x86intrin.h>
>
> diff --git a/gcc/testsuite/gcc.target/i386/sse-13.c
> b/gcc/testsuite/gcc.target/i386/sse-13.c
> index 2b293216c6f..b9cdfb690d1 100644
> --- a/gcc/testsuite/gcc.target/i386/sse-13.c
> +++ b/gcc/testsuite/gcc.target/i386/sse-13.c
> @@ -1,5 +1,5 @@
> /* { dg-do compile } */
> -/* { dg-options "-O2 -Werror-implicit-function-declaration -march=k8 -msse4a
> -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi
> -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw
> -madx -mfxsr -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha
> -mprefetchwt1 -mxsavec -mxsaves -mclflushopt -mavx512vl -mavx512dq -mavx512bw
> -mavx512vbmi -mavx512vbmi2 -mavx512ifma -mavx5124fmaps -mavx5124vnniw
> -mavx512vpopcntdq -mavx512vp2intersect -mclwb -mmwaitx -mclzero -mpku -msgx
> -mrdpid -mgfni -mavx512bitalg -mpconfig -mwbnoinvd -mavx512bf16 -menqcmd
> -mserialize -mtsxldtrk -mamx-tile -mamx-int8 -mamx-bf16 -mkl -mwidekl
> -mavxvnni -mavx512fp16 -mavxifma -mavxvnniint8" } */
> +/* { dg-options "-O2 -Werror-implicit-function-declaration -march=k8 -msse4a
> -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi
> -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw
> -madx -mfxsr -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha
> -mprefetchwt1 -mxsavec -mxsaves -mclflushopt -mavx512vl -mavx512dq -mavx512bw
> -mavx512vbmi -mavx512vbmi2 -mavx512ifma -mavx5124fmaps -mavx5124vnniw
> -mavx512vpopcntdq -mavx512vp2intersect -mclwb -mmwaitx -mclzero -mpku -msgx
> -mrdpid -mgfni -mavx512bitalg -mpconfig -mwbnoinvd -mavx512bf16 -menqcmd
> -mserialize -mtsxldtrk -mamx-tile -mamx-int8 -mamx-bf16 -mkl -mwidekl
> -mavxvnni -mavx512fp16 -mavxifma -mavxvnniint8 -mavxneconvert" } */
> /* { dg-add-options bind_pic_locally } */
>
> #include <mm_malloc.h>
> diff --git a/gcc/testsuite/gcc.target/i386/sse-14.c
> b/gcc/testsuite/gcc.target/i386/sse-14.c
> index 78b51048b90..b6ee3806dcc 100644
> --- a/gcc/testsuite/gcc.target/i386/sse-14.c
> +++ b/gcc/testsuite/gcc.target/i386/sse-14.c
> @@ -1,5 +1,5 @@
> /* { dg-do compile } */
> -/* { dg-options "-O0 -Werror-implicit-function-declaration -march=k8 -msse4a
> -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi
> -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw
> -madx -mfxsr -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha
> -mprefetchwt1 -mxsavec -mxsaves -mclflushopt -mavx512dq -mavx512bw -mavx512vl
> -mavx512ifma -mavx512vbmi -mavx512vbmi2 -mavx5124fmaps -mavx5124vnniw
> -mavx512vpopcntdq -mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni
> -mpconfig -mwbnoinvd -mavx512vl -mavx512bf16 -menqcmd -mavx512vp2intersect
> -mserialize -mtsxldtrk -mamx-tile -mamx-int8 -mamx-bf16 -mkl -mwidekl
> -mavxvnni -mavx512fp16 -mavxifma -mavxvnniint8" } */
> +/* { dg-options "-O0 -Werror-implicit-function-declaration -march=k8 -msse4a
> -m3dnow -mavx -mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi
> -mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -mprfchw
> -madx -mfxsr -mxsaveopt -mavx512f -mavx512er -mavx512cd -mavx512pf -msha
> -mprefetchwt1 -mxsavec -mxsaves -mclflushopt -mavx512dq -mavx512bw -mavx512vl
> -mavx512ifma -mavx512vbmi -mavx512vbmi2 -mavx5124fmaps -mavx5124vnniw
> -mavx512vpopcntdq -mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni
> -mpconfig -mwbnoinvd -mavx512vl -mavx512bf16 -menqcmd -mavx512vp2intersect
> -mserialize -mtsxldtrk -mamx-tile -mamx-int8 -mamx-bf16 -mkl -mwidekl
> -mavxvnni -mavx512fp16 -mavxifma -mavxvnniint8 -mavxneconvert" } */
> /* { dg-add-options bind_pic_locally } */
>
> #include <mm_malloc.h>
> diff --git a/gcc/testsuite/gcc.target/i386/sse-22.c
> b/gcc/testsuite/gcc.target/i386/sse-22.c
> index cc1c8cfa4be..71ac0f3da19 100644
> --- a/gcc/testsuite/gcc.target/i386/sse-22.c
> +++ b/gcc/testsuite/gcc.target/i386/sse-22.c
> @@ -103,7 +103,7 @@
>
>
> #ifndef DIFFERENT_PRAGMAS
> -#pragma GCC target
> ("sse4a,3dnow,avx,avx2,fma4,xop,aes,pclmul,popcnt,abm,lzcnt,bmi,bmi2,tbm,lwp,fsgsbase,rdrnd,f16c,rtm,rdseed,prfchw,adx,fxsr,xsaveopt,avx512f,avx512er,avx512cd,avx512pf,sha,prefetchwt1,avx512vl,avx512bw,avx512dq,avx512vbmi,avx512vbmi2,avx512ifma,avx5124fmaps,avx5124vnniw,avx512vpopcntdq,gfni,avx512bitalg,avx512bf16,avx512vp2intersect,serialize,tsxldtrk,amx-tile,amx-int8,amx-bf16,kl,widekl,avxvnni,avx512fp16,avxifma,avxvnniint8")
> +#pragma GCC target
> ("sse4a,3dnow,avx,avx2,fma4,xop,aes,pclmul,popcnt,abm,lzcnt,bmi,bmi2,tbm,lwp,fsgsbase,rdrnd,f16c,rtm,rdseed,prfchw,adx,fxsr,xsaveopt,avx512f,avx512er,avx512cd,avx512pf,sha,prefetchwt1,avx512vl,avx512bw,avx512dq,avx512vbmi,avx512vbmi2,avx512ifma,avx5124fmaps,avx5124vnniw,avx512vpopcntdq,gfni,avx512bitalg,avx512bf16,avx512vp2intersect,serialize,tsxldtrk,amx-tile,amx-int8,amx-bf16,kl,widekl,avxvnni,avx512fp16,avxifma,avxvnniint8,avxneconvert")
> #endif
>
> /* Following intrinsics require immediate arguments. They
> @@ -220,7 +220,7 @@ test_4 (_mm_cmpestrz, int, __m128i, int, __m128i, int, 1)
>
> /* immintrin.h (AVX/AVX2/RDRND/FSGSBASE/F16C/RTM/AVX512F/SHA) */
> #ifdef DIFFERENT_PRAGMAS
> -#pragma GCC target
> ("avx,avx2,rdrnd,fsgsbase,f16c,rtm,avx512f,avx512er,avx512cd,avx512pf,sha,avx512vl,avx512bw,avx512dq,avx512ifma,avx512vbmi,avx512vbmi2,avx5124fmaps,avx5124vnniw,avx512vpopcntdq,gfni,avx512bitalg,avx512bf16,avx512vp2intersect,serialize,tsxldtrk,amx-tile,amx-int8,amx-bf16,kl,widekl,avxvnni,avx512fp16,avxifma,avxvnniint8")
> +#pragma GCC target
> ("avx,avx2,rdrnd,fsgsbase,f16c,rtm,avx512f,avx512er,avx512cd,avx512pf,sha,avx512vl,avx512bw,avx512dq,avx512ifma,avx512vbmi,avx512vbmi2,avx5124fmaps,avx5124vnniw,avx512vpopcntdq,gfni,avx512bitalg,avx512bf16,avx512vp2intersect,serialize,tsxldtrk,amx-tile,amx-int8,amx-bf16,kl,widekl,avxvnni,avx512fp16,avxifma,avxvnniint8,avxneconvert")
> #endif
> #include <immintrin.h>
> test_1 (_cvtss_sh, unsigned short, float, 1)
> diff --git a/gcc/testsuite/gcc.target/i386/sse-23.c
> b/gcc/testsuite/gcc.target/i386/sse-23.c
> index 270f4483491..898dde80c8f 100644
> --- a/gcc/testsuite/gcc.target/i386/sse-23.c
> +++ b/gcc/testsuite/gcc.target/i386/sse-23.c
> @@ -843,6 +843,6 @@
> #define __builtin_ia32_vpclmulqdq_v2di(A, B, C)
> __builtin_ia32_vpclmulqdq_v2di(A, B, 1)
> #define __builtin_ia32_vpclmulqdq_v8di(A, B, C)
> __builtin_ia32_vpclmulqdq_v8di(A, B, 1)
>
> -#pragma GCC target
> ("sse4a,3dnow,avx,avx2,fma4,xop,aes,pclmul,popcnt,abm,lzcnt,bmi,bmi2,tbm,lwp,fsgsbase,rdrnd,f16c,fma,rtm,rdseed,prfchw,adx,fxsr,xsaveopt,avx512f,avx512er,avx512cd,avx512pf,sha,prefetchwt1,xsavec,xsaves,clflushopt,avx512bw,avx512dq,avx512vl,avx512vbmi,avx512ifma,avx5124fmaps,avx5124vnniw,avx512vpopcntdq,clwb,mwaitx,clzero,pku,sgx,rdpid,gfni,avx512vbmi2,vpclmulqdq,avx512bitalg,pconfig,wbnoinvd,avx512bf16,enqcmd,avx512vp2intersect,serialize,tsxldtrk,amx-tile,amx-int8,amx-bf16,kl,widekl,avxvnni,avx512fp16,avxifma,avxvnniint8")
> +#pragma GCC target
> ("sse4a,3dnow,avx,avx2,fma4,xop,aes,pclmul,popcnt,abm,lzcnt,bmi,bmi2,tbm,lwp,fsgsbase,rdrnd,f16c,fma,rtm,rdseed,prfchw,adx,fxsr,xsaveopt,avx512f,avx512er,avx512cd,avx512pf,sha,prefetchwt1,xsavec,xsaves,clflushopt,avx512bw,avx512dq,avx512vl,avx512vbmi,avx512ifma,avx5124fmaps,avx5124vnniw,avx512vpopcntdq,clwb,mwaitx,clzero,pku,sgx,rdpid,gfni,avx512vbmi2,vpclmulqdq,avx512bitalg,pconfig,wbnoinvd,avx512bf16,enqcmd,avx512vp2intersect,serialize,tsxldtrk,amx-tile,amx-int8,amx-bf16,kl,widekl,avxvnni,avx512fp16,avxifma,avxvnniint8,avxneconvert")
>
> #include <x86intrin.h>
> diff --git a/gcc/testsuite/lib/target-supports.exp
> b/gcc/testsuite/lib/target-supports.exp
> index 64ccfc746bd..9228e810c45 100644
> --- a/gcc/testsuite/lib/target-supports.exp
> +++ b/gcc/testsuite/lib/target-supports.exp
> @@ -9530,6 +9530,18 @@ proc check_effective_target_avxvnniint8 { } {
> } "-O0 -mavxvnniint8" ]
> }
>
> +# Return 1 if avxneconvert instructions can be compiled.
> +proc check_effective_target_avxneconvert { } {
> + return [check_no_compiler_messages avxneconvert object {
> + typedef float __m128 __attribute__ ((__vector_size__ (16),
> __may_alias__));
> + __m128
> + _mm_bcstnebf16_ps (const void *__P)
> + {
> + return (__m128) __builtin_ia32_vbcstnebf162ps128 ((const short *)
> __P);
> + }
> + } "-O0 -mavxneconvert" ]
> +}
> +
> # Return 1 if sse instructions can be compiled.
> proc check_effective_target_sse { } {
> return [check_no_compiler_messages sse object {
> --
> 2.18.1
>
--
BR,
Hongtao
