diff options
| author | Danny <danny@kdrag0n.dev> | 2021-01-09 23:34:32 +0000 |
|---|---|---|
| committer | mosimchah <mosimchah@gmail.com> | 2021-01-22 03:35:20 -0800 |
| commit | 783d21ff74759076d2fc503685ca47d2c29baea3 (patch) | |
| tree | d650cc46cbf7ca53f15c77ced2682e97d492c068 /lib/clang/12.0.0/include/fmaintrin.h | |
| parent | fdbc6f7102056fb52d26bfb2cbc6ea317890ee34 (diff) | |
LLVM commit: https://github.com/llvm/llvm-project/commit/b02eab9058e58782fca32dd8b1e53c27ed93f866
binutils version: 2.35.1
Builder commit: https://github.com/kdrag0n/proton-clang-build/commit/ba42f701467c9103f23fbb90aca4b23858221ee2
Diffstat (limited to 'lib/clang/12.0.0/include/fmaintrin.h')
| -rw-r--r-- | lib/clang/12.0.0/include/fmaintrin.h | 216 |
1 files changed, 216 insertions, 0 deletions
diff --git a/lib/clang/12.0.0/include/fmaintrin.h b/lib/clang/12.0.0/include/fmaintrin.h new file mode 100644 index 0000000..d889b7c --- /dev/null +++ b/lib/clang/12.0.0/include/fmaintrin.h @@ -0,0 +1,216 @@ +/*===---- fmaintrin.h - FMA 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 <fmaintrin.h> directly; include <immintrin.h> instead." +#endif + +#ifndef __FMAINTRIN_H +#define __FMAINTRIN_H + +/* Define the default attributes for the functions in this file. */ +#define __DEFAULT_FN_ATTRS128 __attribute__((__always_inline__, __nodebug__, __target__("fma"), __min_vector_width__(128))) +#define __DEFAULT_FN_ATTRS256 __attribute__((__always_inline__, __nodebug__, __target__("fma"), __min_vector_width__(256))) + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fmadd_ps(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fmadd_pd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddpd((__v2df)__A, (__v2df)__B, (__v2df)__C); +} + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fmadd_ss(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddss3((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fmadd_sd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddsd3((__v2df)__A, (__v2df)__B, (__v2df)__C); +} + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fmsub_ps(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddps((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fmsub_pd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddpd((__v2df)__A, (__v2df)__B, -(__v2df)__C); +} + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fmsub_ss(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddss3((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fmsub_sd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddsd3((__v2df)__A, (__v2df)__B, -(__v2df)__C); +} + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fnmadd_ps(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddps(-(__v4sf)__A, (__v4sf)__B, (__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fnmadd_pd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddpd(-(__v2df)__A, (__v2df)__B, (__v2df)__C); +} + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fnmadd_ss(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddss3((__v4sf)__A, -(__v4sf)__B, (__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fnmadd_sd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddsd3((__v2df)__A, -(__v2df)__B, (__v2df)__C); +} + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fnmsub_ps(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddps(-(__v4sf)__A, (__v4sf)__B, -(__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fnmsub_pd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddpd(-(__v2df)__A, (__v2df)__B, -(__v2df)__C); +} + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fnmsub_ss(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddss3((__v4sf)__A, -(__v4sf)__B, -(__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fnmsub_sd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddsd3((__v2df)__A, -(__v2df)__B, -(__v2df)__C); +} + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fmaddsub_ps(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddsubps((__v4sf)__A, (__v4sf)__B, (__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fmaddsub_pd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddsubpd((__v2df)__A, (__v2df)__B, (__v2df)__C); +} + +static __inline__ __m128 __DEFAULT_FN_ATTRS128 +_mm_fmsubadd_ps(__m128 __A, __m128 __B, __m128 __C) +{ + return (__m128)__builtin_ia32_vfmaddsubps((__v4sf)__A, (__v4sf)__B, -(__v4sf)__C); +} + +static __inline__ __m128d __DEFAULT_FN_ATTRS128 +_mm_fmsubadd_pd(__m128d __A, __m128d __B, __m128d __C) +{ + return (__m128d)__builtin_ia32_vfmaddsubpd((__v2df)__A, (__v2df)__B, -(__v2df)__C); +} + +static __inline__ __m256 __DEFAULT_FN_ATTRS256 +_mm256_fmadd_ps(__m256 __A, __m256 __B, __m256 __C) +{ + return (__m256)__builtin_ia32_vfmaddps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C); +} + +static __inline__ __m256d __DEFAULT_FN_ATTRS256 +_mm256_fmadd_pd(__m256d __A, __m256d __B, __m256d __C) +{ + return (__m256d)__builtin_ia32_vfmaddpd256((__v4df)__A, (__v4df)__B, (__v4df)__C); +} + +static __inline__ __m256 __DEFAULT_FN_ATTRS256 +_mm256_fmsub_ps(__m256 __A, __m256 __B, __m256 __C) +{ + return (__m256)__builtin_ia32_vfmaddps256((__v8sf)__A, (__v8sf)__B, -(__v8sf)__C); +} + +static __inline__ __m256d __DEFAULT_FN_ATTRS256 +_mm256_fmsub_pd(__m256d __A, __m256d __B, __m256d __C) +{ + return (__m256d)__builtin_ia32_vfmaddpd256((__v4df)__A, (__v4df)__B, -(__v4df)__C); +} + +static __inline__ __m256 __DEFAULT_FN_ATTRS256 +_mm256_fnmadd_ps(__m256 __A, __m256 __B, __m256 __C) +{ + return (__m256)__builtin_ia32_vfmaddps256(-(__v8sf)__A, (__v8sf)__B, (__v8sf)__C); +} + +static __inline__ __m256d __DEFAULT_FN_ATTRS256 +_mm256_fnmadd_pd(__m256d __A, __m256d __B, __m256d __C) +{ + return (__m256d)__builtin_ia32_vfmaddpd256(-(__v4df)__A, (__v4df)__B, (__v4df)__C); +} + +static __inline__ __m256 __DEFAULT_FN_ATTRS256 +_mm256_fnmsub_ps(__m256 __A, __m256 __B, __m256 __C) +{ + return (__m256)__builtin_ia32_vfmaddps256(-(__v8sf)__A, (__v8sf)__B, -(__v8sf)__C); +} + +static __inline__ __m256d __DEFAULT_FN_ATTRS256 +_mm256_fnmsub_pd(__m256d __A, __m256d __B, __m256d __C) +{ + return (__m256d)__builtin_ia32_vfmaddpd256(-(__v4df)__A, (__v4df)__B, -(__v4df)__C); +} + +static __inline__ __m256 __DEFAULT_FN_ATTRS256 +_mm256_fmaddsub_ps(__m256 __A, __m256 __B, __m256 __C) +{ + return (__m256)__builtin_ia32_vfmaddsubps256((__v8sf)__A, (__v8sf)__B, (__v8sf)__C); +} + +static __inline__ __m256d __DEFAULT_FN_ATTRS256 +_mm256_fmaddsub_pd(__m256d __A, __m256d __B, __m256d __C) +{ + return (__m256d)__builtin_ia32_vfmaddsubpd256((__v4df)__A, (__v4df)__B, (__v4df)__C); +} + +static __inline__ __m256 __DEFAULT_FN_ATTRS256 +_mm256_fmsubadd_ps(__m256 __A, __m256 __B, __m256 __C) +{ + return (__m256)__builtin_ia32_vfmaddsubps256((__v8sf)__A, (__v8sf)__B, -(__v8sf)__C); +} + +static __inline__ __m256d __DEFAULT_FN_ATTRS256 +_mm256_fmsubadd_pd(__m256d __A, __m256d __B, __m256d __C) +{ + return (__m256d)__builtin_ia32_vfmaddsubpd256((__v4df)__A, (__v4df)__B, -(__v4df)__C); +} + +#undef __DEFAULT_FN_ATTRS128 +#undef __DEFAULT_FN_ATTRS256 + +#endif /* __FMAINTRIN_H */ |
