1//===-- x86_64 implementations of the fma function --------------*- C++ -*-===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8
9#ifndef LLVM_LIBC_SRC___SUPPORT_FPUTIL_X86_64_FMA_H
10#define LLVM_LIBC_SRC___SUPPORT_FPUTIL_X86_64_FMA_H
11
12#include "src/__support/macros/attributes.h" // LIBC_INLINE
13#include "src/__support/macros/properties/architectures.h"
14#include "src/__support/macros/properties/cpu_features.h" // LIBC_TARGET_CPU_HAS_FMA
15
16#if !defined(LIBC_TARGET_ARCH_IS_X86_64)
17#error "Invalid include"
18#endif
19
20#if !defined(LIBC_TARGET_CPU_HAS_FMA)
21#error "FMA instructions are not supported"
22#endif
23
24#include "src/__support/CPP/type_traits.h"
25#include <immintrin.h>
26
27namespace LIBC_NAMESPACE {
28namespace fputil {
29
30template <typename T>
31LIBC_INLINE cpp::enable_if_t<cpp::is_same_v<T, float>, T> fma(T x, T y, T z) {
32 float result;
33 __m128 xmm = _mm_load_ss(&x); // NOLINT
34 __m128 ymm = _mm_load_ss(&y); // NOLINT
35 __m128 zmm = _mm_load_ss(&z); // NOLINT
36 __m128 r = _mm_fmadd_ss(A: xmm, B: ymm, C: zmm); // NOLINT
37 _mm_store_ss(p: &result, a: r); // NOLINT
38 return result;
39}
40
41template <typename T>
42LIBC_INLINE cpp::enable_if_t<cpp::is_same_v<T, double>, T> fma(T x, T y, T z) {
43 double result;
44 __m128d xmm = _mm_load_sd(&x); // NOLINT
45 __m128d ymm = _mm_load_sd(&y); // NOLINT
46 __m128d zmm = _mm_load_sd(&z); // NOLINT
47 __m128d r = _mm_fmadd_sd(A: xmm, B: ymm, C: zmm); // NOLINT
48 _mm_store_sd(dp: &result, a: r); // NOLINT
49 return result;
50}
51
52} // namespace fputil
53} // namespace LIBC_NAMESPACE
54
55#endif // LLVM_LIBC_SRC___SUPPORT_FPUTIL_X86_64_FMA_H
56

source code of libc/src/__support/FPUtil/x86_64/FMA.h