- simd[meta header]
- std::simd[meta namespace]
- function template[meta id-type]
- cpp26[meta cpp]
namespace std::simd {
template<math-floating-point V>
constexpr deduced-vec-t<V> fmod(const V& x, const V& y); // (1) C++26
template<math-floating-point V>
constexpr deduced-vec-t<V> fmod(const deduced-vec-t<V>& x, const V& y); // (2) C++26
template<math-floating-point V>
constexpr deduced-vec-t<V> fmod(const V& x, const deduced-vec-t<V>& y); // (3) C++26
}- deduced-vec-t[link /reference/simd/deduced-vec-t.md]
- math-floating-point[link /reference/simd/math-floating-point.md]
basic_vecの各要素について、xをyで割った剰余(x - n * y、nはx / yを0方向へ丸めた整数)を求める。
制約math-floating-pointは、Vが浮動小数点要素のbasic_vec、またはそこからbasic_vecを導出可能なスカラー浮動小数点型であることを表す説明専用のコンセプトである。deduced-vec-t<V>は、Vに対応するbasic_vec型(Vがスカラー型のときはそれを要素型とするbasic_vec)を表す。
- (1) : 両引数が同じ型
Vの場合。 - (2), (3) : 一方の引数を
basic_vec、他方をスカラー値として、スカラー側を全要素に対して適用する場合。
各要素iについて、x・yの対応する要素に<cmath>のfmodを適用した結果で初期化されたdeduced-vec-t<V>型のオブジェクトを返す。
ある要素iについて定義域エラー・極エラー・値域エラーが発生する場合、その要素の値は未規定である。
errnoにアクセスするか否かは未規定である。
#include <simd>
#include <print>
namespace simd = std::simd;
int main()
{
simd::vec<float, 4> x{[](int i) { return i + 5.0f; }}; // {5, 6, 7, 8}
// 各要素を3.0で割った剰余(スカラーとの演算)
simd::vec<float, 4> r = simd::fmod(x, 3.0f);
for (int i = 0; i < r.size(); ++i)
std::print("{} ", r[i]);
std::println("");
}- simd::fmod[color ff0000]
- simd::vec[link basic_vec.md]
2 0 1 2
- C++26
- Clang: 22 [mark noimpl]
- GCC: 16.1 [mark noimpl]
- Visual C++: 2026 Update 2 [mark noimpl]
- P1928R15 std::simd — merge data-parallel types from the Parallelism TS 2
- C++26で追加された
- P3844R4 Reword simd.math for consteval conversions
<cmath>関数と同様に引数の変換を呼び出し側で行うよう、<simd>の数学関数が複数のオーバーロードとして再規定された(比較関数の戻り値型の修正・不足していたオーバーロードの追加を含む)