1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175
//! This module defines some SIMD constants and provides a missing floor() implementaiton for
//! packed_simd
use cfg_if::cfg_if;
/// All the intermediate calculations are performed on this type.
#[allow(dead_code)]
pub trait SimdCompat: Copy + Sized {
/// Number of SIMD lanes, if no SIMD types are used this equals to 1.
const LANES: usize;
/// A function to wrap float value primitives for SIMD enabled calculations
fn sc_splat(v: f32) -> Self;
/// Return the element-wise minimum with other
fn sc_min(self, other: Self) -> Self;
/// Return the element-wise maximum with other
fn sc_max(self, other: Self) -> Self;
}
cfg_if! {if #[cfg(all(any(target_arch = "x86", target_arch = "x86_64"), feature = "use-simd"))] {
pub use std::simd::{StdFloat, cmp::SimdPartialOrd, prelude::{SimdFloat, mask32x8, u32x8, f32x8}};
// #[cfg(all(target_arch = "x86", target_feature = "sse4.1"))]
// use std::arch::x86::*;
// #[cfg(all(target_arch = "x86_64", target_feature = "sse4.1"))]
// use std::arch::x86_64::*;
#[allow(non_camel_case_types)]
pub type m32s = mask32x8;
#[allow(non_camel_case_types)]
pub type u32s = u32x8;
#[allow(non_camel_case_types)]
pub type f32s = f32x8;
#[allow(non_camel_case_types)]
pub type f32tuple = [f32;f32s::LEN];
pub type Flt = f32s;
pub const fn csplat(v: f32) -> Flt { Flt::from_array([v;Flt::LANES]) }
impl SimdCompat for f32s {
const LANES: usize = f32s::LEN;
#[inline(always)]
fn sc_splat(v: f32) -> Self { Self::splat(v) }
#[inline(always)]
fn sc_min(self, other: Self) -> Self { self.simd_min(other) }
#[inline(always)]
fn sc_max(self, other: Self) -> Self { self.simd_max(other) }
}
macro_rules! simd_new_consecutive {
// ($name:ident, $v:expr) => ($name::new($v, $v+1, $v+2, $v+3, $v+4, $v+5, $v+6, $v+7,
// $v+8, $v+9, $v+10, $v+11, $v+12, $v+13, $v+14 ,$v+15));
($name:ident, $v:expr) => ($name::from_array([$v, $v+1, $v+2, $v+3, $v+4, $v+5, $v+6, $v+7]));
// ($name:ident, $v:expr) => ($name::new($v, $v+1, $v+2, $v+3));
}
pub(crate) use simd_new_consecutive;
// /* floor SIMD */
// cfg_if! {if #[cfg(target_feature = "avx")] {
// use std::mem::transmute;
// #[inline]
// pub fn floor(val: f32x8) -> f32x8 {
// unsafe {
// transmute(_mm256_floor_ps(transmute(val)))
// }
// }
// }
// else if #[cfg(target_feature = "sse4.1")] {
// use std::mem::transmute;
// #[inline]
// pub fn floor(val: f32x8) -> f32x8 {
// unsafe {
// union U {
// vec: f32x8,
// halves: [__m128; 2]
// }
// let mut halves = U { vec: val }.halves;
// *halves.get_unchecked_mut(0) =
// transmute(_mm_floor_ps(transmute(*halves.get_unchecked(0))));
// *halves.get_unchecked_mut(1) =
// transmute(_mm_floor_ps(transmute(*halves.get_unchecked(1))));
// U { halves }.vec
// }
// }
// }
// else {
// #[inline]
// pub fn floor(val: f32x8) -> f32x8 {
// unsafe {
// union U {
// vec: f32x8,
// scalars: [f32; 8]
// }
// let mut scalars = U { vec: val }.scalars;
// for i in &mut scalars {
// *i = (*i).floor();
// }
// U { scalars }.vec
// }
// }
// }}
}
else if #[cfg(all(target_arch = "aarch64", feature = "use-simd"))] {
pub use std::simd::{StdFloat, cmp::SimdPartialOrd, prelude::{SimdFloat, mask32x4, u32x4, f32x4}};
// use core::ops::Sub;
#[allow(non_camel_case_types)]
pub type m32s = mask32x4;
#[allow(non_camel_case_types)]
pub type u32s = u32x4;
#[allow(non_camel_case_types)]
pub type f32s = f32x4;
#[allow(non_camel_case_types)]
pub type f32tuple = [f32;f32s::LEN];
pub type Flt = f32s;
pub const fn csplat(v: f32) -> Flt { Flt::from_array([v;Flt::LANES]) }
impl SimdCompat for f32s {
const LANES: usize = f32s::LEN;
#[inline(always)]
fn sc_splat(v: f32) -> Self { Self::splat(v) }
#[inline(always)]
fn sc_min(self, other: Self) -> Self { self.simd_min(other) }
#[inline(always)]
fn sc_max(self, other: Self) -> Self { self.simd_max(other) }
}
macro_rules! simd_new_consecutive {
($name:ident, $v:expr) => ($name::from_array([$v, $v+1, $v+2, $v+3]));
}
pub(crate) use simd_new_consecutive;
// /* floor SIMD */
// cfg_if! {if #[cfg(target_feature = "neon")] {
// use std::mem::transmute;
// #[inline]
// pub fn floor(val: f32x4) -> f32x4 {
// unsafe {
// transmute(core::arch::aarch64::vrndnq_f32(transmute(val.sub(splat(0.5)))))
// }
// }
// }
// else {
// #[inline]
// pub fn floor(val: f32x4) -> f32x4 {
// unsafe {
// union U {
// vec: f32x4,
// scalars: [f32; 4]
// }
// let mut scalars = U { vec: val }.scalars;
// for i in &mut scalars {
// *i = (*i).floor();
// }
// U { scalars }.vec
// }
// }
// }}
}
else {
/// All the intermediate calculations are performed on this type.
pub type Flt = f32;
pub const fn csplat(v: f32) -> Flt { v }
impl SimdCompat for f32 {
const LANES: usize = 1;
#[inline(always)]
fn sc_splat(v: f32) -> Self { v }
#[inline(always)]
fn sc_min(self, other: Self) -> Self { self.min(other) }
#[inline(always)]
fn sc_max(self, other: Self) -> Self { self.max(other) }
}
}}