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
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
use super::sealed::Sealed;
use crate::simd::{
    intrinsics, LaneCount, Mask, Simd, SimdElement, SimdPartialOrd, SupportedLaneCount,
};

/// 对有符号整数的 SIMD vectors 的操作。
pub trait SimdInt: Copy + Sealed {
    /// 用于操作此 SIMD vector 类型的掩码类型。
    type Mask;

    /// 此 SIMD vector 类型包含的标量类型。
    type Scalar;

    /// Lanewise 饱和加法。
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::{Simd, SimdInt};
    /// use core::i32::{MIN, MAX};
    /// let x = Simd::from_array([MIN, 0, 1, MAX]);
    /// let max = Simd::splat(MAX);
    /// let unsat = x + max;
    /// let sat = x.saturating_add(max);
    /// assert_eq!(unsat, Simd::from_array([-1, MAX, MIN, -2]));
    /// assert_eq!(sat, Simd::from_array([-1, MAX, MAX, MAX]));
    /// ```
    fn saturating_add(self, second: Self) -> Self;

    /// Lanewise 饱和减法。
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::{Simd, SimdInt};
    /// use core::i32::{MIN, MAX};
    /// let x = Simd::from_array([MIN, -2, -1, MAX]);
    /// let max = Simd::splat(MAX);
    /// let unsat = x - max;
    /// let sat = x.saturating_sub(max);
    /// assert_eq!(unsat, Simd::from_array([1, MAX, MIN, 0]));
    /// assert_eq!(sat, Simd::from_array([MIN, MIN, MIN, 0]));
    fn saturating_sub(self, second: Self) -> Self;

    /// Lanewise 绝对值,在 Rust 中实现。
    /// 每个 lane 都成为它的绝对值。
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::{Simd, SimdInt};
    /// use core::i32::{MIN, MAX};
    /// let xs = Simd::from_array([MIN, MIN +1, -5, 0]);
    /// assert_eq!(xs.abs(), Simd::from_array([MIN, MAX, 5, 0]));
    /// ```
    fn abs(self) -> Self;

    /// Lanewise 饱和绝对值,在 Rust 中实现。
    /// 作为 abs(),除了 MIN 值变为 MAX 而不是它本身。
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::{Simd, SimdInt};
    /// use core::i32::{MIN, MAX};
    /// let xs = Simd::from_array([MIN, -2, 0, 3]);
    /// let unsat = xs.abs();
    /// let sat = xs.saturating_abs();
    /// assert_eq!(unsat, Simd::from_array([MIN, 2, 0, 3]));
    /// assert_eq!(sat, Simd::from_array([MAX, 2, 0, 3]));
    /// ```
    fn saturating_abs(self) -> Self;

    /// Lanewise 饱和否定,在 Rust 中实现。
    /// 作为 neg(),除了 MIN 值变为 MAX 而不是它本身。
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::{Simd, SimdInt};
    /// use core::i32::{MIN, MAX};
    /// let x = Simd::from_array([MIN, -2, 3, MAX]);
    /// let unsat = -x;
    /// let sat = x.saturating_neg();
    /// assert_eq!(unsat, Simd::from_array([MIN, 2, -3, MIN + 1]));
    /// assert_eq!(sat, Simd::from_array([MAX, 2, -3, MIN + 1]));
    /// ```
    fn saturating_neg(self) -> Self;

    /// 对于每个正 lane 返回真,如果为零或负则返回假。
    fn is_positive(self) -> Self::Mask;

    /// 对于每个负 lane 返回 true,如果为零或正则返回 false。
    fn is_negative(self) -> Self::Mask;

    /// 返回代表每个 lane 符号的数字。
    /// * `0` 如果数字为零
    /// * `1` 如果数字是正数
    /// * `-1` 如果数字是负数
    fn signum(self) -> Self;

    /// 返回 vector 的 lane 总和,带包装加法。
    ///
    /// # Examples
    ///
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::{i32x4, SimdInt};
    /// let v = i32x4::from_array([1, 2, 3, 4]);
    /// assert_eq!(v.reduce_sum(), 10);
    ///
    /// // SIMD 整数加法总是换行
    /// let v = i32x4::from_array([i32::MAX, 1, 0, 0]);
    /// assert_eq!(v.reduce_sum(), i32::MIN);
    /// ```
    fn reduce_sum(self) -> Self::Scalar;

    /// 返回 vector 的 lane 的乘积,带包装乘法。
    ///
    /// # Examples
    ///
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::{i32x4, SimdInt};
    /// let v = i32x4::from_array([1, 2, 3, 4]);
    /// assert_eq!(v.reduce_product(), 24);
    ///
    /// // SIMD 整数乘法总是换行
    /// let v = i32x4::from_array([i32::MAX, 2, 1, 1]);
    /// assert!(v.reduce_product() < i32::MAX);
    /// ```
    fn reduce_product(self) -> Self::Scalar;

    /// 返回 vector 中的最大 lane。
    ///
    /// # Examples
    ///
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::{i32x4, SimdInt};
    /// let v = i32x4::from_array([1, 2, 3, 4]);
    /// assert_eq!(v.reduce_max(), 4);
    /// ```
    fn reduce_max(self) -> Self::Scalar;

    /// 返回 vector 中的最小 lane。
    ///
    /// # Examples
    ///
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::{i32x4, SimdInt};
    /// let v = i32x4::from_array([1, 2, 3, 4]);
    /// assert_eq!(v.reduce_min(), 1);
    /// ```
    fn reduce_min(self) -> Self::Scalar;

    /// 返回跨 vector lane 的累积按位与。
    fn reduce_and(self) -> Self::Scalar;

    /// 返回跨 vector lane 的累积按位或。
    fn reduce_or(self) -> Self::Scalar;

    /// 返回跨 vector lane 的累积按位异或。
    fn reduce_xor(self) -> Self::Scalar;
}

macro_rules! impl_trait {
    { $($ty:ty),* } => {
        $(
        impl<const LANES: usize> Sealed for Simd<$ty, LANES>
        where
            LaneCount<LANES>: SupportedLaneCount,
        {
        }

        impl<const LANES: usize> SimdInt for Simd<$ty, LANES>
        where
            LaneCount<LANES>: SupportedLaneCount,
        {
            type Mask = Mask<<$ty as SimdElement>::Mask, LANES>;
            type Scalar = $ty;

            #[inline]
            fn saturating_add(self, second: Self) -> Self {
                // 安全性: `self` 是 vector
                unsafe { intrinsics::simd_saturating_add(self, second) }
            }

            #[inline]
            fn saturating_sub(self, second: Self) -> Self {
                // 安全性: `self` 是 vector
                unsafe { intrinsics::simd_saturating_sub(self, second) }
            }

            #[inline]
            fn abs(self) -> Self {
                const SHR: $ty = <$ty>::BITS as $ty - 1;
                let m = self >> Simd::splat(SHR);
                (self^m) - m
            }

            #[inline]
            fn saturating_abs(self) -> Self {
                // 基于符号位的 -1 或 0 掩码的算术移位,给出 2s 补码
                const SHR: $ty = <$ty>::BITS as $ty - 1;
                let m = self >> Simd::splat(SHR);
                (self^m).saturating_sub(m)
            }

            #[inline]
            fn saturating_neg(self) -> Self {
                Self::splat(0).saturating_sub(self)
            }

            #[inline]
            fn is_positive(self) -> Self::Mask {
                self.simd_gt(Self::splat(0))
            }

            #[inline]
            fn is_negative(self) -> Self::Mask {
                self.simd_lt(Self::splat(0))
            }

            #[inline]
            fn signum(self) -> Self {
                self.is_positive().select(
                    Self::splat(1),
                    self.is_negative().select(Self::splat(-1), Self::splat(0))
                )
            }

            #[inline]
            fn reduce_sum(self) -> Self::Scalar {
                // 安全性: `self` 是整数 vector
                unsafe { intrinsics::simd_reduce_add_ordered(self, 0) }
            }

            #[inline]
            fn reduce_product(self) -> Self::Scalar {
                // 安全性: `self` 是整数 vector
                unsafe { intrinsics::simd_reduce_mul_ordered(self, 1) }
            }

            #[inline]
            fn reduce_max(self) -> Self::Scalar {
                // 安全性: `self` 是整数 vector
                unsafe { intrinsics::simd_reduce_max(self) }
            }

            #[inline]
            fn reduce_min(self) -> Self::Scalar {
                // 安全性: `self` 是整数 vector
                unsafe { intrinsics::simd_reduce_min(self) }
            }

            #[inline]
            fn reduce_and(self) -> Self::Scalar {
                // 安全性: `self` 是整数 vector
                unsafe { intrinsics::simd_reduce_and(self) }
            }

            #[inline]
            fn reduce_or(self) -> Self::Scalar {
                // 安全性: `self` 是整数 vector
                unsafe { intrinsics::simd_reduce_or(self) }
            }

            #[inline]
            fn reduce_xor(self) -> Self::Scalar {
                // 安全性: `self` 是整数 vector
                unsafe { intrinsics::simd_reduce_xor(self) }
            }
        }
        )*
    }
}

impl_trait! { i8, i16, i32, i64, isize }