core/portable-simd/crates/core_simd/src/simd/num/
int.rs

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
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
use super::sealed::Sealed;
use crate::simd::{
    cmp::SimdPartialOrd, num::SimdUint, LaneCount, Mask, Simd, SimdCast, SimdElement,
    SupportedLaneCount,
};

/// Operations on SIMD vectors of signed integers.
pub trait SimdInt: Copy + Sealed {
    /// Mask type used for manipulating this SIMD vector type.
    type Mask;

    /// Scalar type contained by this SIMD vector type.
    type Scalar;

    /// A SIMD vector of unsigned integers with the same element size.
    type Unsigned;

    /// A SIMD vector with a different element type.
    type Cast<T: SimdElement>;

    /// Performs elementwise conversion of this vector's elements to another SIMD-valid type.
    ///
    /// This follows the semantics of Rust's `as` conversion for casting integers (wrapping to
    /// other integer types, and saturating to float types).
    #[must_use]
    fn cast<T: SimdCast>(self) -> Self::Cast<T>;

    /// Lanewise saturating add.
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::prelude::*;
    /// 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 saturating subtract.
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::prelude::*;
    /// 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 absolute value, implemented in Rust.
    /// Every element becomes its absolute value.
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::prelude::*;
    /// 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 saturating absolute value, implemented in Rust.
    /// As abs(), except the MIN value becomes MAX instead of itself.
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::prelude::*;
    /// 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 saturating negation, implemented in Rust.
    /// As neg(), except the MIN value becomes MAX instead of itself.
    ///
    /// # Examples
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::prelude::*;
    /// 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;

    /// Returns true for each positive element and false if it is zero or negative.
    fn is_positive(self) -> Self::Mask;

    /// Returns true for each negative element and false if it is zero or positive.
    fn is_negative(self) -> Self::Mask;

    /// Returns numbers representing the sign of each element.
    /// * `0` if the number is zero
    /// * `1` if the number is positive
    /// * `-1` if the number is negative
    fn signum(self) -> Self;

    /// Returns the sum of the elements of the vector, with wrapping addition.
    ///
    /// # Examples
    ///
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::prelude::*;
    /// let v = i32x4::from_array([1, 2, 3, 4]);
    /// assert_eq!(v.reduce_sum(), 10);
    ///
    /// // SIMD integer addition is always wrapping
    /// let v = i32x4::from_array([i32::MAX, 1, 0, 0]);
    /// assert_eq!(v.reduce_sum(), i32::MIN);
    /// ```
    fn reduce_sum(self) -> Self::Scalar;

    /// Returns the product of the elements of the vector, with wrapping multiplication.
    ///
    /// # Examples
    ///
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::prelude::*;
    /// let v = i32x4::from_array([1, 2, 3, 4]);
    /// assert_eq!(v.reduce_product(), 24);
    ///
    /// // SIMD integer multiplication is always wrapping
    /// let v = i32x4::from_array([i32::MAX, 2, 1, 1]);
    /// assert!(v.reduce_product() < i32::MAX);
    /// ```
    fn reduce_product(self) -> Self::Scalar;

    /// Returns the maximum element in the vector.
    ///
    /// # Examples
    ///
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::prelude::*;
    /// let v = i32x4::from_array([1, 2, 3, 4]);
    /// assert_eq!(v.reduce_max(), 4);
    /// ```
    fn reduce_max(self) -> Self::Scalar;

    /// Returns the minimum element in the vector.
    ///
    /// # Examples
    ///
    /// ```
    /// # #![feature(portable_simd)]
    /// # #[cfg(feature = "as_crate")] use core_simd::simd;
    /// # #[cfg(not(feature = "as_crate"))] use core::simd;
    /// # use simd::prelude::*;
    /// let v = i32x4::from_array([1, 2, 3, 4]);
    /// assert_eq!(v.reduce_min(), 1);
    /// ```
    fn reduce_min(self) -> Self::Scalar;

    /// Returns the cumulative bitwise "and" across the elements of the vector.
    fn reduce_and(self) -> Self::Scalar;

    /// Returns the cumulative bitwise "or" across the elements of the vector.
    fn reduce_or(self) -> Self::Scalar;

    /// Returns the cumulative bitwise "xor" across the elements of the vector.
    fn reduce_xor(self) -> Self::Scalar;

    /// Reverses the byte order of each element.
    fn swap_bytes(self) -> Self;

    /// Reverses the order of bits in each elemnent.
    /// The least significant bit becomes the most significant bit, second least-significant bit becomes second most-significant bit, etc.
    fn reverse_bits(self) -> Self;

    /// Returns the number of leading zeros in the binary representation of each element.
    fn leading_zeros(self) -> Self::Unsigned;

    /// Returns the number of trailing zeros in the binary representation of each element.
    fn trailing_zeros(self) -> Self::Unsigned;

    /// Returns the number of leading ones in the binary representation of each element.
    fn leading_ones(self) -> Self::Unsigned;

    /// Returns the number of trailing ones in the binary representation of each element.
    fn trailing_ones(self) -> Self::Unsigned;
}

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

        impl<const N: usize> SimdInt for Simd<$ty, N>
        where
            LaneCount<N>: SupportedLaneCount,
        {
            type Mask = Mask<<$ty as SimdElement>::Mask, N>;
            type Scalar = $ty;
            type Unsigned = Simd<$unsigned, N>;
            type Cast<T: SimdElement> = Simd<T, N>;

            #[inline]
            fn cast<T: SimdCast>(self) -> Self::Cast<T> {
                // Safety: supported types are guaranteed by SimdCast
                unsafe { core::intrinsics::simd::simd_as(self) }
            }

            #[inline]
            fn saturating_add(self, second: Self) -> Self {
                // Safety: `self` is a vector
                unsafe { core::intrinsics::simd::simd_saturating_add(self, second) }
            }

            #[inline]
            fn saturating_sub(self, second: Self) -> Self {
                // Safety: `self` is a vector
                unsafe { core::intrinsics::simd::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 {
                // arith shift for -1 or 0 mask based on sign bit, giving 2s complement
                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 {
                // Safety: `self` is an integer vector
                unsafe { core::intrinsics::simd::simd_reduce_add_ordered(self, 0) }
            }

            #[inline]
            fn reduce_product(self) -> Self::Scalar {
                // Safety: `self` is an integer vector
                unsafe { core::intrinsics::simd::simd_reduce_mul_ordered(self, 1) }
            }

            #[inline]
            fn reduce_max(self) -> Self::Scalar {
                // Safety: `self` is an integer vector
                unsafe { core::intrinsics::simd::simd_reduce_max(self) }
            }

            #[inline]
            fn reduce_min(self) -> Self::Scalar {
                // Safety: `self` is an integer vector
                unsafe { core::intrinsics::simd::simd_reduce_min(self) }
            }

            #[inline]
            fn reduce_and(self) -> Self::Scalar {
                // Safety: `self` is an integer vector
                unsafe { core::intrinsics::simd::simd_reduce_and(self) }
            }

            #[inline]
            fn reduce_or(self) -> Self::Scalar {
                // Safety: `self` is an integer vector
                unsafe { core::intrinsics::simd::simd_reduce_or(self) }
            }

            #[inline]
            fn reduce_xor(self) -> Self::Scalar {
                // Safety: `self` is an integer vector
                unsafe { core::intrinsics::simd::simd_reduce_xor(self) }
            }

            #[inline]
            fn swap_bytes(self) -> Self {
                // Safety: `self` is an integer vector
                unsafe { core::intrinsics::simd::simd_bswap(self) }
            }

            #[inline]
            fn reverse_bits(self) -> Self {
                // Safety: `self` is an integer vector
                unsafe { core::intrinsics::simd::simd_bitreverse(self) }
            }

            #[inline]
            fn leading_zeros(self) -> Self::Unsigned {
                self.cast::<$unsigned>().leading_zeros()
            }

            #[inline]
            fn trailing_zeros(self) -> Self::Unsigned {
                self.cast::<$unsigned>().trailing_zeros()
            }

            #[inline]
            fn leading_ones(self) -> Self::Unsigned {
                self.cast::<$unsigned>().leading_ones()
            }

            #[inline]
            fn trailing_ones(self) -> Self::Unsigned {
                self.cast::<$unsigned>().trailing_ones()
            }
        }
        )*
    }
}

impl_trait! { i8 (u8), i16 (u16), i32 (u32), i64 (u64), isize (usize) }