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
|
//! LLVM's platform intrinsics
#![allow(dead_code)]
use crate::sealed::Shuffle;
#[allow(unused_imports)] // FIXME: spurious warning?
use crate::sealed::Simd;
// Shuffle intrinsics: expanded in users' crates, therefore public.
extern "platform-intrinsic" {
pub fn simd_shuffle2<T, U>(x: T, y: T, idx: [u32; 2]) -> U;
pub fn simd_shuffle4<T, U>(x: T, y: T, idx: [u32; 4]) -> U;
pub fn simd_shuffle8<T, U>(x: T, y: T, idx: [u32; 8]) -> U;
pub fn simd_shuffle16<T, U>(x: T, y: T, idx: [u32; 16]) -> U;
pub fn simd_shuffle32<T, U>(x: T, y: T, idx: [u32; 32]) -> U;
pub fn simd_shuffle64<T, U>(x: T, y: T, idx: [u32; 64]) -> U;
}
#[allow(clippy::missing_safety_doc)]
#[inline]
pub unsafe fn __shuffle_vector2<const IDX: [u32; 2], T, U>(x: T, y: T) -> U
where
T: Simd,
<T as Simd>::Element: Shuffle<[u32; 2], Output = U>,
{
simd_shuffle2(x, y, IDX)
}
#[allow(clippy::missing_safety_doc)]
#[inline]
pub unsafe fn __shuffle_vector4<const IDX: [u32; 4], T, U>(x: T, y: T) -> U
where
T: Simd,
<T as Simd>::Element: Shuffle<[u32; 4], Output = U>,
{
simd_shuffle4(x, y, IDX)
}
#[allow(clippy::missing_safety_doc)]
#[inline]
pub unsafe fn __shuffle_vector8<const IDX: [u32; 8], T, U>(x: T, y: T) -> U
where
T: Simd,
<T as Simd>::Element: Shuffle<[u32; 8], Output = U>,
{
simd_shuffle8(x, y, IDX)
}
#[allow(clippy::missing_safety_doc)]
#[inline]
pub unsafe fn __shuffle_vector16<const IDX: [u32; 16], T, U>(x: T, y: T) -> U
where
T: Simd,
<T as Simd>::Element: Shuffle<[u32; 16], Output = U>,
{
simd_shuffle16(x, y, IDX)
}
#[allow(clippy::missing_safety_doc)]
#[inline]
pub unsafe fn __shuffle_vector32<const IDX: [u32; 32], T, U>(x: T, y: T) -> U
where
T: Simd,
<T as Simd>::Element: Shuffle<[u32; 32], Output = U>,
{
simd_shuffle32(x, y, IDX)
}
#[allow(clippy::missing_safety_doc)]
#[inline]
pub unsafe fn __shuffle_vector64<const IDX: [u32; 64], T, U>(x: T, y: T) -> U
where
T: Simd,
<T as Simd>::Element: Shuffle<[u32; 64], Output = U>,
{
simd_shuffle64(x, y, IDX)
}
extern "platform-intrinsic" {
pub(crate) fn simd_eq<T, U>(x: T, y: T) -> U;
pub(crate) fn simd_ne<T, U>(x: T, y: T) -> U;
pub(crate) fn simd_lt<T, U>(x: T, y: T) -> U;
pub(crate) fn simd_le<T, U>(x: T, y: T) -> U;
pub(crate) fn simd_gt<T, U>(x: T, y: T) -> U;
pub(crate) fn simd_ge<T, U>(x: T, y: T) -> U;
pub(crate) fn simd_insert<T, U>(x: T, idx: u32, val: U) -> T;
pub(crate) fn simd_extract<T, U>(x: T, idx: u32) -> U;
pub(crate) fn simd_cast<T, U>(x: T) -> U;
pub(crate) fn simd_add<T>(x: T, y: T) -> T;
pub(crate) fn simd_sub<T>(x: T, y: T) -> T;
pub(crate) fn simd_mul<T>(x: T, y: T) -> T;
pub(crate) fn simd_div<T>(x: T, y: T) -> T;
pub(crate) fn simd_rem<T>(x: T, y: T) -> T;
pub(crate) fn simd_shl<T>(x: T, y: T) -> T;
pub(crate) fn simd_shr<T>(x: T, y: T) -> T;
pub(crate) fn simd_and<T>(x: T, y: T) -> T;
pub(crate) fn simd_or<T>(x: T, y: T) -> T;
pub(crate) fn simd_xor<T>(x: T, y: T) -> T;
pub(crate) fn simd_reduce_add_unordered<T, U>(x: T) -> U;
pub(crate) fn simd_reduce_mul_unordered<T, U>(x: T) -> U;
pub(crate) fn simd_reduce_add_ordered<T, U>(x: T, acc: U) -> U;
pub(crate) fn simd_reduce_mul_ordered<T, U>(x: T, acc: U) -> U;
pub(crate) fn simd_reduce_min<T, U>(x: T) -> U;
pub(crate) fn simd_reduce_max<T, U>(x: T) -> U;
pub(crate) fn simd_reduce_min_nanless<T, U>(x: T) -> U;
pub(crate) fn simd_reduce_max_nanless<T, U>(x: T) -> U;
pub(crate) fn simd_reduce_and<T, U>(x: T) -> U;
pub(crate) fn simd_reduce_or<T, U>(x: T) -> U;
pub(crate) fn simd_reduce_xor<T, U>(x: T) -> U;
pub(crate) fn simd_reduce_all<T>(x: T) -> bool;
pub(crate) fn simd_reduce_any<T>(x: T) -> bool;
pub(crate) fn simd_select<M, T>(m: M, a: T, b: T) -> T;
pub(crate) fn simd_fmin<T>(a: T, b: T) -> T;
pub(crate) fn simd_fmax<T>(a: T, b: T) -> T;
pub(crate) fn simd_fsqrt<T>(a: T) -> T;
pub(crate) fn simd_fma<T>(a: T, b: T, c: T) -> T;
pub(crate) fn simd_gather<T, P, M>(value: T, pointers: P, mask: M) -> T;
pub(crate) fn simd_scatter<T, P, M>(value: T, pointers: P, mask: M);
pub(crate) fn simd_bitmask<T, U>(value: T) -> U;
}
|
