thermite 0.2.0

High-performance, generic, ISA-portable SIMD library with a policy-configurable transcendental math library
Documentation 
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

use super::arch;

use crate::register::{
    CastRegister, ConcatRegister, IndexableRegister, NumericRegister, Register, SaturatingCastRegister, Storage,
    reduced::{HalfRegister2, ReducedRegister},
};

pub type F32x2V1 = HalfRegister2<super::F32x4V1>;
pub type I32x2V1 = HalfRegister2<super::I32x4V1>;
pub type U32x2V1 = HalfRegister2<super::U32x4V1>;

#[thermite_macros::inline_always]
impl ConcatRegister<f32> for F32x2V1 {
    fn concat(lo: Storage<f32>, hi: Storage<f32>) -> Storage<Self> {
        ReducedRegister::new(unsafe { arch::_mm_setr_ps(lo, hi, 0.0, 0.0) })
    }

    fn split(value: Storage<Self>) -> (Storage<f32>, Storage<f32>) {
        let mut arr = [0.0f32; 4];
        unsafe { arch::_mm_storeu_ps(arr.as_mut_ptr(), value.0) };
        (arr[0], arr[1])
    }
}

// `ExtendRegister<f32> for F32x2V1` comes from the generic extend-from-scalar blanket
// in `register/reduced.rs` (F32x4V1::single / ::extract::<0>, same codegen).

#[thermite_macros::inline_always]
impl ConcatRegister<F32x2V1> for super::F32x4V1 {
    fn concat(lo: Storage<F32x2V1>, hi: Storage<F32x2V1>) -> Storage<Self> {
        unsafe { arch::_mm_movelh_ps(lo.0, hi.0) }
    }

    fn split(value: Storage<Self>) -> (Storage<F32x2V1>, Storage<F32x2V1>) {
        (
            ReducedRegister::new(value),
            ReducedRegister::new(unsafe { arch::_mm_movehl_ps(value, value) }),
        )
    }
}

#[thermite_macros::inline_always]
impl ConcatRegister<i32> for I32x2V1 {
    fn concat(lo: Storage<i32>, hi: Storage<i32>) -> Storage<Self> {
        ReducedRegister::new(unsafe { arch::_mm_setr_epi32(lo, hi, 0, 0) })
    }

    fn split(value: Storage<Self>) -> (Storage<i32>, Storage<i32>) {
        let mut arr = [0i32; 4];
        unsafe { arch::_mm_storeu_si128(arr.as_mut_ptr() as *mut _, value.0) };
        (arr[0], arr[1])
    }
}

// See the F32x2V1 note above: supplied by the extend-from-scalar blanket.

#[thermite_macros::inline_always]
impl ConcatRegister<I32x2V1> for super::I32x4V1 {
    fn concat(lo: Storage<I32x2V1>, hi: Storage<I32x2V1>) -> Storage<Self> {
        unsafe { arch::_mm_unpacklo_epi64(lo.0, hi.0) }
    }

    fn split(value: Storage<Self>) -> (Storage<I32x2V1>, Storage<I32x2V1>) {
        (
            ReducedRegister::new(value),
            ReducedRegister::new(unsafe { arch::_mm_unpackhi_epi64(value, value) }),
        )
    }
}

#[thermite_macros::inline_always]
impl ConcatRegister<u32> for U32x2V1 {
    fn concat(lo: Storage<u32>, hi: Storage<u32>) -> Storage<Self> {
        ReducedRegister::new(unsafe { arch::_mm_setr_epi32(lo as i32, hi as i32, 0, 0) })
    }

    fn split(value: Storage<Self>) -> (Storage<u32>, Storage<u32>) {
        let mut arr = [0i32; 4];
        unsafe { arch::_mm_storeu_si128(arr.as_mut_ptr() as *mut _, value.0) };
        (arr[0] as u32, arr[1] as u32)
    }
}

// See the F32x2V1 note above: supplied by the extend-from-scalar blanket.

#[thermite_macros::inline_always]
impl ConcatRegister<U32x2V1> for super::U32x4V1 {
    fn concat(lo: Storage<U32x2V1>, hi: Storage<U32x2V1>) -> Storage<Self> {
        unsafe { arch::_mm_unpacklo_epi64(lo.0, hi.0) }
    }

    fn split(value: Storage<Self>) -> (Storage<U32x2V1>, Storage<U32x2V1>) {
        (
            ReducedRegister::new(value),
            ReducedRegister::new(unsafe { arch::_mm_unpackhi_epi64(value, value) }),
        )
    }
}

#[thermite_macros::inline_always]
impl CastRegister<F32x2V1> for super::F64x2V1 {
    fn cast_from(value: Storage<F32x2V1>) -> Storage<Self> {
        unsafe { arch::_mm_cvtps_pd(value.0) }
    }
}

#[thermite_macros::inline_always]
impl CastRegister<super::F64x2V1> for F32x2V1 {
    fn cast_from(value: Storage<super::F64x2V1>) -> Storage<Self> {
        ReducedRegister::new(unsafe { arch::_mm_cvtpd_ps(value) })
    }
}

#[thermite_macros::inline_always]
impl CastRegister<super::F64x2V1> for I32x2V1 {
    fn cast_from(value: Storage<super::F64x2V1>) -> Storage<Self> {
        ReducedRegister::new(unsafe { arch::_mm_cvttpd_epi32(value) })
    }
}

#[thermite_macros::inline_always]
impl CastRegister<super::F64x2V1> for U32x2V1 {
    fn cast_from(value: Storage<super::F64x2V1>) -> Storage<Self> {
        ReducedRegister::new(unsafe { arch::_mm_cvtpd_epu32x_v1(value) })
    }
}

#[thermite_macros::inline_always]
impl CastRegister<I32x2V1> for super::F64x2V1 {
    fn cast_from(value: Storage<I32x2V1>) -> Storage<Self> {
        unsafe { arch::_mm_cvtepi32_pd(value.0) }
    }
}

#[thermite_macros::inline_always]
impl CastRegister<U32x2V1> for super::U64x2V1 {
    fn cast_from(value: Storage<U32x2V1>) -> Storage<Self> {
        unsafe { arch::_mm_cvtepu32_epi64x_v1(value.0) }
    }
}

#[thermite_macros::inline_always]
impl CastRegister<super::U64x2V1> for U32x2V1 {
    fn cast_from(value: Storage<super::U64x2V1>) -> Storage<Self> {
        ReducedRegister::new(unsafe { arch::_mm_shuffle_epi32(value, 0b10_00_10_00) })
    }
}

#[thermite_macros::inline_always]
impl CastRegister<I32x2V1> for super::I64x2V1 {
    fn cast_from(value: Storage<I32x2V1>) -> Storage<Self> {
        unsafe { arch::_mm_cvtepi32_epi64x_v1(value.0) }
    }
}

#[thermite_macros::inline_always]
impl CastRegister<super::I64x2V1> for I32x2V1 {
    fn cast_from(value: Storage<super::I64x2V1>) -> Storage<Self> {
        ReducedRegister::new(unsafe { arch::_mm_shuffle_epi32(value, 0b10_00_10_00) })
    }
}

// Saturating narrow i64 -> i32: no SSE 64-bit pack, so clamp (polyfilled 64-bit min/max) + narrow.
macro_rules! sat_clamp_narrow {
    ($(($from:ty, $fe:ty, $into:ty, $ie:ty)),* $(,)?) => {$(
        #[thermite_macros::inline_always]
        impl SaturatingCastRegister<$from> for $into {
            fn saturating_cast_from(value: Storage<$from>) -> Storage<Self> {
                let lo = <$from as Register>::splat(<$ie>::MIN as $fe);
                let hi = <$from as Register>::splat(<$ie>::MAX as $fe);
                let clamped = <$from as NumericRegister>::min(<$from as NumericRegister>::max(value, lo), hi);
                <Self as CastRegister<$from>>::cast_from(clamped)
            }
        }
    )*};
}
sat_clamp_narrow! {
    (super::I64x2V1, i64, I32x2V1, i32),
    (super::U64x2V1, u64, U32x2V1, u32),
}

#[thermite_macros::inline_always]
impl IndexableRegister<super::U64x2V1> for F32x2V1 {}
impl IndexableRegister<super::U64x2V1> for U32x2V1 {}
impl IndexableRegister<super::U64x2V1> for I32x2V1 {}