blob: c1e379c34241fff11b8446557c6555429784e971 (
plain)
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
|
//! Vertical floating-point `sqrt`
#![allow(unused)]
// FIXME 64-bit 1 elem vectors sqrte
use crate::llvm::simd_fsqrt;
use crate::*;
crate trait Sqrte {
fn sqrte(self) -> Self;
}
gen_unary_impl_table!(Sqrte, sqrte);
cfg_if! {
if #[cfg(all(target_arch = "x86_64", feature = "sleef-sys"))] {
use sleef_sys::*;
cfg_if! {
if #[cfg(target_feature = "avx2")] {
impl_unary!(f32x2[t => f32x4]: Sleef_sqrtf4_u35avx2128);
impl_unary!(f32x16[h => f32x8]: Sleef_sqrtf8_u35avx2);
impl_unary!(f64x8[h => f64x4]: Sleef_sqrtd4_u35avx2);
impl_unary!(f32x4: Sleef_sqrtf4_u35avx2128);
impl_unary!(f32x8: Sleef_sqrtf8_u35avx2);
impl_unary!(f64x2: Sleef_sqrtd2_u35avx2128);
impl_unary!(f64x4: Sleef_sqrtd4_u35avx2);
} else if #[cfg(target_feature = "avx")] {
impl_unary!(f32x2[t => f32x4]: Sleef_sqrtf4_u35sse4);
impl_unary!(f32x16[h => f32x8]: Sleef_sqrtf8_u35avx);
impl_unary!(f64x8[h => f64x4]: Sleef_sqrtd4_u35avx);
impl_unary!(f32x4: Sleef_sqrtf4_u35sse4);
impl_unary!(f32x8: Sleef_sqrtf8_u35avx);
impl_unary!(f64x2: Sleef_sqrtd2_u35sse4);
impl_unary!(f64x4: Sleef_sqrtd4_u35avx);
} else if #[cfg(target_feature = "sse4.2")] {
impl_unary!(f32x2[t => f32x4]: Sleef_sqrtf4_u35sse4);
impl_unary!(f32x16[q => f32x4]: Sleef_sqrtf4_u35sse4);
impl_unary!(f64x8[q => f64x2]: Sleef_sqrtd2_u35sse4);
impl_unary!(f32x4: Sleef_sqrtf4_u35sse4);
impl_unary!(f32x8[h => f32x4]: Sleef_sqrtf4_u35sse4);
impl_unary!(f64x2: Sleef_sqrtd2_u35sse4);
impl_unary!(f64x4[h => f64x2]: Sleef_sqrtd2_u35sse4);
} else {
impl_unary!(f32x2[g]: simd_fsqrt);
impl_unary!(f32x16[g]: simd_fsqrt);
impl_unary!(f64x8[g]: simd_fsqrt);
impl_unary!(f32x4[g]: simd_fsqrt);
impl_unary!(f32x8[g]: simd_fsqrt);
impl_unary!(f64x2[g]: simd_fsqrt);
impl_unary!(f64x4[g]: simd_fsqrt);
}
}
} else {
impl_unary!(f32x2[g]: simd_fsqrt);
impl_unary!(f32x4[g]: simd_fsqrt);
impl_unary!(f32x8[g]: simd_fsqrt);
impl_unary!(f32x16[g]: simd_fsqrt);
impl_unary!(f64x2[g]: simd_fsqrt);
impl_unary!(f64x4[g]: simd_fsqrt);
impl_unary!(f64x8[g]: simd_fsqrt);
}
}
|
