user0/rust
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

convert _mm256_permute4x64_pd to const generics

Browse source
This commit is contained in:
Rémy Rakic 2021-03-06 02:33:55 +01:00 • committed by Amanieu d'Antras
parent 913f5b9af4
commit fb871cd7c9
1 changed files with 15 additions and 47 deletions
Download patch file Download diff file Expand all files Collapse all files

62
library/stdarch/crates/core_arch/src/x86/avx2.rs
View file

@ -2351,53 +2351,21 @@ pub unsafe fn _mm256_permute2x128_si256<const IMM8: i32>(a: __m256i, b: __m256i)
/// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_permute4x64_pd)
#[inline]
#[target_feature(enable = "avx2")]
#[cfg_attr(test, assert_instr(vpermpd, imm8 = 1))]
#[rustc_args_required_const(1)]
#[cfg_attr(test, assert_instr(vpermpd, IMM8 = 1))]
#[rustc_legacy_const_generics(1)]
#[stable(feature = "simd_x86", since = "1.27.0")]
pub unsafe fn _mm256_permute4x64_pd(a: __m256d, imm8: i32) -> __m256d {
let imm8 = (imm8 & 0xFF) as u8;
let undef = _mm256_undefined_pd();
macro_rules! shuffle_done {
($x01:expr, $x23:expr, $x45:expr, $x67:expr) => {
simd_shuffle4(a, undef, [$x01, $x23, $x45, $x67])
};
}
macro_rules! shuffle_x67 {
($x01:expr, $x23:expr, $x45:expr) => {
match (imm8 >> 6) & 0b11 {
0b00 => shuffle_done!($x01, $x23, $x45, 0),
0b01 => shuffle_done!($x01, $x23, $x45, 1),
0b10 => shuffle_done!($x01, $x23, $x45, 2),
_ => shuffle_done!($x01, $x23, $x45, 3),
}
};
}
macro_rules! shuffle_x45 {
($x01:expr, $x23:expr) => {
match (imm8 >> 4) & 0b11 {
0b00 => shuffle_x67!($x01, $x23, 0),
0b01 => shuffle_x67!($x01, $x23, 1),
0b10 => shuffle_x67!($x01, $x23, 2),
_ => shuffle_x67!($x01, $x23, 3),
}
};
}
macro_rules! shuffle_x23 {
($x01:expr) => {
match (imm8 >> 2) & 0b11 {
0b00 => shuffle_x45!($x01, 0),
0b01 => shuffle_x45!($x01, 1),
0b10 => shuffle_x45!($x01, 2),
_ => shuffle_x45!($x01, 3),
}
};
}
match imm8 & 0b11 {
0b00 => shuffle_x23!(0),
0b01 => shuffle_x23!(1),
0b10 => shuffle_x23!(2),
_ => shuffle_x23!(3),
}
pub unsafe fn _mm256_permute4x64_pd<const IMM8: i32>(a: __m256d) -> __m256d {
static_assert_imm8!(IMM8);
simd_shuffle4(
a,
_mm256_undefined_pd(),
[
IMM8 as u32 & 0b11,
(IMM8 as u32 >> 2) & 0b11,
(IMM8 as u32 >> 4) & 0b11,
(IMM8 as u32 >> 6) & 0b11,
],
)
}
/// Shuffles eight 32-bit foating-point elements in `a` across lanes using
@ -5406,7 +5374,7 @@ mod tests {
#[simd_test(enable = "avx2")]
unsafe fn test_mm256_permute4x64_pd() {
let a = _mm256_setr_pd(1., 2., 3., 4.);
let r = _mm256_permute4x64_pd(a, 0b00_01_00_11);
let r = _mm256_permute4x64_pd::<0b00_01_00_11>(a);
let e = _mm256_setr_pd(4., 1., 2., 1.);
assert_eq_m256d(r, e);
}