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

convert _mm256_mask_i64gather_pd to const generics

Browse source
This commit is contained in:
Rémy Rakic 2021-03-06 01:48:28 +01:00 • committed by Amanieu d'Antras
parent b0fcace623
commit 7ca45bebaa
1 changed files with 6 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

18
library/stdarch/crates/core_arch/src/x86/avx2.rs
View file

@ -1797,24 +1797,19 @@ pub unsafe fn _mm256_i64gather_pd<const SCALE: i32>(
/// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_mask_i64gather_pd)
#[inline]
#[target_feature(enable = "avx2")]
#[cfg_attr(test, assert_instr(vgatherqpd, scale = 1))]
#[rustc_args_required_const(4)]
#[cfg_attr(test, assert_instr(vgatherqpd, SCALE = 1))]
#[rustc_legacy_const_generics(4)]
#[stable(feature = "simd_x86", since = "1.27.0")]
pub unsafe fn _mm256_mask_i64gather_pd(
pub unsafe fn _mm256_mask_i64gather_pd<const SCALE: i32>(
src: __m256d,
slice: *const f64,
offsets: __m256i,
mask: __m256d,
scale: i32,
) -> __m256d {
static_assert_imm8_scale!(SCALE);
let slice = slice as *const i8;
let offsets = offsets.as_i64x4();
macro_rules! call {
($imm8:expr) => {
vpgatherqpd(src, slice, offsets, mask, $imm8)
};
}
constify_imm8_gather!(scale, call)
vpgatherqpd(src, slice, offsets, mask, SCALE as i8)
}
/// Copies `a` to `dst`, then insert 128 bits (of integer data) from `b` at the
@ -5917,12 +5912,11 @@ mod tests {
j += 1.0;
}
// A multiplier of 8 is word-addressing for f64s
let r = _mm256_mask_i64gather_pd(
let r = _mm256_mask_i64gather_pd::<8>(
_mm256_set1_pd(256.0),
arr.as_ptr(),
_mm256_setr_epi64x(0, 16, 64, 96),
_mm256_setr_pd(-1.0, -1.0, -1.0, 0.0),
8,
);
assert_eq_m256d(r, _mm256_setr_pd(0.0, 16.0, 64.0, 256.0));
}