convert _mm256_i32gather_ps to const generics

This commit is contained in:
Rémy Rakic 2021-03-06 00:48:03 +01:00 committed by Amanieu d'Antras
parent 13201ca35b
commit 2eaa3f783d

View file

@ -1225,20 +1225,16 @@ pub unsafe fn _mm_mask_i32gather_ps<const SCALE: i32>(
/// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm256_i32gather_ps)
#[inline]
#[target_feature(enable = "avx2")]
#[cfg_attr(test, assert_instr(vgatherdps, scale = 1))]
#[rustc_args_required_const(2)]
#[cfg_attr(test, assert_instr(vgatherdps, SCALE = 1))]
#[rustc_legacy_const_generics(2)]
#[stable(feature = "simd_x86", since = "1.27.0")]
pub unsafe fn _mm256_i32gather_ps(slice: *const f32, offsets: __m256i, scale: i32) -> __m256 {
pub unsafe fn _mm256_i32gather_ps<const SCALE: i32>(slice: *const f32, offsets: __m256i) -> __m256 {
static_assert_imm8_scale!(SCALE);
let zero = _mm256_setzero_ps();
let neg_one = _mm256_set1_ps(-1.0);
let offsets = offsets.as_i32x8();
let slice = slice as *const i8;
macro_rules! call {
($imm8:expr) => {
vpgatherdps(zero, slice, offsets, neg_one, $imm8)
};
}
constify_imm8_gather!(scale, call)
vpgatherdps(zero, slice, offsets, neg_one, SCALE as i8)
}
/// Returns values from `slice` at offsets determined by `offsets * scale`,
@ -5640,11 +5636,8 @@ mod tests {
j += 1.0;
}
// A multiplier of 4 is word-addressing for f32s
let r = _mm256_i32gather_ps(
arr.as_ptr(),
_mm256_setr_epi32(0, 16, 32, 48, 1, 2, 3, 4),
4,
);
let r =
_mm256_i32gather_ps::<4>(arr.as_ptr(), _mm256_setr_epi32(0, 16, 32, 48, 1, 2, 3, 4));
assert_eq_m256(r, _mm256_setr_ps(0.0, 16.0, 32.0, 48.0, 1.0, 2.0, 3.0, 4.0));
}