Convert some SSE2 intrinsics to const generics (#1021)
This commit is contained in:
Rémy Rakic
GitHub
parent
78ab9042cb
commit
5438f1fb7e
5 changed files with 87 additions and 219 deletions
|
|
@ -1,5 +1,21 @@
|
|||
//! Utility macros.
|
||||
|
||||
// Helper struct used to trigger const eval errors when a const generic immediate value is
|
||||
// out of range.
|
||||
pub(crate) struct ValidateConstImm8<const imm8: i32>();
|
||||
impl<const imm8: i32> ValidateConstImm8<imm8> {
|
||||
pub(crate) const VALID: () = {
|
||||
let _ = 1 / ((imm8 >= 0 && imm8 <= 255) as usize);
|
||||
};
|
||||
}
|
||||
|
||||
#[allow(unused)]
|
||||
macro_rules! static_assert_imm8 {
|
||||
($imm:ident) => {
|
||||
let _ = $crate::core_arch::macros::ValidateConstImm8::<$imm>::VALID;
|
||||
};
|
||||
}
|
||||
|
||||
#[allow(unused)]
|
||||
macro_rules! static_assert {
|
||||
($imm:ident : $ty:ty where $e:expr) => {
|
||||
|
|
@ -320,48 +336,6 @@ macro_rules! constify_imm5 {
|
|||
};
|
||||
}
|
||||
|
||||
//immediate value: -16:15
|
||||
#[allow(unused)]
|
||||
macro_rules! constify_imm5 {
|
||||
($imm8:expr, $expand:ident) => {
|
||||
#[allow(overflowing_literals)]
|
||||
match ($imm8) & 0b1_1111 {
|
||||
0 => $expand!(0),
|
||||
1 => $expand!(1),
|
||||
2 => $expand!(2),
|
||||
3 => $expand!(3),
|
||||
4 => $expand!(4),
|
||||
5 => $expand!(5),
|
||||
6 => $expand!(6),
|
||||
7 => $expand!(7),
|
||||
8 => $expand!(8),
|
||||
9 => $expand!(9),
|
||||
10 => $expand!(10),
|
||||
11 => $expand!(11),
|
||||
12 => $expand!(12),
|
||||
13 => $expand!(13),
|
||||
14 => $expand!(14),
|
||||
15 => $expand!(15),
|
||||
16 => $expand!(16),
|
||||
17 => $expand!(17),
|
||||
18 => $expand!(18),
|
||||
19 => $expand!(19),
|
||||
20 => $expand!(20),
|
||||
21 => $expand!(21),
|
||||
22 => $expand!(22),
|
||||
23 => $expand!(23),
|
||||
24 => $expand!(24),
|
||||
25 => $expand!(25),
|
||||
26 => $expand!(26),
|
||||
27 => $expand!(27),
|
||||
28 => $expand!(28),
|
||||
29 => $expand!(29),
|
||||
30 => $expand!(30),
|
||||
_ => $expand!(31),
|
||||
}
|
||||
};
|
||||
}
|
||||
|
||||
//immediate value: 0:16
|
||||
#[allow(unused)]
|
||||
macro_rules! constify_imm4 {
|
||||
|
|
|
|||
|
|
@ -5858,7 +5858,7 @@ pub unsafe fn _mm256_maskz_srai_epi16(k: __mmask16, a: __m256i, imm8: u32) -> __
|
|||
pub unsafe fn _mm_mask_srai_epi16(src: __m128i, k: __mmask8, a: __m128i, imm8: u32) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_srai_epi16(a, $imm8)
|
||||
_mm_srai_epi16::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let shf = constify_imm8_sae!(imm8, call);
|
||||
|
|
@ -5875,7 +5875,7 @@ pub unsafe fn _mm_mask_srai_epi16(src: __m128i, k: __mmask8, a: __m128i, imm8: u
|
|||
pub unsafe fn _mm_maskz_srai_epi16(k: __mmask8, a: __m128i, imm8: u32) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_srai_epi16(a, $imm8)
|
||||
_mm_srai_epi16::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let shf = constify_imm8_sae!(imm8, call);
|
||||
|
|
@ -7414,7 +7414,7 @@ pub unsafe fn _mm_mask_shufflelo_epi16(
|
|||
) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_shufflelo_epi16(a, $imm8)
|
||||
_mm_shufflelo_epi16::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let shuffle = constify_imm8_sae!(imm8, call);
|
||||
|
|
@ -7431,7 +7431,7 @@ pub unsafe fn _mm_mask_shufflelo_epi16(
|
|||
pub unsafe fn _mm_maskz_shufflelo_epi16(k: __mmask8, a: __m128i, imm8: i32) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_shufflelo_epi16(a, $imm8)
|
||||
_mm_shufflelo_epi16::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let shuffle = constify_imm8_sae!(imm8, call);
|
||||
|
|
@ -7592,7 +7592,7 @@ pub unsafe fn _mm_mask_shufflehi_epi16(
|
|||
) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_shufflehi_epi16(a, $imm8)
|
||||
_mm_shufflehi_epi16::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let shuffle = constify_imm8_sae!(imm8, call);
|
||||
|
|
@ -7609,7 +7609,7 @@ pub unsafe fn _mm_mask_shufflehi_epi16(
|
|||
pub unsafe fn _mm_maskz_shufflehi_epi16(k: __mmask8, a: __m128i, imm8: i32) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_shufflehi_epi16(a, $imm8)
|
||||
_mm_shufflehi_epi16::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let shuffle = constify_imm8_sae!(imm8, call);
|
||||
|
|
|
|||
|
|
@ -19238,7 +19238,7 @@ pub unsafe fn _mm256_maskz_srai_epi32(k: __mmask8, a: __m256i, imm8: u32) -> __m
|
|||
pub unsafe fn _mm_mask_srai_epi32(src: __m128i, k: __mmask8, a: __m128i, imm8: u32) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_srai_epi32(a, $imm8)
|
||||
_mm_srai_epi32::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let shf = constify_imm8_sae!(imm8, call);
|
||||
|
|
@ -19255,7 +19255,7 @@ pub unsafe fn _mm_mask_srai_epi32(src: __m128i, k: __mmask8, a: __m128i, imm8: u
|
|||
pub unsafe fn _mm_maskz_srai_epi32(k: __mmask8, a: __m128i, imm8: u32) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_srai_epi32(a, $imm8)
|
||||
_mm_srai_epi32::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let shf = constify_imm8_sae!(imm8, call);
|
||||
|
|
@ -22495,7 +22495,7 @@ pub unsafe fn _mm_mask_shuffle_epi32(
|
|||
) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_shuffle_epi32(a, $imm8)
|
||||
_mm_shuffle_epi32::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let r = constify_imm8_sae!(imm8, call);
|
||||
|
|
@ -22512,7 +22512,7 @@ pub unsafe fn _mm_mask_shuffle_epi32(
|
|||
pub unsafe fn _mm_maskz_shuffle_epi32(k: __mmask8, a: __m128i, imm8: _MM_PERM_ENUM) -> __m128i {
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
_mm_shuffle_epi32(a, $imm8)
|
||||
_mm_shuffle_epi32::<$imm8>(a)
|
||||
};
|
||||
}
|
||||
let r = constify_imm8_sae!(imm8, call);
|
||||
|
|
|
|||
|
|
@ -1010,7 +1010,7 @@ pub const fn _MM_SHUFFLE(z: u32, y: u32, x: u32, w: u32) -> i32 {
|
|||
#[rustc_legacy_const_generics(2)]
|
||||
#[stable(feature = "simd_x86", since = "1.27.0")]
|
||||
pub unsafe fn _mm_shuffle_ps<const mask: i32>(a: __m128, b: __m128) -> __m128 {
|
||||
static_assert!(mask: i32 where mask >= 0 && mask <= 255);
|
||||
static_assert_imm8!(mask);
|
||||
simd_shuffle4(
|
||||
a,
|
||||
b,
|
||||
|
|
|
|||
|
|
@ -594,16 +594,11 @@ pub unsafe fn _mm_sll_epi64(a: __m128i, count: __m128i) -> __m128i {
|
|||
#[inline]
|
||||
#[target_feature(enable = "sse2")]
|
||||
#[cfg_attr(test, assert_instr(psraw, imm8 = 1))]
|
||||
#[rustc_args_required_const(1)]
|
||||
#[rustc_legacy_const_generics(1)]
|
||||
#[stable(feature = "simd_x86", since = "1.27.0")]
|
||||
pub unsafe fn _mm_srai_epi16(a: __m128i, imm8: i32) -> __m128i {
|
||||
let a = a.as_i16x8();
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
transmute(psraiw(a, $imm8))
|
||||
};
|
||||
}
|
||||
constify_imm8!(imm8, call)
|
||||
pub unsafe fn _mm_srai_epi16<const imm8: i32>(a: __m128i) -> __m128i {
|
||||
static_assert_imm8!(imm8);
|
||||
transmute(psraiw(a.as_i16x8(), imm8))
|
||||
}
|
||||
|
||||
/// Shifts packed 16-bit integers in `a` right by `count` while shifting in sign
|
||||
|
|
@ -625,16 +620,11 @@ pub unsafe fn _mm_sra_epi16(a: __m128i, count: __m128i) -> __m128i {
|
|||
#[inline]
|
||||
#[target_feature(enable = "sse2")]
|
||||
#[cfg_attr(test, assert_instr(psrad, imm8 = 1))]
|
||||
#[rustc_args_required_const(1)]
|
||||
#[rustc_legacy_const_generics(1)]
|
||||
#[stable(feature = "simd_x86", since = "1.27.0")]
|
||||
pub unsafe fn _mm_srai_epi32(a: __m128i, imm8: i32) -> __m128i {
|
||||
let a = a.as_i32x4();
|
||||
macro_rules! call {
|
||||
($imm8:expr) => {
|
||||
transmute(psraid(a, $imm8))
|
||||
};
|
||||
}
|
||||
constify_imm8!(imm8, call)
|
||||
pub unsafe fn _mm_srai_epi32<const imm8: i32>(a: __m128i) -> __m128i {
|
||||
static_assert_imm8!(imm8);
|
||||
transmute(psraid(a.as_i32x4(), imm8))
|
||||
}
|
||||
|
||||
/// Shifts packed 32-bit integers in `a` right by `count` while shifting in sign
|
||||
|
|
@ -1461,60 +1451,21 @@ pub unsafe fn _mm_movemask_epi8(a: __m128i) -> i32 {
|
|||
#[inline]
|
||||
#[target_feature(enable = "sse2")]
|
||||
#[cfg_attr(test, assert_instr(pshufd, imm8 = 9))]
|
||||
#[rustc_args_required_const(1)]
|
||||
#[rustc_legacy_const_generics(1)]
|
||||
#[stable(feature = "simd_x86", since = "1.27.0")]
|
||||
pub unsafe fn _mm_shuffle_epi32(a: __m128i, imm8: i32) -> __m128i {
|
||||
// simd_shuffleX requires that its selector parameter be made up of
|
||||
// constant values, but we can't enforce that here. In spirit, we need
|
||||
// to write a `match` on all possible values of a byte, and for each value,
|
||||
// hard-code the correct `simd_shuffleX` call using only constants. We
|
||||
// then hope for LLVM to do the rest.
|
||||
//
|
||||
// Of course, that's... awful. So we try to use macros to do it for us.
|
||||
let imm8 = (imm8 & 0xFF) as u8;
|
||||
pub unsafe fn _mm_shuffle_epi32<const imm8: i32>(a: __m128i) -> __m128i {
|
||||
static_assert_imm8!(imm8);
|
||||
let a = a.as_i32x4();
|
||||
|
||||
macro_rules! shuffle_done {
|
||||
($x01:expr, $x23:expr, $x45:expr, $x67:expr) => {
|
||||
simd_shuffle4(a, a, [$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),
|
||||
}
|
||||
};
|
||||
}
|
||||
let x: i32x4 = match imm8 & 0b11 {
|
||||
0b00 => shuffle_x23!(0),
|
||||
0b01 => shuffle_x23!(1),
|
||||
0b10 => shuffle_x23!(2),
|
||||
_ => shuffle_x23!(3),
|
||||
};
|
||||
let x: i32x4 = simd_shuffle4(
|
||||
a,
|
||||
a,
|
||||
[
|
||||
imm8 as u32 & 0b11,
|
||||
(imm8 as u32 >> 2) & 0b11,
|
||||
(imm8 as u32 >> 4) & 0b11,
|
||||
(imm8 as u32 >> 6) & 0b11,
|
||||
],
|
||||
);
|
||||
transmute(x)
|
||||
}
|
||||
|
||||
|
|
@ -1528,53 +1479,25 @@ pub unsafe fn _mm_shuffle_epi32(a: __m128i, imm8: i32) -> __m128i {
|
|||
#[inline]
|
||||
#[target_feature(enable = "sse2")]
|
||||
#[cfg_attr(test, assert_instr(pshufhw, imm8 = 9))]
|
||||
#[rustc_args_required_const(1)]
|
||||
#[rustc_legacy_const_generics(1)]
|
||||
#[stable(feature = "simd_x86", since = "1.27.0")]
|
||||
pub unsafe fn _mm_shufflehi_epi16(a: __m128i, imm8: i32) -> __m128i {
|
||||
// See _mm_shuffle_epi32.
|
||||
let imm8 = (imm8 & 0xFF) as u8;
|
||||
pub unsafe fn _mm_shufflehi_epi16<const imm8: i32>(a: __m128i) -> __m128i {
|
||||
static_assert_imm8!(imm8);
|
||||
let a = a.as_i16x8();
|
||||
macro_rules! shuffle_done {
|
||||
($x01:expr, $x23:expr, $x45:expr, $x67:expr) => {
|
||||
simd_shuffle8(a, a, [0, 1, 2, 3, $x01 + 4, $x23 + 4, $x45 + 4, $x67 + 4])
|
||||
};
|
||||
}
|
||||
macro_rules! shuffle_x67 {
|
||||
($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),
|
||||
}
|
||||
};
|
||||
}
|
||||
let x: i16x8 = match imm8 & 0b11 {
|
||||
0b00 => shuffle_x23!(0),
|
||||
0b01 => shuffle_x23!(1),
|
||||
0b10 => shuffle_x23!(2),
|
||||
_ => shuffle_x23!(3),
|
||||
};
|
||||
let x: i16x8 = simd_shuffle8(
|
||||
a,
|
||||
a,
|
||||
[
|
||||
0,
|
||||
1,
|
||||
2,
|
||||
3,
|
||||
(imm8 as u32 & 0b11) + 4,
|
||||
((imm8 as u32 >> 2) & 0b11) + 4,
|
||||
((imm8 as u32 >> 4) & 0b11) + 4,
|
||||
((imm8 as u32 >> 6) & 0b11) + 4,
|
||||
],
|
||||
);
|
||||
transmute(x)
|
||||
}
|
||||
|
||||
|
|
@ -1588,54 +1511,25 @@ pub unsafe fn _mm_shufflehi_epi16(a: __m128i, imm8: i32) -> __m128i {
|
|||
#[inline]
|
||||
#[target_feature(enable = "sse2")]
|
||||
#[cfg_attr(test, assert_instr(pshuflw, imm8 = 9))]
|
||||
#[rustc_args_required_const(1)]
|
||||
#[rustc_legacy_const_generics(1)]
|
||||
#[stable(feature = "simd_x86", since = "1.27.0")]
|
||||
pub unsafe fn _mm_shufflelo_epi16(a: __m128i, imm8: i32) -> __m128i {
|
||||
// See _mm_shuffle_epi32.
|
||||
let imm8 = (imm8 & 0xFF) as u8;
|
||||
pub unsafe fn _mm_shufflelo_epi16<const imm8: i32>(a: __m128i) -> __m128i {
|
||||
static_assert_imm8!(imm8);
|
||||
let a = a.as_i16x8();
|
||||
|
||||
macro_rules! shuffle_done {
|
||||
($x01:expr, $x23:expr, $x45:expr, $x67:expr) => {
|
||||
simd_shuffle8(a, a, [$x01, $x23, $x45, $x67, 4, 5, 6, 7])
|
||||
};
|
||||
}
|
||||
macro_rules! shuffle_x67 {
|
||||
($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),
|
||||
}
|
||||
};
|
||||
}
|
||||
let x: i16x8 = match imm8 & 0b11 {
|
||||
0b00 => shuffle_x23!(0),
|
||||
0b01 => shuffle_x23!(1),
|
||||
0b10 => shuffle_x23!(2),
|
||||
_ => shuffle_x23!(3),
|
||||
};
|
||||
let x: i16x8 = simd_shuffle8(
|
||||
a,
|
||||
a,
|
||||
[
|
||||
imm8 as u32 & 0b11,
|
||||
(imm8 as u32 >> 2) & 0b11,
|
||||
(imm8 as u32 >> 4) & 0b11,
|
||||
(imm8 as u32 >> 6) & 0b11,
|
||||
4,
|
||||
5,
|
||||
6,
|
||||
7,
|
||||
],
|
||||
);
|
||||
transmute(x)
|
||||
}
|
||||
|
||||
|
|
@ -3594,7 +3488,7 @@ mod tests {
|
|||
|
||||
#[simd_test(enable = "sse2")]
|
||||
unsafe fn test_mm_srai_epi16() {
|
||||
let r = _mm_srai_epi16(_mm_set1_epi16(-1), 1);
|
||||
let r = _mm_srai_epi16::<1>(_mm_set1_epi16(-1));
|
||||
assert_eq_m128i(r, _mm_set1_epi16(-1));
|
||||
}
|
||||
|
||||
|
|
@ -3608,7 +3502,7 @@ mod tests {
|
|||
|
||||
#[simd_test(enable = "sse2")]
|
||||
unsafe fn test_mm_srai_epi32() {
|
||||
let r = _mm_srai_epi32(_mm_set1_epi32(-1), 1);
|
||||
let r = _mm_srai_epi32::<1>(_mm_set1_epi32(-1));
|
||||
assert_eq_m128i(r, _mm_set1_epi32(-1));
|
||||
}
|
||||
|
||||
|
|
@ -4107,7 +4001,7 @@ mod tests {
|
|||
#[simd_test(enable = "sse2")]
|
||||
unsafe fn test_mm_shuffle_epi32() {
|
||||
let a = _mm_setr_epi32(5, 10, 15, 20);
|
||||
let r = _mm_shuffle_epi32(a, 0b00_01_01_11);
|
||||
let r = _mm_shuffle_epi32::<0b00_01_01_11>(a);
|
||||
let e = _mm_setr_epi32(20, 10, 10, 5);
|
||||
assert_eq_m128i(r, e);
|
||||
}
|
||||
|
|
@ -4115,7 +4009,7 @@ mod tests {
|
|||
#[simd_test(enable = "sse2")]
|
||||
unsafe fn test_mm_shufflehi_epi16() {
|
||||
let a = _mm_setr_epi16(1, 2, 3, 4, 5, 10, 15, 20);
|
||||
let r = _mm_shufflehi_epi16(a, 0b00_01_01_11);
|
||||
let r = _mm_shufflehi_epi16::<0b00_01_01_11>(a);
|
||||
let e = _mm_setr_epi16(1, 2, 3, 4, 20, 10, 10, 5);
|
||||
assert_eq_m128i(r, e);
|
||||
}
|
||||
|
|
@ -4123,7 +4017,7 @@ mod tests {
|
|||
#[simd_test(enable = "sse2")]
|
||||
unsafe fn test_mm_shufflelo_epi16() {
|
||||
let a = _mm_setr_epi16(5, 10, 15, 20, 1, 2, 3, 4);
|
||||
let r = _mm_shufflelo_epi16(a, 0b00_01_01_11);
|
||||
let r = _mm_shufflelo_epi16::<0b00_01_01_11>(a);
|
||||
let e = _mm_setr_epi16(20, 10, 10, 5, 1, 2, 3, 4);
|
||||
assert_eq_m128i(r, e);
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue