Skip to content

use simd_shuffle macros on wasm32 #1162

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
May 13, 2021
Merged

use simd_shuffle macros on wasm32 #1162

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions crates/core_arch/src/macros.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,7 +95,7 @@ macro_rules! types {

#[allow(unused_macros)]
macro_rules! simd_shuffle2 {
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+> $idx:expr $(,)?) => {{
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+ $(,)?> $idx:expr $(,)?) => {{
struct ConstParam<$(const $imm: $ty),+>;
impl<$(const $imm: $ty),+> ConstParam<$($imm),+> {
const IDX: [u32; 2] = $idx;
Expand All @@ -111,7 +111,7 @@ macro_rules! simd_shuffle2 {

#[allow(unused_macros)]
macro_rules! simd_shuffle4 {
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+> $idx:expr $(,)?) => {{
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+ $(,)?> $idx:expr $(,)?) => {{
struct ConstParam<$(const $imm: $ty),+>;
impl<$(const $imm: $ty),+> ConstParam<$($imm),+> {
const IDX: [u32; 4] = $idx;
Expand All @@ -127,7 +127,7 @@ macro_rules! simd_shuffle4 {

#[allow(unused_macros)]
macro_rules! simd_shuffle8 {
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+> $idx:expr $(,)?) => {{
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+ $(,)?> $idx:expr $(,)?) => {{
struct ConstParam<$(const $imm: $ty),+>;
impl<$(const $imm: $ty),+> ConstParam<$($imm),+> {
const IDX: [u32; 8] = $idx;
Expand All @@ -143,7 +143,7 @@ macro_rules! simd_shuffle8 {

#[allow(unused_macros)]
macro_rules! simd_shuffle16 {
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+> $idx:expr $(,)?) => {{
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+ $(,)?> $idx:expr $(,)?) => {{
struct ConstParam<$(const $imm: $ty),+>;
impl<$(const $imm: $ty),+> ConstParam<$($imm),+> {
const IDX: [u32; 16] = $idx;
Expand All @@ -159,7 +159,7 @@ macro_rules! simd_shuffle16 {

#[allow(unused_macros)]
macro_rules! simd_shuffle32 {
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+> $idx:expr $(,)?) => {{
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+> $(,)? $idx:expr $(,)?) => {{
struct ConstParam<$(const $imm: $ty),+>;
impl<$(const $imm: $ty),+> ConstParam<$($imm),+> {
const IDX: [u32; 32] = $idx;
Expand All @@ -175,7 +175,7 @@ macro_rules! simd_shuffle32 {

#[allow(unused_macros)]
macro_rules! simd_shuffle64 {
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+> $idx:expr $(,)?) => {{
($x:expr, $y:expr, <$(const $imm:ident : $ty:ty),+ $(,)?> $idx:expr $(,)?) => {{
struct ConstParam<$(const $imm: $ty),+>;
impl<$(const $imm: $ty),+> ConstParam<$($imm),+> {
const IDX: [u32; 64] = $idx;
Expand Down
128 changes: 82 additions & 46 deletions crates/core_arch/src/wasm32/simd128.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -769,10 +769,27 @@ pub unsafe fn i8x16_shuffle<
static_assert!(I13: usize where I13 < 32);
static_assert!(I14: usize where I14 < 32);
static_assert!(I15: usize where I15 < 32);
let shuf = simd_shuffle16::<simd::u8x16, simd::u8x16>(
let shuf: simd::u8x16 = simd_shuffle16!(
a.as_u8x16(),
b.as_u8x16(),
[
<
const I0: usize,
const I1: usize,
const I2: usize,
const I3: usize,
const I4: usize,
const I5: usize,
const I6: usize,
const I7: usize,
const I8: usize,
const I9: usize,
const I10: usize,
const I11: usize,
const I12: usize,
const I13: usize,
const I14: usize,
const I15: usize,
> [
I0 as u32, I1 as u32, I2 as u32, I3 as u32, I4 as u32, I5 as u32, I6 as u32, I7 as u32,
I8 as u32, I9 as u32, I10 as u32, I11 as u32, I12 as u32, I13 as u32, I14 as u32,
I15 as u32,
Expand Down Expand Up @@ -825,10 +842,19 @@ pub unsafe fn i16x8_shuffle<
static_assert!(I5: usize where I5 < 16);
static_assert!(I6: usize where I6 < 16);
static_assert!(I7: usize where I7 < 16);
let shuf = simd_shuffle8::<simd::u16x8, simd::u16x8>(
let shuf: simd::u16x8 = simd_shuffle8!(
a.as_u16x8(),
b.as_u16x8(),
[
<
const I0: usize,
const I1: usize,
const I2: usize,
const I3: usize,
const I4: usize,
const I5: usize,
const I6: usize,
const I7: usize,
> [
I0 as u32, I1 as u32, I2 as u32, I3 as u32, I4 as u32, I5 as u32, I6 as u32, I7 as u32,
],
);
Expand All @@ -854,10 +880,10 @@ pub unsafe fn i32x4_shuffle<const I0: usize, const I1: usize, const I2: usize, c
static_assert!(I1: usize where I1 < 8);
static_assert!(I2: usize where I2 < 8);
static_assert!(I3: usize where I3 < 8);
let shuf = simd_shuffle4::<simd::u32x4, simd::u32x4>(
let shuf: simd::u32x4 = simd_shuffle4!(
a.as_u32x4(),
b.as_u32x4(),
[I0 as u32, I1 as u32, I2 as u32, I3 as u32],
<const I0: usize, const I1: usize, const I2: usize, const I3: usize> [I0 as u32, I1 as u32, I2 as u32, I3 as u32],
);
transmute(shuf)
}
Expand All @@ -876,10 +902,10 @@ pub unsafe fn i32x4_shuffle<const I0: usize, const I1: usize, const I2: usize, c
pub unsafe fn i64x2_shuffle<const I0: usize, const I1: usize>(a: v128, b: v128) -> v128 {
static_assert!(I0: usize where I0 < 4);
static_assert!(I1: usize where I1 < 4);
let shuf = simd_shuffle2::<simd::u64x2, simd::u64x2>(
let shuf: simd::u64x2 = simd_shuffle2!(
a.as_u64x2(),
b.as_u64x2(),
[I0 as u32, I1 as u32],
<const I0: usize, const I1: usize> [I0 as u32, I1 as u32],
);
transmute(shuf)
}
Expand Down Expand Up @@ -2288,7 +2314,7 @@ pub unsafe fn u16x8_narrow_i32x4(a: v128, b: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i16x8.extend_low_i8x16_s"))]
pub unsafe fn i16x8_extend_low_i8x16(a: v128) -> v128 {
transmute(simd_cast::<_, simd::i16x8>(simd_shuffle8::<_, simd::i8x8>(
transmute(simd_cast::<simd::i8x8, simd::i16x8>(simd_shuffle8!(
a.as_i8x16(),
a.as_i8x16(),
[0, 1, 2, 3, 4, 5, 6, 7],
Expand All @@ -2302,7 +2328,7 @@ pub unsafe fn i16x8_extend_low_i8x16(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i16x8.extend_high_i8x16_s"))]
pub unsafe fn i16x8_extend_high_i8x16(a: v128) -> v128 {
transmute(simd_cast::<_, simd::i16x8>(simd_shuffle8::<_, simd::i8x8>(
transmute(simd_cast::<simd::i8x8, simd::i16x8>(simd_shuffle8!(
a.as_i8x16(),
a.as_i8x16(),
[8, 9, 10, 11, 12, 13, 14, 15],
Expand All @@ -2316,7 +2342,7 @@ pub unsafe fn i16x8_extend_high_i8x16(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i16x8.extend_low_i8x16_u"))]
pub unsafe fn i16x8_extend_low_u8x16(a: v128) -> v128 {
transmute(simd_cast::<_, simd::u16x8>(simd_shuffle8::<_, simd::u8x8>(
transmute(simd_cast::<simd::u8x8, simd::u16x8>(simd_shuffle8!(
a.as_u8x16(),
a.as_u8x16(),
[0, 1, 2, 3, 4, 5, 6, 7],
Expand All @@ -2330,7 +2356,7 @@ pub unsafe fn i16x8_extend_low_u8x16(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i16x8.extend_high_i8x16_u"))]
pub unsafe fn i16x8_extend_high_u8x16(a: v128) -> v128 {
transmute(simd_cast::<_, simd::u16x8>(simd_shuffle8::<_, simd::u8x8>(
transmute(simd_cast::<simd::u8x8, simd::u16x8>(simd_shuffle8!(
a.as_u8x16(),
a.as_u8x16(),
[8, 9, 10, 11, 12, 13, 14, 15],
Expand Down Expand Up @@ -2618,9 +2644,11 @@ pub unsafe fn i32x4_bitmask(a: v128) -> i32 {
#[target_feature(enable = "simd128")]
#[doc(alias("i32x4.extend_low_i16x8_s"))]
pub unsafe fn i32x4_extend_low_i16x8(a: v128) -> v128 {
transmute(simd_cast::<_, simd::i32x4>(
simd_shuffle4::<_, simd::i16x4>(a.as_i16x8(), a.as_i16x8(), [0, 1, 2, 3]),
))
transmute(simd_cast::<simd::i16x4, simd::i32x4>(simd_shuffle4!(
a.as_i16x8(),
a.as_i16x8(),
[0, 1, 2, 3]
)))
}

/// Converts high half of the smaller lane vector to a larger lane
Expand All @@ -2630,9 +2658,11 @@ pub unsafe fn i32x4_extend_low_i16x8(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i32x4.extend_high_i16x8_s"))]
pub unsafe fn i32x4_extend_high_i16x8(a: v128) -> v128 {
transmute(simd_cast::<_, simd::i32x4>(
simd_shuffle4::<_, simd::i16x4>(a.as_i16x8(), a.as_i16x8(), [4, 5, 6, 7]),
))
transmute(simd_cast::<simd::i16x4, simd::i32x4>(simd_shuffle4!(
a.as_i16x8(),
a.as_i16x8(),
[4, 5, 6, 7]
)))
}

/// Converts low half of the smaller lane vector to a larger lane
Expand All @@ -2642,9 +2672,11 @@ pub unsafe fn i32x4_extend_high_i16x8(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i32x4.extend_low_i16x8_u"))]
pub unsafe fn i32x4_extend_low_u16x8(a: v128) -> v128 {
transmute(simd_cast::<_, simd::u32x4>(
simd_shuffle4::<_, simd::u16x4>(a.as_u16x8(), a.as_u16x8(), [0, 1, 2, 3]),
))
transmute(simd_cast::<simd::u16x4, simd::u32x4>(simd_shuffle4!(
a.as_u16x8(),
a.as_u16x8(),
[0, 1, 2, 3]
)))
}

/// Converts high half of the smaller lane vector to a larger lane
Expand All @@ -2654,9 +2686,11 @@ pub unsafe fn i32x4_extend_low_u16x8(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i32x4.extend_high_i16x8_u"))]
pub unsafe fn i32x4_extend_high_u16x8(a: v128) -> v128 {
transmute(simd_cast::<_, simd::u32x4>(
simd_shuffle4::<_, simd::u16x4>(a.as_u16x8(), a.as_u16x8(), [4, 5, 6, 7]),
))
transmute(simd_cast::<simd::u16x4, simd::u32x4>(simd_shuffle4!(
a.as_u16x8(),
a.as_u16x8(),
[4, 5, 6, 7]
)))
}

/// Shifts each lane to the left by the specified number of bits.
Expand Down Expand Up @@ -2881,9 +2915,11 @@ pub unsafe fn i64x2_bitmask(a: v128) -> i32 {
#[target_feature(enable = "simd128")]
#[doc(alias("i64x2.extend_low_i32x4_s"))]
pub unsafe fn i64x2_extend_low_i32x4(a: v128) -> v128 {
transmute(simd_cast::<_, simd::i64x2>(
simd_shuffle2::<_, simd::i32x2>(a.as_i32x4(), a.as_i32x4(), [0, 1]),
))
transmute(simd_cast::<simd::i32x2, simd::i64x2>(simd_shuffle2!(
a.as_i32x4(),
a.as_i32x4(),
[0, 1]
)))
}

/// Converts high half of the smaller lane vector to a larger lane
Expand All @@ -2893,9 +2929,11 @@ pub unsafe fn i64x2_extend_low_i32x4(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i64x2.extend_high_i32x4_s"))]
pub unsafe fn i64x2_extend_high_i32x4(a: v128) -> v128 {
transmute(simd_cast::<_, simd::i64x2>(
simd_shuffle2::<_, simd::i32x2>(a.as_i32x4(), a.as_i32x4(), [2, 3]),
))
transmute(simd_cast::<simd::i32x2, simd::i64x2>(simd_shuffle2!(
a.as_i32x4(),
a.as_i32x4(),
[2, 3]
)))
}

/// Converts low half of the smaller lane vector to a larger lane
Expand All @@ -2905,9 +2943,11 @@ pub unsafe fn i64x2_extend_high_i32x4(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i64x2.extend_low_i32x4_u"))]
pub unsafe fn i64x2_extend_low_u32x4(a: v128) -> v128 {
transmute(simd_cast::<_, simd::i64x2>(
simd_shuffle2::<_, simd::u32x2>(a.as_u32x4(), a.as_u32x4(), [0, 1]),
))
transmute(simd_cast::<simd::u32x2, simd::i64x2>(simd_shuffle2!(
a.as_u32x4(),
a.as_u32x4(),
[0, 1]
)))
}

/// Converts high half of the smaller lane vector to a larger lane
Expand All @@ -2917,9 +2957,11 @@ pub unsafe fn i64x2_extend_low_u32x4(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i64x2.extend_high_i32x4_u"))]
pub unsafe fn i64x2_extend_high_u32x4(a: v128) -> v128 {
transmute(simd_cast::<_, simd::i64x2>(
simd_shuffle2::<_, simd::u32x2>(a.as_u32x4(), a.as_u32x4(), [2, 3]),
))
transmute(simd_cast::<simd::u32x2, simd::i64x2>(simd_shuffle2!(
a.as_u32x4(),
a.as_u32x4(),
[2, 3]
)))
}

/// Shifts each lane to the left by the specified number of bits.
Expand Down Expand Up @@ -3386,7 +3428,7 @@ pub unsafe fn f32x4_convert_u32x4(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i32x4.trunc_sat_f64x2_s_zero"))]
pub unsafe fn i32x4_trunc_sat_f64x2_zero(a: v128) -> v128 {
transmute(simd_shuffle4::<simd::i32x2, simd::i32x4>(
transmute::<simd::i32x4, v128>(simd_shuffle4!(
llvm_i32x2_trunc_sat_f64x2_s(a.as_f64x2()),
simd::i32x2::splat(0),
[0, 1, 2, 3],
Expand All @@ -3406,7 +3448,7 @@ pub unsafe fn i32x4_trunc_sat_f64x2_zero(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("i32x4.trunc_sat_f64x2_u_zero"))]
pub unsafe fn u32x4_trunc_sat_f64x2_zero(a: v128) -> v128 {
transmute(simd_shuffle4::<simd::i32x2, simd::i32x4>(
transmute::<simd::i32x4, v128>(simd_shuffle4!(
llvm_i32x2_trunc_sat_f64x2_u(a.as_f64x2()),
simd::i32x2::splat(0),
[0, 1, 2, 3],
Expand All @@ -3419,10 +3461,7 @@ pub unsafe fn u32x4_trunc_sat_f64x2_zero(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("f64x2.convert_low_i32x4_s"))]
pub unsafe fn f64x2_convert_low_i32x4(a: v128) -> v128 {
transmute(simd_cast::<_, simd::f64x2>(simd_shuffle2::<
simd::i32x4,
simd::i32x2,
>(
transmute(simd_cast::<simd::i32x2, simd::f64x2>(simd_shuffle2!(
a.as_i32x4(),
a.as_i32x4(),
[0, 1],
Expand All @@ -3435,10 +3474,7 @@ pub unsafe fn f64x2_convert_low_i32x4(a: v128) -> v128 {
#[target_feature(enable = "simd128")]
#[doc(alias("f64x2.convert_low_i32x4_u"))]
pub unsafe fn f64x2_convert_low_u32x4(a: v128) -> v128 {
transmute(simd_cast::<_, simd::f64x2>(simd_shuffle2::<
simd::u32x4,
simd::u32x2,
>(
transmute(simd_cast::<simd::u32x2, simd::f64x2>(simd_shuffle2!(
a.as_u32x4(),
a.as_u32x4(),
[0, 1],
Expand Down