Skip to content

Use new bitmask intrinsics with byte arrays #159

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
workingjubilee merged 1 commit into from
Nov 13, 2021
Merged

Use new bitmask intrinsics with byte arrays #159

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
9 changes: 0 additions & 9 deletions crates/core_simd/src/lane_count.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,34 +15,25 @@ impl<const LANES: usize> LaneCount<LANES> {
pub trait SupportedLaneCount: Sealed {
#[doc(hidden)]
type BitMask: Copy + Default + AsRef<[u8]> + AsMut<[u8]>;

#[doc(hidden)]
type IntBitMask;
}

impl<const LANES: usize> Sealed for LaneCount<LANES> {}

impl SupportedLaneCount for LaneCount<1> {
type BitMask = [u8; 1];
type IntBitMask = u8;
}
impl SupportedLaneCount for LaneCount<2> {
type BitMask = [u8; 1];
type IntBitMask = u8;
}
impl SupportedLaneCount for LaneCount<4> {
type BitMask = [u8; 1];
type IntBitMask = u8;
}
impl SupportedLaneCount for LaneCount<8> {
type BitMask = [u8; 1];
type IntBitMask = u8;
}
impl SupportedLaneCount for LaneCount<16> {
type BitMask = [u8; 2];
type IntBitMask = u16;
}
impl SupportedLaneCount for LaneCount<32> {
type BitMask = [u8; 4];
type IntBitMask = u32;
}
19 changes: 6 additions & 13 deletions crates/core_simd/src/masks/bitmask.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,24 +101,17 @@ where
#[inline]
pub fn to_int(self) -> Simd<T, LANES> {
unsafe {
let mask: <LaneCount<LANES> as SupportedLaneCount>::IntBitMask =
core::mem::transmute_copy(&self);
intrinsics::simd_select_bitmask(mask, Simd::splat(T::TRUE), Simd::splat(T::FALSE))
crate::intrinsics::simd_select_bitmask(
self.0,
Simd::splat(T::TRUE),
Simd::splat(T::FALSE),
)
}
}

#[inline]
pub unsafe fn from_int_unchecked(value: Simd<T, LANES>) -> Self {
// TODO remove the transmute when rustc is more flexible
assert_eq!(
core::mem::size_of::<<LaneCount::<LANES> as SupportedLaneCount>::BitMask>(),
core::mem::size_of::<<LaneCount::<LANES> as SupportedLaneCount>::IntBitMask>(),
);
unsafe {
let mask: <LaneCount<LANES> as SupportedLaneCount>::IntBitMask =
intrinsics::simd_bitmask(value);
Self(core::mem::transmute_copy(&mask), PhantomData)
}
Self(crate::intrinsics::simd_bitmask(value), PhantomData)
}

#[cfg(feature = "generic_const_exprs")]
Expand Down
19 changes: 2 additions & 17 deletions crates/core_simd/src/masks/full_masks.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -106,15 +106,8 @@ where
#[inline]
pub fn to_bitmask(self) -> [u8; LaneCount::<LANES>::BITMASK_LEN] {
unsafe {
// TODO remove the transmute when rustc can use arrays of u8 as bitmasks
assert_eq!(
core::mem::size_of::<<LaneCount::<LANES> as SupportedLaneCount>::IntBitMask>(),
LaneCount::<LANES>::BITMASK_LEN,
);
let bitmask: <LaneCount<LANES> as SupportedLaneCount>::IntBitMask =
intrinsics::simd_bitmask(self.0);
let mut bitmask: [u8; LaneCount::<LANES>::BITMASK_LEN] =
core::mem::transmute_copy(&bitmask);
crate::intrinsics::simd_bitmask(self.0);

// There is a bug where LLVM appears to implement this operation with the wrong
// bit order.
Expand Down Expand Up @@ -142,15 +135,7 @@ where
}
}

// TODO remove the transmute when rustc can use arrays of u8 as bitmasks
assert_eq!(
core::mem::size_of::<<LaneCount::<LANES> as SupportedLaneCount>::IntBitMask>(),
LaneCount::<LANES>::BITMASK_LEN,
);
let bitmask: <LaneCount<LANES> as SupportedLaneCount>::IntBitMask =
core::mem::transmute_copy(&bitmask);

Self::from_int_unchecked(intrinsics::simd_select_bitmask(
Self::from_int_unchecked(crate::intrinsics::simd_select_bitmask(
bitmask,
Self::splat(true).to_int(),
Self::splat(false).to_int(),
Expand Down