Skip to content
New issue

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

#

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

Already on GitHub? # to your account

Jump to bottom

optimize simd widening mul #1247

Merged
merged 1 commit into from
Aug 13, 2022
Merged

optimize simd widening mul #1247

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
15 changes: 7 additions & 8 deletions src/distributions/utils.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ macro_rules! wmul_impl {
};

// simd bulk implementation
($(($ty:ident, $wide:ident),)+, $shift:expr) => {
($(($ty:ident, $wide:ty),)+, $shift:expr) => {
$(
impl WideningMultiply for $ty {
type Output = ($ty, $ty);
Expand Down Expand Up @@ -152,7 +152,8 @@ mod simd_wmul {
(u8x4, u16x4),
(u8x8, u16x8),
(u8x16, u16x16),
(u8x32, u16x32),,
(u8x32, u16x32),
(u8x64, Simd<u16, 64>),,
8
}

Expand All @@ -162,6 +163,8 @@ mod simd_wmul {
wmul_impl! { (u16x8, u32x8),, 16 }
#[cfg(not(target_feature = "avx2"))]
wmul_impl! { (u16x16, u32x16),, 16 }
#[cfg(not(target_feature = "avx512bw"))]
wmul_impl! { (u16x32, Simd<u32, 32>),, 16 }

// 16-bit lane widths allow use of the x86 `mulhi` instructions, which
// means `wmul` can be implemented with only two instructions.
Expand Down Expand Up @@ -191,15 +194,11 @@ mod simd_wmul {
wmul_impl! {
(u32x2, u64x2),
(u32x4, u64x4),
(u32x8, u64x8),,
(u32x8, u64x8),
(u32x16, Simd<u64, 16>),,
32
}

// TODO: optimize, this seems to seriously slow things down
wmul_impl_large! { (u8x64,) u8, 4 }
#[cfg(not(target_feature = "avx512bw"))]
wmul_impl_large! { (u16x32,) u16, 8 }
wmul_impl_large! { (u32x16,) u32, 16 }
wmul_impl_large! { (u64x2, u64x4, u64x8,) u64, 32 }
}

Expand Down