Skip to content

ARM NEON support #32

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 9 commits into from
Sep 4, 2019
Merged

ARM NEON support #32

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
bab0acc
Put something in the readme so we can have a PR
Licenser Jul 30, 2019
489477b
Add drone file
Licenser Jul 30, 2019
e2abd7a
update build status
Licenser Jul 30, 2019
5113c8f
unguard for sse4.2 to allow rust to polyfill on older platforms
Licenser Jul 30, 2019
50ba38a
Add more simd tests
Licenser Aug 16, 2019
9f7b8a9
RFC: Neon support (pretty much working) (#35)
sunnygleason Aug 16, 2019
ebf9ca9
Update extq intrinsics
Licenser Aug 16, 2019
4acbe17
Use simd-lite (#39)
Licenser Sep 3, 2019
3ee0aa9
add simd-lite real version
sunnygleason Sep 4, 2019
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
3 changes: 2 additions & 1 deletion src/lib.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,8 @@
platform_intrinsics,
stmt_expr_attributes,
simd_ffi,
link_llvm_intrinsics
link_llvm_intrinsics,
rustc_attrs
)
)]

Expand Down
219 changes: 206 additions & 13 deletions src/neon/intrinsics.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
//use std::arch::

use crate::neon::simd_llvm;

use crate::neon::simd_llvm::{self, *};
use std::hint::unreachable_unchecked;
use std::mem;
use core;

Expand Down Expand Up @@ -67,21 +67,214 @@ pub unsafe fn vnegq_s8(a: int8x16_t) -> int8x16_t {
}


/// Extract vector from pair of vectors
//uint8x16_t vextq_s8 (uint8x16_t a, uint8x16_t b, const int n)
#[inline]
fn rotate_(a: u128, b: u128, n: u128) -> u128 {
let az = a >> (n * 8);
let bz = b << (128 - (n * 8));
az | bz
}

#[inline]
pub unsafe fn vextq_u8(a: uint8x16_t, b: uint8x16_t, n: u8) -> uint8x16_t {
mem::transmute(rotate_(mem::transmute(a), mem::transmute(b), n as u128))
#[target_feature(enable = "neon")]
#[cfg_attr(target_arch = "arm", target_feature(enable = "v7"))]
#[rustc_args_required_const(2)]
pub unsafe fn vextq_u8(a: uint8x16_t, b: uint8x16_t, n: i32) -> uint8x16_t {
if n < 0 || n > 15 {
unreachable_unchecked();
};
match n & 0b1111 {
0 => simd_shuffle16(a, b, [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]),
1 => simd_shuffle16(
a,
b,
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
),
2 => simd_shuffle16(
a,
b,
[2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17],
),
3 => simd_shuffle16(
a,
b,
[3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18],
),
4 => simd_shuffle16(
a,
b,
[4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
),
5 => simd_shuffle16(
a,
b,
[5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20],
),
6 => simd_shuffle16(
a,
b,
[6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21],
),
7 => simd_shuffle16(
a,
b,
[7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22],
),
8 => simd_shuffle16(
a,
b,
[8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23],
),
9 => simd_shuffle16(
a,
b,
[
9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24,
],
),
10 => simd_shuffle16(
a,
b,
[
10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,
],
),
11 => simd_shuffle16(
a,
b,
[
11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26,
],
),
12 => simd_shuffle16(
a,
b,
[
12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27,
],
),
13 => simd_shuffle16(
a,
b,
[
13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28,
],
),
14 => simd_shuffle16(
a,
b,
[
14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29,
],
),
15 => simd_shuffle16(
a,
b,
[
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30,
],
),
_ => unreachable_unchecked(),
}
}

/// Extract vector from pair of vectors
//int8x16_t vextq_s8 (int8x16_t a, int8x16_t b, const int n)
#[inline]
pub unsafe fn vextq_s8(a: int8x16_t, b: int8x16_t, n: u8) -> int8x16_t {
mem::transmute(rotate_(mem::transmute(a), mem::transmute(b), n as u128))
#[target_feature(enable = "neon")]
#[cfg_attr(target_arch = "arm", target_feature(enable = "v7"))]
#[rustc_args_required_const(2)]
pub unsafe fn vextq_s8(a: int8x16_t, b: int8x16_t, n: i32) -> int8x16_t {
if n < 0 || n > 15 {
unreachable_unchecked();
};
match n & 0b1111 {
0 => simd_shuffle16(a, b, [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]),
1 => simd_shuffle16(
a,
b,
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
),
2 => simd_shuffle16(
a,
b,
[2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17],
),
3 => simd_shuffle16(
a,
b,
[3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18],
),
4 => simd_shuffle16(
a,
b,
[4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
),
5 => simd_shuffle16(
a,
b,
[5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20],
),
6 => simd_shuffle16(
a,
b,
[6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21],
),
7 => simd_shuffle16(
a,
b,
[7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22],
),
8 => simd_shuffle16(
a,
b,
[8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23],
),
9 => simd_shuffle16(
a,
b,
[
9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24,
],
),
10 => simd_shuffle16(
a,
b,
[
10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,
],
),
11 => simd_shuffle16(
a,
b,
[
11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26,
],
),
12 => simd_shuffle16(
a,
b,
[
12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27,
],
),
13 => simd_shuffle16(
a,
b,
[
13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28,
],
),
14 => simd_shuffle16(
a,
b,
[
14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29,
],
),
15 => simd_shuffle16(
a,
b,
[
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30,
],
),
_ => unreachable_unchecked(),
}
}

#[inline]
Expand Down
4 changes: 2 additions & 2 deletions src/neon/simd_llvm.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@ extern "platform-intrinsic" {
// pub fn simd_shuffle2<T, U>(x: T, y: T, idx: [u32; 2]) -> U;
// pub fn simd_shuffle4<T, U>(x: T, y: T, idx: [u32; 4]) -> U;
// pub fn simd_shuffle8<T, U>(x: T, y: T, idx: [u32; 8]) -> U;
// pub fn simd_shuffle16<T, U>(x: T, y: T, idx: [u32; 16]) -> U;
pub fn simd_shuffle16<T, U>(x: T, y: T, idx: [u32; 16]) -> U;
// pub fn simd_shuffle32<T, U>(x: T, y: T, idx: [u32; 32]) -> U;
// pub fn simd_shuffle64<T, U>(x: T, y: T, idx: [u32; 64]) -> U;
// pub fn simd_shuffle128<T, U>(x: T, y: T, idx: [u32; 128]) -> U;
Expand Down Expand Up @@ -51,4 +51,4 @@ extern "platform-intrinsic" {
//
// pub fn simd_fsqrt<T>(a: T) -> T;
// pub fn simd_fma<T>(a: T, b: T, c: T) -> T;
}
}