Skip to content

fix: remove avx2 runtime detection #53

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
DaniPopes merged 6 commits into from
Feb 3, 2026
Merged

fix: remove avx2 runtime detection #53

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
452d95c
tmp: disable x86 path
DaniPopes Feb 3, 2026
0d89bd4
fix: extract to target_feature fn
DaniPopes Feb 3, 2026
f15dd2c
bench with znver3
DaniPopes Feb 3, 2026
2ce3528
Revert "fix: extract to target_feature fn"
DaniPopes Feb 3, 2026
2162b5e
no runtime detection
DaniPopes Feb 3, 2026
0e9c616
chore: drop rustflags
DaniPopes Feb 3, 2026
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
42 changes: 18 additions & 24 deletions src/nibbles.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1283,30 +1283,24 @@ fn longest_prefix_bit(a: &U256, b: &U256) -> usize {
#[inline]
fn longest_prefix<const EXACT: bool>(a: &U256, b: &U256) -> usize {
cfg_if! {
if #[cfg(target_arch = "x86_64")] {
#[cfg(feature = "std")]
let enabled = std::is_x86_feature_detected!("avx2");
#[cfg(not(feature = "std"))]
let enabled = cfg!(target_feature = "avx2");
if enabled {
return unsafe {
use core::arch::x86_64::*;
let x = _mm256_loadu_si256(a.as_limbs().as_ptr().cast());
let y = _mm256_loadu_si256(b.as_limbs().as_ptr().cast());
let diff = _mm256_cmpeq_epi8(x, y);
let mask = _mm256_movemask_epi8(diff);
let bytes = mask.leading_ones() as usize;
if !EXACT || bytes == 32 {
return bytes * 8;
}
let le_idx = 31 - bytes;
let a = *a.as_le_slice().get_unchecked(le_idx);
let b = *b.as_le_slice().get_unchecked(le_idx);
let diff = a ^ b;
let bits = diff.leading_zeros() as usize;
bytes * 8 + bits
};
}
if #[cfg(all(target_arch = "x86_64", target_feature = "avx2"))] {
return unsafe {
use core::arch::x86_64::*;
let x = _mm256_loadu_si256(a.as_limbs().as_ptr().cast());
let y = _mm256_loadu_si256(b.as_limbs().as_ptr().cast());
let diff = _mm256_cmpeq_epi8(x, y);
let mask = _mm256_movemask_epi8(diff);
let bytes = mask.leading_ones() as usize;
if !EXACT || bytes == 32 {
return bytes * 8;
}
let le_idx = 31 - bytes;
let a = *a.as_le_slice().get_unchecked(le_idx);
let b = *b.as_le_slice().get_unchecked(le_idx);
let diff = a ^ b;
let bits = diff.leading_zeros() as usize;
bytes * 8 + bits
};
}
}

Expand Down