Crate haagenti_simd 

Haagenti SIMD

SIMD-accelerated primitives for compression algorithms.

Provides optimized implementations of common compression operations using platform-specific SIMD instructions.

Supported Architectures

• x86_64: SSE4.2, AVX2, AVX-512
• aarch64: NEON, SVE
• Fallback: Scalar implementation

Example

use haagenti_simd::{detect_simd, SimdLevel};

match detect_simd() {
    SimdLevel::Avx512 => println!("Using AVX-512"),
    SimdLevel::Avx2 => println!("Using AVX2"),
    SimdLevel::Neon => println!("Using NEON"),
    SimdLevel::None => println!("Scalar fallback"),
}

Enums

SimdLevel

SIMD feature level.

Functions

byte_histogram

Compute byte histogram (frequency of each byte value 0-255).

byte_histogram_simd

SIMD-accelerated histogram using scatter/gather approach.

copy_match

Copy a match from earlier in the output buffer.

copy_within_extend

Copy bytes from source to destination with potential overlap.

detect_simd

Detect available SIMD level at runtime.

fill_repeat

Fill buffer with a repeating pattern.

find_match_length

Find the length of matching bytes between two slices.

find_match_length_safe

Safe version that doesn’t use unsafe operations. Slightly slower but guaranteed safe.

has_avx2

Check if AVX2 is available.

has_avx512

Check if AVX-512 is available.

has_neon

Check if NEON is available.

hash4_scalar

Compute hash for a single 4-byte sequence (scalar).

hash4x4

Compute hashes for 4 consecutive positions using best available SIMD.

hash4x8

Compute hashes for 8 consecutive positions using AVX2.

simd_level

Get the current SIMD level (cached).