Crate rustalign_simd 

RustAlign SIMD - SIMD-accelerated alignment operations

This crate provides SIMD-accelerated implementations of the Smith-Waterman alignment algorithm, matching the C++ SSE/AVX2 code.

Platform Support

• x86_64: SSE2, SSE4.2, AVX2 via core::arch::x86_64
• AArch64: NEON via core::arch::aarch64
• Other: Scalar fallback implementations

Runtime Dispatch

The crate supports runtime CPU feature detection and dispatch, similar to the C++ __builtin_cpu_supports approach.

Structs

CpuFeatures

Detected CPU features for SIMD dispatch

DpMatrix

SIMD-accelerated DP matrix

MatrixQuartet

Helper structure for working with matrix quartets

SseReg

SIMD register type wrapper

Constants

AVX_REG_WIDTH

AVX register width in bytes

I16_PER_AVX

Number of i16 elements per AVX register

I16_PER_SSE

Number of i16 elements per SSE register

SSE_REG_WIDTH

SIMD register width in bytes (16 for SSE, 32 for AVX2)

U8_PER_AVX

Number of u8 elements per AVX register

U8_PER_SSE

Number of u8 elements per SSE register

Functions

aligned_vec

Create a new SIMD-aligned vector

has_avx2

Check if AVX2 is available (x86_64)

has_neon

Check if NEON is available (always false on x86_64)

has_popcnt

Check if POPCNT is available

has_sse2

Convenience functions for individual feature detection

has_sse4_2

Check if SSE4.2 is available

Type Aliases

AlignedVec

SIMD-aligned vector for SSE operations