Enum SimdBackend 

pub enum SimdBackend {
    Avx2,
    Avx512,
    Neon,
    Scalar,
}

SIMD backend selection (Jidoka auto-detection)

Variants

Avx2

AVX2: 256-bit vectors, 32 int8 ops/cycle

Avx512

AVX-512: 512-bit vectors, 64 int8 ops/cycle

Neon

ARM NEON: 128-bit vectors, 16 int8 ops/cycle

Scalar

Scalar fallback (Jidoka degradation)

Implementations

impl SimdBackend

pub fn detect() -> Self

Auto-detect best available SIMD backend (Jidoka)

pub fn from_x86_features(has_avx512: bool, has_avx2: bool) -> Self

Select backend from x86 feature flags (testable)

pub fn dot_i8(&self, a: &[i8], b: &[i8]) -> i32

Compute dot product of two i8 vectors

Returns i32 to prevent overflow (127^2 x 4096 < i32::MAX)

Safety rationale

Calls unsafe SIMD intrinsics guarded by is_x86_feature_detected!.

pub fn dot_f32_i8(&self, query: &[f32], doc: &[i8], scale: f32) -> f32

Compute dot product of f32 query with i8 document (rescoring)

Used in stage 2 for 99%+ accuracy retention.

Trait Implementations

impl Clone for SimdBackend

fn clone(&self) -> SimdBackend

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for SimdBackend

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for SimdBackend

fn eq(&self, other: &SimdBackend) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for SimdBackend

impl Eq for SimdBackend

impl StructuralPartialEq for SimdBackend