Enum CpuTier 

pub enum CpuTier {
    Scalar,
    Avx2,
    Avx512,
    Avx512Vnni,
    Neon,
    NeonDotprod,
    NeonI8mm,
}

Ordered CPU capability tier. Higher is more capable.

Ord is derived from declaration order, so within a single architecture the comparison is meaningful (Scalar < Avx2 < Avx512, Scalar < Neon < NeonDotprod < NeonI8mm). Cross-architecture comparisons are nonsensical but harmless — only one architecture’s variants are ever produced at runtime.

Variants

Scalar

Portable scalar reference path. Always available.

Avx2

x86_64 AVX2 + FMA. Also the threshold for the x86 int8 GEMM/GEMV path: dpbusd is emulated with maddubs+madd here, so every tier from this one up runs the same int8 kernels for Q4_0/Q8_0/Q4_K/Q6_K.

Avx512

x86_64 AVX-512 (needs only avx512f). Produced when the default-on avx512 crate feature is enabled; disable it to cap the x86 tier at AVX2.

Its own contribution is now narrow: the 512-bit f32 vec_dot for Q8_0/Q4_0, plus the AVX-512 activation quantizer (which needs avx512vl on top). The production GEMM and GEMV at this tier are the CpuTier::Avx2 int8 kernels — see cpu::avx512_quantizer_available.

Avx512Vnni

x86_64 AVX-512 + VNNI (avx512vnni + avx512vl) — the x86 analogue of CpuTier::NeonI8mm. Runs the same int8 kernels as CpuTier::Avx2 with a native dpbusd instead of the emulation, wider tiles, and a 512-bit activation quantizer. This tier is a speed difference, not a capability one — the dtype coverage is identical.

Neon

aarch64 baseline NEON.

NeonDotprod

aarch64 NEON + dotprod (FEAT_DotProd, vdotq_s32).

NeonI8mm

aarch64 NEON + i8mm (FEAT_I8MM, vmmlaq_s32) — Q8_0 and Q4_0 GEMM; other ops use the dotprod path (i8mm implies dotprod).

Implementations

impl CpuTier

pub fn label(self) -> &'static str

Lowercase label used by CpuFeatures::report and parsed by the CERA_CPU_TIER override.

Trait Implementations

impl Clone for CpuTier

fn clone(&self) -> CpuTier

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for CpuTier

impl Debug for CpuTier

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

Formats the value using the given formatter.

impl Eq for CpuTier

impl Hash for CpuTier

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl Ord for CpuTier

fn cmp(&self, other: &CpuTier) -> Ordering

This method returns an Ordering between self and other.

fn max(self, other: Self) -> Self

where Self: Sized,

Compares and returns the maximum of two values.

fn min(self, other: Self) -> Self

where Self: Sized,

Compares and returns the minimum of two values.

fn clamp(self, min: Self, max: Self) -> Self

where Self: Sized,

Restrict a value to a certain interval.

fn clamp_to<R>(self, range: R) -> Self

where Self: Sized, R: ClampBounds<Self>,

🔬This is a nightly-only experimental API. (clamp_to)

Restrict a value to a certain range.

impl PartialEq for CpuTier

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

Equality operator ==.

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

Inequality operator !=.

impl PartialOrd for CpuTier

fn partial_cmp(&self, other: &CpuTier) -> Option<Ordering>

This method returns an ordering between self and other values if one exists.

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

Tests less than (for self and other) and is used by the < operator.

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

Tests less than or equal to (for self and other) and is used by the <= operator.

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

Tests greater than (for self and other) and is used by the > operator.

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

Tests greater than or equal to (for self and other) and is used by the >= operator.

impl StructuralPartialEq for CpuTier