Module simd 

SIMD-optimized preset function library.

Hardware-accelerated batch operations (arithmetic, hashing) with runtime feature detection and scalar fallback. SIMD-optimized preset function library.

The slice-iterating wrappers in this module dispatch to the 4-lane kernels under crate::asm_presets. Every kernel emits explicit core::arch::asm! instructions for its target ISA — there is no reliance on the compiler’s auto-vectorizer (simd_review §1).

• arithmetic — Batch add, mul, scalar-add, pow, cmp-eq, FMA, coefficient-merge.
• hash — AES-NI based batch hashing for dedup.
• detect — Runtime CPU feature detection (SSE4.2, AVX2, NEON).

Re-exports

pub use arithmetic::BatchError;

pub use arithmetic::batch_add;

pub use arithmetic::batch_add_scalar;

pub use arithmetic::batch_cmp_eq;

pub use arithmetic::batch_coef_merge;

pub use arithmetic::batch_eval;

pub use arithmetic::batch_eval_parallel;

pub use arithmetic::batch_fma;

pub use arithmetic::batch_mul;

pub use arithmetic::batch_pow;

pub use detect::has_avx2;

pub use detect::has_neon;

pub use detect::has_sse42;

pub use hash::batch_hash;

pub use kernel::ScalarKernel;

pub use kernel::SimdKernel;

pub use kernel::global_kernel;

Modules

arithmetic

SIMD-accelerated batch arithmetic operations.

detect

Runtime CPU feature detection.

hash

Vectorized hashing for batch node deduplication.

kernel

Runtime-dispatched SIMD kernel abstraction.