Expand description
SIMD-aligned memory allocation utilities.
Provides aligned vectors for optimal SIMD performance. Uses 64-byte alignment to support all SIMD instruction sets (NEON: 16B, AVX2: 32B, AVX-512: 64B).
Constants§
- SIMD_
ALIGN - Alignment for SIMD operations (64 bytes = AVX-512 width). Also works for NEON (16B) and AVX2 (32B) since 64 is a multiple of both.
Functions§
- alloc_
aligned - Allocate a SIMD-aligned vector with the given capacity.
- alloc_
aligned_ zeroed - Allocate a SIMD-aligned vector filled with zeros.
- from_
slice - Convert a slice to an aligned vector (copies data).
- to_vec
- Convert an aligned vector to a regular Vec.
Type Aliases§
- Aligned
Vec - Type alias for 64-byte aligned vector.