Module wide_utils 

SIMD utilities and trait extensions for the wide crate

This module provides optimized platform-specific implementations for common SIMD operations that are not directly available in the wide crate, including:

• Widening operations (u32x8 → u64x8)
• Bitmask to vector conversion
• Shuffle/permute operations
• Vector splitting (high/low extraction)
• Cross-platform optimizations for x86_64 and aarch64

Examples

use simd_lookup::wide_utils::{WideUtilsExt, FromBitmask};
use wide::{u32x8, u64x8};

let input = u32x8::from([1, 2, 3, 4, 5, 6, 7, 8]);
let widened: u64x8 = input.widen_to_u64x8();

let mask = 0b10101010u8;
let mask_vec: u64x8 = u64x8::from_bitmask(mask);

Statics

SHUFFLE_COMPRESS_IDX_U8_HI

High byte shuffle indices (with +8 offset baked in)

SHUFFLE_COMPRESS_IDX_U8_LO

Shuffle indices for compressing u8x16 (low half).

SHUFFLE_COMPRESS_IDX_U32

Legacy alias for backwards compatibility

SHUFFLE_COMPRESS_IDX_U32X8

Shuffle indices for compressing u32x8 based on an 8-bit mask. Each entry contains the indices to shuffle selected elements to the front. Stored as arrays; use get_compress_indices_u32x8() for zero-cost SIMD access.

Traits

FromBitmask

Trait for creating SIMD vectors from bitmasks

SimdSplit

Trait for splitting SIMD vectors into high/low halves efficiently

WideUtilsExt

Trait extension for wide SIMD types providing additional utility operations

Functions

get_compress_indices_u32x8

Get compress indices as u32x8 (zero-cost transmute from array)