Crate lav 

Lane-Associated Vector (LAV): Portable SIMD vector trait as GAT of SIMD lane trait.

NOTE: This crate requires nightly Rust.

This example uses SIMD generically over floating-point types while hiding it from the user.

Features

• SIMD lane trait Real abstracting over f32 and f64.
• SIMD vector trait SimdReal<Real, N> abstracting over Simd<f32, N> and Simd<f64, N>.
• Generic associated type (GAT) Real::Simd<N> as part of SIMD lane trait Real implementing SIMD vector trait SimdReal<Self, N> for itself as lane type where the GAT is generic over the number of SIMD vector lanes N.
• AoS/SoA/AoSoA via Real::as_simd/Real::as_simd_mut abstracting over as_simd/as_simd_mut of f32 and f64 slices.
• Lanewise approximate equality test wrt to epsilon and ULP SIMD vectors.
• ApproxEq trait complementing PartialEq.

Optional Features

Following features are disabled by default unless their feature gate is enabled:

• target-features: Provides native number of SIMD vector lanes Real::NATIVE_LANE_COUNT for the current build target.
• libm: Enables no_std without loss of functionality.

Modules

example

Portably SIMD-optimized 3D rotator implementation generic over lane type f32 and f64.

Macros

swizzle

Constructs vector by copying lanes from selected lanes of one or two input vectors.

Traits

ApproxEq

Tests for approximate equality.

Bits

Bits representation of Real number with associated SimdBits vector.

Real

Real number of f32 or f64 with associated Bits representation and SimdReal vector.

Select

Selects lanes from two vectors by mask vector.

SimdBits

Bits representation vector of SimdReal vector with associated SimdMask vector.

SimdMask

Mask vector of Mask<i32, N> or Mask<i64, N>.

SimdReal

Real number vector of Simd<f32, N> or Simd<f64, N> with associated SimdBits and SimdMask vector.