Struct AutoDispatcher 

pub struct AutoDispatcher { /* private fields */ }

Automatic dispatcher that chooses CPU (SIMD) or GPU based on batch size.

For small batches (< threshold), SIMD-optimized CPU operations are used. For large batches (>= threshold), GPU compute shaders are used.

Implementations

impl AutoDispatcher

pub async fn new() -> Self

Create a new auto dispatcher, attempting to initialize GPU.

pub async fn with_threshold(threshold: usize) -> Self

Create with a custom threshold.

pub fn cpu_only() -> Self

Create a CPU-only dispatcher (no GPU).

pub fn has_gpu(&self) -> bool

Check if GPU is available.

pub fn threshold(&self) -> usize

Get the current dispatch threshold.

pub async fn geometric_product( &self, a: &[GA3], b: &[GA3], ) -> Result<Vec<GA3>, GpuError>

Batch geometric product with automatic dispatch.

Uses GPU for large batches, SIMD-optimized CPU for small batches.

pub async fn addition(&self, a: &[GA3], b: &[GA3]) -> Result<Vec<GA3>, GpuError>

Batch addition with automatic dispatch.

Uses GPU for large batches, SIMD-optimized CPU for small batches.

pub async fn sandwich( &self, rotors: &[GA3], vectors: &[GA3], ) -> Result<Vec<GA3>, GpuError>

Batch sandwich product with automatic dispatch.

Uses GPU for large batches, SIMD-optimized CPU for small batches.

pub async fn exp(&self, a: &[GA3]) -> Result<Vec<GA3>, GpuError>

Batch exponential with automatic dispatch.

Uses GPU for large batches, SIMD-optimized CPU for small batches.

pub async fn rotor_slerp( &self, a: &[GA3], b: &[GA3], t: f32, ) -> Result<Vec<GA3>, GpuError>

Batch rotor SLERP with automatic dispatch.

Uses GPU for large batches, SIMD-optimized CPU for small batches.