Struct HybridDispatcher 

pub struct HybridDispatcher { /* private fields */ }

Dispatcher for routing workloads between CPU and GPU.

The dispatcher uses the configured ProcessingMode to decide where to execute each workload. In Adaptive mode, it learns the optimal threshold from runtime measurements.

Implementations

impl HybridDispatcher

pub fn new(config: HybridConfig) -> Self

Creates a new hybrid dispatcher.

pub fn with_defaults() -> Self

Creates a dispatcher with default configuration.

pub fn should_use_gpu(&self, workload_size: usize) -> bool

Returns whether GPU should be used for the given workload size.

pub fn execute<W: HybridWorkload>( &self, workload: &W, ) -> HybridResult<W::Result>

Executes a workload using the appropriate backend.

Returns the result and records execution statistics.

pub fn execute_measured<W: HybridWorkload>( &self, workload: &W, ) -> HybridResult<W::Result>

where W::Result: Clone,

Executes a workload and measures both backends for comparison.

In Adaptive mode, this updates the threshold based on measurements. Returns the result from the faster backend.

pub fn update_adaptive_threshold( &self, _workload_size: usize, cpu_time: Duration, gpu_time: Duration, )

Updates the adaptive threshold based on runtime measurements.

pub fn execute_cpu<W: HybridWorkload>(&self, workload: &W) -> W::Result

Forces execution on CPU regardless of mode.

pub fn execute_gpu<W: HybridWorkload>( &self, workload: &W, ) -> HybridResult<W::Result>

Forces execution on GPU regardless of mode.

pub fn config(&self) -> &HybridConfig

Returns the configuration.

pub fn stats(&self) -> &HybridStats

Returns the execution statistics.

pub fn stats_arc(&self) -> Arc<HybridStats>

Returns a shared reference to the statistics.

pub fn adaptive_threshold(&self) -> usize

Returns the current adaptive threshold.

pub fn set_adaptive_threshold(&self, threshold: usize)

Manually sets the adaptive threshold.