Struct ComputeOverlapScheduler 

pub struct ComputeOverlapScheduler {
    pub critical_count: usize,
    pub background_count: usize,
    /* private fields */
}

Schedules compute dispatches to maximise GPU utilisation through overlapping.

Classifies dispatches as either “latency-critical” (must finish before the next frame) or “background” (can run async on a secondary queue).

Fields

critical_count: usize

Number of latency-critical dispatches submitted this frame.

background_count: usize

Number of background dispatches submitted this frame.

Implementations

impl ComputeOverlapScheduler

pub fn new() -> Self

Create a new scheduler.

pub fn submit_critical(&mut self, batch: DispatchBatch)

Submit a latency-critical dispatch (→ main queue).

pub fn submit_background(&mut self, batch: DispatchBatch)

Submit a background dispatch (→ async compute queue).

pub fn end_frame(&mut self) -> usize

Execute all pending work and reset frame counters.

pub fn has_pending(&self) -> bool

Returns true if there is any pending work.

Trait Implementations

impl Debug for ComputeOverlapScheduler

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for ComputeOverlapScheduler

fn default() -> Self

Returns the “default value” for a type.