Struct BatchDispatchConfig 

pub struct BatchDispatchConfig {

    pub workgroup_size_x: u32,
    pub worker_groups: u32,
    pub hit_capacity: u32,
    pub timeout: Duration,
    pub graph_node_count: u32,
    pub graph_edge_count: u32,
    pub frontier_density_bps: u16,
    pub memory_pressure_bps: u16,
    pub resident_device_bytes: u64,
    pub device_memory_budget_bytes: u64,
    pub hot_opcode_count: u32,
    pub hot_window_count: u32,
    pub requeue_count: u64,
    pub max_priority_age: u32,
}

Immutable pipeline + launch geometry for batched megakernel scans.

Fields

workgroup_size_x: u32

Worker lanes per workgroup.

worker_groups: u32

Number of workgroups to launch for each batch.

hit_capacity: u32

Maximum sparse hits retained in the output ring.

timeout: Duration

Per-dispatch timeout budget.

graph_node_count: u32

Optional graph-node count hint for topology selection.

graph_edge_count: u32

Optional graph-edge count hint for topology selection.

frontier_density_bps: u16

Optional active-frontier density in basis points.

memory_pressure_bps: u16

Optional memory-pressure estimate in basis points.

resident_device_bytes: u64

Additional device-resident bytes already committed for this dispatch family.

The dispatcher adds its fixed queue-state resident footprint when building the shared launch-policy request.

device_memory_budget_bytes: u64

Hard device-memory budget for policy planning. Zero means unbounded.

hot_opcode_count: u32

Hot opcode count observed by the caller or runtime telemetry.

hot_window_count: u32

Hot window count observed by the caller or runtime telemetry.

requeue_count: u64

Requeued continuation count observed by the caller or runtime telemetry.

max_priority_age: u32

Maximum priority age observed by the caller or runtime telemetry.

Implementations

impl BatchDispatchConfig

pub const fn with_graph_hints( self, graph_node_count: u32, graph_edge_count: u32, frontier_density_bps: u16, memory_pressure_bps: u16, ) -> Self

Attach graph-topology hints used by the shared megakernel policy.

pub const fn with_device_memory_budget( self, resident_device_bytes: u64, device_memory_budget_bytes: u64, ) -> Self

Attach hard device-memory budget hints used by the shared launch policy.

pub const fn with_execution_hints( self, hot_opcode_count: u32, hot_window_count: u32, requeue_count: u64, max_priority_age: u32, ) -> Self

Attach execution hotness hints used by interpreter/JIT routing.

pub fn launch_recommendation( &self, limits: &Limits, queue_len: u32, ) -> Result<MegakernelLaunchRecommendation, PipelineError>

Return the shared launch-policy recommendation for this batch shape.

Errors

Returns PipelineError::Backend when adapter limits are malformed.

Trait Implementations

impl Clone for BatchDispatchConfig

fn clone(&self) -> BatchDispatchConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for BatchDispatchConfig

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

Formats the value using the given formatter.

impl Default for BatchDispatchConfig

fn default() -> Self

Returns the “default value” for a type.

impl Eq for BatchDispatchConfig

impl PartialEq for BatchDispatchConfig

fn eq(&self, other: &BatchDispatchConfig) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for BatchDispatchConfig