Struct MegakernelLaunchRequest 

pub struct MegakernelLaunchRequest {

    pub queue_len: u32,
    pub requested_worker_groups: u32,
    pub max_workgroup_size_x: u32,
    pub max_compute_workgroups_per_dimension: u32,
    pub max_compute_invocations_per_workgroup: u32,
    pub requested_hit_capacity: u32,
    pub expected_hits_per_item: u32,
    pub hot_opcode_count: u32,
    pub hot_window_count: u32,
    pub requeue_count: u64,
    pub max_priority_age: u32,
    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,
}

Inputs for one launch-policy recommendation.

Fields

queue_len: u32

Logical ring slots or work items queued for this launch.

requested_worker_groups: u32

Caller-requested worker workgroup ceiling. Zero means derive from occupancy.

max_workgroup_size_x: u32

Adapter maximum workgroup size in the x dimension.

max_compute_workgroups_per_dimension: u32

Adapter maximum compute workgroups per dimension.

max_compute_invocations_per_workgroup: u32

Adapter maximum invocations per compute workgroup.

requested_hit_capacity: u32

Caller-requested sparse-hit capacity. Zero means derive from queue shape.

expected_hits_per_item: u32

Expected sparse hits per queued item when deriving hit capacity.

hot_opcode_count: u32

Count of opcodes observed hot enough for promotion.

hot_window_count: u32

Count of ticketed route windows observed hot enough for promotion.

requeue_count: u64

Slots requeued by priority scheduling since the last recommendation.

max_priority_age: u32

Maximum priority age observed since the last recommendation.

graph_node_count: u32

Nodes in the resident dependency graph. Zero means the caller has no graph-shape telemetry for this launch.

graph_edge_count: u32

Edges in the resident dependency graph. Zero means the caller has no graph-shape telemetry for this launch.

frontier_density_bps: u16

Active frontier density in basis points relative to graph nodes.

memory_pressure_bps: u16

Device-memory pressure in basis points relative to the active budget.

resident_device_bytes: u64

Device-resident bytes already required by this dispatch family.

device_memory_budget_bytes: u64

Hard device-memory budget for this launch. Zero means unbounded.

Implementations

impl MegakernelLaunchRequest

pub const fn direct( queue_len: u32, requested_worker_groups: u32, max_workgroup_size_x: u32, ) -> Self

Construct a direct-dispatch request with conservative defaults.

Trait Implementations

impl Clone for MegakernelLaunchRequest

fn clone(&self) -> MegakernelLaunchRequest

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for MegakernelLaunchRequest

impl Debug for MegakernelLaunchRequest

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

Formats the value using the given formatter.

impl Eq for MegakernelLaunchRequest

impl Hash for MegakernelLaunchRequest

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for MegakernelLaunchRequest

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for MegakernelLaunchRequest