Struct SchedulerStat

pub struct SchedulerStat {
    work_id_name_map: HashMap<TypeId, &'static str>,
    work_counts: HashMap<TypeId, usize>,
    work_counters: HashMap<TypeId, Vec<Vec<Box<dyn WorkCounter>>>>,
}

Merge and print the work-packet level statistics from all worker threads

Fields

work_id_name_map: HashMap<TypeId, &'static str>

Map work packet type IDs to work packet names

work_counts: HashMap<TypeId, usize>

Count the number of work packets executed for different types

work_counters: HashMap<TypeId, Vec<Vec<Box<dyn WorkCounter>>>>

Collect work counters from work threads. Two dimensional vectors are used, e.g. [[foo_0, ..., foo_n], ..., [bar_0, ..., bar_n]]. The first dimension is for different types of work counters, (foo and bar in the above example). The second dimension if for work counters of the same type but from different threads (foo_0 and bar_0 are from the same thread). The order of insertion is determined by when SchedulerStat::merge is called for each WorkerLocalStat. We assume different threads have the same set of work counters (in the same order).

Implementations

impl SchedulerStat

fn work_name(&self, name: &str) -> String

Extract the work-packet name from the full type name. i.e. simplifies crate::scheduler::gc_work::SomeWorkPacket<Semispace> to SomeWorkPacket.

pub fn harness_stat(&self) -> HashMap<String, String>

Used during statistics printing at crate::memory_manager::harness_end

pub fn merge<C>(&mut self, stat: &WorkerLocalStat<C>)

Merge work counters from different worker threads

Trait Implementations

impl Default for SchedulerStat

fn default() -> SchedulerStat

Returns the “default value” for a type.