Struct dslab_compute::multicore::Compute

pub struct Compute { /* private fields */ }

Models computing resource with multiple cores which supports execution of parallel tasks.

In this model, the computation request can specify the minimum and maximum number of used cores, and provide a function which defines the dependence of parallel speedup on the number of used cores. Each core can only be used by one computation. The cores allocation for each computation is computed upon the request arrival and is not changed afterwards. This model also supports the manual allocation and release of cores and memory.

Implementations

impl Compute

pub fn new(speed: f64, cores: u32, memory: u64, ctx: SimulationContext) -> Self

Creates a new computing resource.

pub fn id(&self) -> Id

Returns id of corresponding simulation component.

pub fn speed(&self) -> f64

Returns the core speed.

pub fn cores_total(&self) -> u32

Returns the total number of cores.

pub fn cores_available(&self) -> u32

Returns the number of available cores.

pub fn memory_total(&self) -> u64

Returns the total amount of memory.

pub fn memory_available(&self) -> u64

Returns the amount of available memory.

pub fn min_compute_time( &self, flops: f64, min_cores: u32, max_cores: u32, cores_dependency: CoresDependency, ) -> Result<f64, &str>

Returns the minimum compute time for a workload with given flops, cores and cores dependency.

pub fn fraction_done(&self, comp_id: EventId) -> Result<f64, &str>

Returns workload fraction done for a given computation.

pub fn run( &mut self, flops: f64, memory: u64, min_cores: u32, max_cores: u32, cores_dependency: CoresDependency, requester: Id, ) -> u64

Starts computation with given parameters and returns computation id.

pub fn cancel_computation(&mut self, comp_id: u64)

Cancels computation.

pub fn preempt_computation(&mut self, comp_id: u64)

Preempts computation which is currently running.

pub fn resume_computation(&mut self, comp_id: u64)

Resumes computation which has been preempted.

pub fn allocate(&mut self, cores: u32, memory: u64, requester: Id) -> u64

Requests resource allocation with given parameters and returns allocation id.

pub fn deallocate(&mut self, cores: u32, memory: u64, requester: Id) -> u64

Requests resource deallocation with given parameters and returns deallocation id.

Trait Implementations

impl EventHandler for Compute

fn on(&mut self, event: Event)

Processes event.