Represents the various scheduling strategies for parallel for loops. Detailed explanations of each scheduling strategy are provided alongside each getter. If no schedule is specified, the default is Schedule.Static. More...

Inheritance diagram for DotMP.Schedule:

[legend]

Collaboration diagram for DotMP.Schedule:

[legend]

Public Member Functions
abstract void	LoopInit (int start, int end, uint num_threads, uint chunk_size)
	Abstract method for builtin schedulers to override for implementing IScheduler. More...

abstract void	LoopNext (int thread_id, out int start, out int end)
	Abstract method for builtin schedulers to override for implementing IScheduler. More...

Properties
static Schedule	Static `[get]`
	The static scheduling strategy. Iterations are divided amongst threads in round-robin fashion. Each thread gets a 'chunk' of iterations, determined by the chunk size. If no chunk size is specified, it's computed as total iterations divided by number of threads. More...

static Schedule	Dynamic `[get]`
	The dynamic scheduling strategy. Iterations are managed in a central queue. Threads fetch chunks of iterations from this queue when they have no assigned work. If no chunk size is defined, a basic heuristic is used to determine a chunk size. More...

static Schedule	Guided `[get]`
	The guided scheduling strategy. Similar to dynamic, but the chunk size starts larger and shrinks as iterations are consumed. The shrinking formula is based on the remaining iterations divided by the number of threads. The chunk size parameter sets a minimum chunk size. More...

static Schedule	Runtime `[get]`
	Runtime-defined scheduling strategy. Schedule is determined by the 'OMP_SCHEDULE' environment variable. Expected format: "schedule[,chunk_size]", e.g., "static,128", "guided", or "dynamic,3". More...

static Schedule	WorkStealing `[get]`
	The work-stealing scheduling strategy. Each thread gets its own local queue of iterations to execute. If a thread's queue is empty, it randomly selects another thread's queue as its "victim" and steals half of its remaining iterations. The chunk size parameter specifies how many iterations a thread should execute from its local queue at a time. More...

Static Private Attributes
static Schedulers.StaticScheduler	static_scheduler = new Schedulers.StaticScheduler()
	Internal holder for StaticScheduler object. More...

static Schedulers.DynamicScheduler	dynamic_scheduler = new Schedulers.DynamicScheduler()
	Internal holder for the DynamicScheduler object. More...

static Schedulers.GuidedScheduler	guided_scheduler = new Schedulers.GuidedScheduler()
	Internal holder for the GuidedScheduler object. More...

static Schedulers.RuntimeScheduler	runtime_scheduler = new Schedulers.RuntimeScheduler()
	Internal holder for the RuntimeScheduler object. More...

static Schedulers.WorkStealingScheduler	workstealing_scheduler = new Schedulers.WorkStealingScheduler()
	Internal holder for the WorkStealingScheduler object. More...

Detailed Description

Represents the various scheduling strategies for parallel for loops. Detailed explanations of each scheduling strategy are provided alongside each getter. If no schedule is specified, the default is Schedule.Static.

Member Function Documentation

◆ LoopInit()

abstract void DotMP.Schedule.LoopInit	(	int	start,
		int	end,
		uint	num_threads,
		uint	chunk_size
	)

pure virtual

Abstract method for builtin schedulers to override for implementing IScheduler.

Parameters

start	The start of the loop, inclusive.
end	The end of the loop, exclusive.
num_threads	The number of threads.
chunk_size	The chunk size.

Implements DotMP.IScheduler.

Implemented in DotMP.Schedulers.WorkStealingScheduler, DotMP.Schedulers.RuntimeScheduler, DotMP.Schedulers.GuidedScheduler, DotMP.Schedulers.DynamicScheduler, and DotMP.Schedulers.StaticScheduler.

◆ LoopNext()

abstract void DotMP.Schedule.LoopNext	(	int	thread_id,
		out int	start,
		out int	end
	)

pure virtual

Abstract method for builtin schedulers to override for implementing IScheduler.

Parameters

thread_id	The thread ID.
start	The start of the chunk, inclusive.
end	The end of the chunk, exclusive.

Implements DotMP.IScheduler.

Implemented in DotMP.Schedulers.WorkStealingScheduler, DotMP.Schedulers.RuntimeScheduler, DotMP.Schedulers.GuidedScheduler, DotMP.Schedulers.DynamicScheduler, and DotMP.Schedulers.StaticScheduler.

Member Data Documentation

◆ dynamic_scheduler

Schedulers.DynamicScheduler DotMP.Schedule.dynamic_scheduler = new Schedulers.DynamicScheduler()

staticprivate

Internal holder for the DynamicScheduler object.

◆ guided_scheduler

Schedulers.GuidedScheduler DotMP.Schedule.guided_scheduler = new Schedulers.GuidedScheduler()

staticprivate

Internal holder for the GuidedScheduler object.

◆ runtime_scheduler

Schedulers.RuntimeScheduler DotMP.Schedule.runtime_scheduler = new Schedulers.RuntimeScheduler()

staticprivate

Internal holder for the RuntimeScheduler object.

◆ static_scheduler

Schedulers.StaticScheduler DotMP.Schedule.static_scheduler = new Schedulers.StaticScheduler()

staticprivate

Internal holder for StaticScheduler object.

◆ workstealing_scheduler

Schedulers.WorkStealingScheduler DotMP.Schedule.workstealing_scheduler = new Schedulers.WorkStealingScheduler()

staticprivate

Internal holder for the WorkStealingScheduler object.

Property Documentation

◆ Dynamic

Schedule DotMP.Schedule.Dynamic

staticget

The dynamic scheduling strategy. Iterations are managed in a central queue. Threads fetch chunks of iterations from this queue when they have no assigned work. If no chunk size is defined, a basic heuristic is used to determine a chunk size.

Pros:

Better load balancing.

Cons:

Increased overhead.

◆ Guided

Schedule DotMP.Schedule.Guided

staticget

The guided scheduling strategy. Similar to dynamic, but the chunk size starts larger and shrinks as iterations are consumed. The shrinking formula is based on the remaining iterations divided by the number of threads. The chunk size parameter sets a minimum chunk size.

Pros:

Adaptable to workloads.

Cons:

Might not handle loops with early heavy load imbalance efficiently.

◆ Runtime

Schedule DotMP.Schedule.Runtime

staticget

Runtime-defined scheduling strategy. Schedule is determined by the 'OMP_SCHEDULE' environment variable. Expected format: "schedule[,chunk_size]", e.g., "static,128", "guided", or "dynamic,3".

◆ Static

Schedule DotMP.Schedule.Static

staticget

The static scheduling strategy. Iterations are divided amongst threads in round-robin fashion. Each thread gets a 'chunk' of iterations, determined by the chunk size. If no chunk size is specified, it's computed as total iterations divided by number of threads.

Pros:

Reduced overhead.

Cons:

Potential for load imbalance.

Note: This is the default strategy if none is chosen.

◆ WorkStealing

Schedule DotMP.Schedule.WorkStealing

staticget

The work-stealing scheduling strategy. Each thread gets its own local queue of iterations to execute. If a thread's queue is empty, it randomly selects another thread's queue as its "victim" and steals half of its remaining iterations. The chunk size parameter specifies how many iterations a thread should execute from its local queue at a time.

Pros:

Good approximation of optimal load balancing.
No contention over a shared queue.

Cons:

Stealing can be an expensive operation.

The documentation for this class was generated from the following file:

DotMP/Schedule.cs

Public Member Functions

Properties

Static Private Attributes

Detailed Description

Member Function Documentation

◆ LoopInit()

◆ LoopNext()

Member Data Documentation

◆ dynamic_scheduler

◆ guided_scheduler

◆ runtime_scheduler

◆ static_scheduler

◆ workstealing_scheduler

Property Documentation

◆ Dynamic

◆ Guided

◆ Runtime

◆ Static

◆ WorkStealing