Execution Model

Launching tasks and kernels. More...

Topics
	Reduction Operations
	Types to use with reduce, reduceall, and reduction_accessor.
	Kernels
	Local concurrent operations.
	Legion Execution
	Potentially remote task execution.
	MPI Execution
	Direct task execution.

Classes
struct	agent< S >
	A node-local context for potentially parallel operations. More...
struct	bind_accessors< Proc >
	Handling for low-level special task parameters/arguments. More...
struct	blocks< S, T, B >
	Parallel operations with thread configuration. More...
struct	cpu
	Single-core execution space. More...
struct	executor< S, T, B >
	Parallel operations given a name for debugging or profiling. More...
struct	executor_base< D >
	Executor derivation. More...
struct	future< Return, Launch >
	Single or multiple future. More...
struct	future< Return >
	Single-valued future. More...
struct	future< Return, exec::launch_type_t::index >
	Multi-valued future from an index launch. More...
struct	gpu
	GPU execution space. More...
struct	trace::guard
	RAII guard for executing a trace. More...
struct	launch_domain
	An explicit launch domain size. More...
struct	multi< class >
	A sequence of accessors obtained from a launch::mapping. More...
struct	omp
	OpenMP execution space. More...
struct	partial< F, AA >
	A simple version of C++20's bind_front. More...
struct	prolog< Proc >
	Analyzes task arguments and updates data objects before launching a task. More...
struct	scheduler
	Launches tasks according to their execution-space (template) parameters. More...
struct	space< S >
	Execution space operations. More...
struct	space_base
	An execution space. More...
struct	task_local< T >
	A global variable with a task-specific value. More...
struct	task_prologue< Proc >
	Handling for low-level special task parameters/arguments. More...
struct	trace
	Records execution of a loop whose iterations all execute the same sequence of tasks. More...

Typedefs
using	accelerator
	The available accelerated execution space.

Enumerations
enum class	processor : size_t
	Enumeration of processor types. More...
enum	task_attributes_mask_t : TaskAttributes { }
	Task attribute flags. More...

Functions
template<class T , class... AA>
auto &	allocate (std::unique_ptr< topology< T > > &p, const typename T::coloring &c, AA &&... aa)
	Create a topology instance with specialization support.
template<auto & TASK, TaskAttributes ATTRIBUTES = flecsi::loc | flecsi::leaf, typename... ARGS>
auto	execute (ARGS &&... args)
	Execute a task.
template<class Return >
future< std::decay_t< Return > >	make_future (Return &&)
	Generate a new future from a value.
template<auto & F, class... AA>
constexpr exec::partial< F, std::decay_t< AA >... >	make_partial (AA &&... aa)
	Partially apply a function.
template<auto & Task, class Reduction , TaskAttributes Attributes = flecsi::loc | flecsi::leaf, typename... Args>
auto	reduce (Args &&...)
	Execute a reduction task.
template<class V , class... AA>
int	test (AA &&... aa)
	Execute a variant of a test task.
template<auto & TASK, TaskAttributes ATTRIBUTES = flecsi::loc | flecsi::leaf, typename... ARGS>
int	test (ARGS &&... args)
	Execute a test task.

Variables
constexpr auto	default_accelerator
	The task attribute to use for tasks that use the on-node parallelism interface.
constexpr on_t	on
	Placeholder argument that corresponds to an execution-space task parameter.

Detailed Description

Launching tasks and kernels.

Tasks are coarse-grained and use distributed-memory with restricted side effects; kernels are fine-grained and data-parallel, possibly using an accelerator.

Contents are in flecsi::exec unless otherwise specified.

#include "flecsi/execution.hh"

Typedef Documentation

accelerator

using accelerator

The available accelerated execution space.

Defined as gpu or omp if support for one of those is available, otherwise cpu.

Enumeration Type Documentation

task_attributes_mask_t

enum task_attributes_mask_t : TaskAttributes

Task attribute flags.

Appears directly in flecsi.

Enumerator
idempotent	Ignored. Deprecated No effect.
loc	Run on a Latency-Optimized Core (a CPU).
toc	Run on a Throughput-Optimized Core (a GPU). The task function itself still runs on the host, but a GPU is reserved for its use and field data is made available there. Warning MPI backend: Running one process per node likely leads to poor performance.
omp	Run as an OpenMP task. Note Legion backend: Can improve OpenMP task execution, since Legion knows to assign an entire node to such a task Warning MPI backend: Running one process per core likely leads to poor performance.
mpi	Run simultaneously on all processes with the obvious color mapping; allow MPI communication among point tasks, at the cost of significant startup overhead.

processor

enum class processor : size_t

strong

Enumeration of processor types.

Function Documentation

make_future()

template<class Return >

future< std::decay_t< Return > > make_future

(

Return &&

val

)

Generate a new future from a value.

Appears directly in flecsi.

reduce()

template<auto & Task, class Reduction , TaskAttributes Attributes = flecsi::loc | flecsi::leaf, typename... Args>

auto reduce ( Args && ... )

nodiscard

Execute a reduction task.

Template Parameters

Reduction

The reduction operation type.

Returns

a future providing the reduced return value

See also

execute about parameter and argument types.

Appears directly in flecsi.

execute()

template<auto & TASK, TaskAttributes ATTRIBUTES = flecsi::loc | flecsi::leaf, typename... ARGS>

auto execute

(

ARGS &&...

args

)

Execute a task.

Template Parameters

TASK	The user task. Its parameters must be copyable or a reference to a const, movable type. Any that is a pointer must be to a const type or to a function. If ATTRIBUTES specifies an MPI task, parameters need merely be movable.
ATTRIBUTES	The task attributes mask.
ARGS	The user-specified task arguments, implicitly converted to the parameter types for TASK. Certain FleCSI-defined parameter types accept particular, different argument types that serve as selectors for information stored by the backend; each type involved documents the correspondence. Additionally, a parameter may be a std::vector of such a type or a std::tuple that includes such a type; it accepts a std::vector of or a std::tuple including the corresponding argument type.

Returns

a future providing the value(s) returned from the task

Appears directly in flecsi.

make_partial()

template<auto & F, class... AA>

exec::partial< F, std::decay_t< AA >... > make_partial ( AA &&... aa )

constexpr

Partially apply a function.

Template Parameters

F	function to call
AA	leading arguments

Returns

a function object

Note

A task that accepts the result will usually be a function template:

void func(/*...*/);
template<class F>
void task(F f) {f(/* ... */);}
void client(scheduler &s) {
  auto p = make_partial<func>(/*...*/);
  s.execute<task<decltype(p)>>(p);  // note explicit template argument
}

Appears directly in flecsi.

Deprecated

Use a lambda or std::bind.

test() [1/2]

template<auto & TASK, TaskAttributes ATTRIBUTES = flecsi::loc | flecsi::leaf, typename... ARGS>

int test ( ARGS &&... args )

nodiscard

Execute a test task.

This interface is provided for FleCSI's unit testing framework. Test tasks must return an integer that is non-zero on failure, and zero otherwise.

See also

execute about parameter and argument types.

Returns

zero on success, non-zero on failure.

Appears directly in flecsi.

test() [2/2]

template<class V , class... AA>

int test ( AA &&... aa )

nodiscard

Execute a variant of a test task.

allocate()

template<class T , class... AA>

auto & allocate	(	std::unique_ptr< topology< T > > &	p,
		const typename T::coloring &	c,
		AA &&...	aa )

Create a topology instance with specialization support.

Calls the specialization's initialize on the topology instance.

Parameters

p	where to store the topology
c	coloring (perhaps from an mpi_coloring)
aa	further specialization-specific parameters

Returns

the new instance

Variable Documentation

on

on_t on

constexpr

Placeholder argument that corresponds to an execution-space task parameter.

default_accelerator

auto default_accelerator

constexpr

The task attribute to use for tasks that use the on-node parallelism interface.

Warning

Using toc causes field data to be placed on the device, so that it is accessible only via the parallelism interface.

Appears directly in flecsi.

Deprecated

Use accelerator.