Inheritance diagram for Arcane::TBBTaskImplementation:

Collaboration diagram for Arcane::TBBTaskImplementation:

Classes
class	MDParallelForExecute
class	TaskInfoLockGuard
	Class for positioning TaskThreadInfo::taskIndex(). More...
class	Impl
class	ParallelForExecute

Public Member Functions
class	ARCCORE_ALIGNAS_PACKED (64) TaskThreadInfo
void	build ()
void	initialize (Int32 nb_thread) override
void	terminate () override
ITask *	createRootTask (ITaskFunctor *f) override
	Creates a root task. The implementation must copy the value of f, which is either a TaskFunctor or a TaskFunctorWithContext.
void	executeParallelFor (Int32 begin, Int32 size, const ParallelLoopOptions &options, IRangeFunctor *f) final
	Executes the functor f in parallel.
void	executeParallelFor (Int32 begin, Int32 size, Integer grain_size, IRangeFunctor *f) final
	Executes the functor f in parallel.
void	executeParallelFor (Int32 begin, Int32 size, IRangeFunctor *f) final
	Executes the functor f in parallel.
void	executeParallelFor (const ParallelFor1DLoopInfo &loop_info) override
	Executes the loop loop_info in parallel.
void	executeParallelFor (const ComplexForLoopRanges< 1 > &loop_ranges, const ForLoopRunInfo &run_info, IMDRangeFunctor< 1 > *functor) final
	Executes a 1D loop in parallel.
void	executeParallelFor (const ComplexForLoopRanges< 2 > &loop_ranges, const ForLoopRunInfo &run_info, IMDRangeFunctor< 2 > *functor) final
	Executes a 2D loop in parallel.
void	executeParallelFor (const ComplexForLoopRanges< 3 > &loop_ranges, const ForLoopRunInfo &run_info, IMDRangeFunctor< 3 > *functor) final
	Executes a 3D loop in parallel.
void	executeParallelFor (const ComplexForLoopRanges< 4 > &loop_ranges, const ForLoopRunInfo &run_info, IMDRangeFunctor< 4 > *functor) final
	Executes a 4D loop in parallel.
bool	isActive () const final
	Indicates if the implementation is active.
Int32	currentTaskThreadIndex () const final
	Implementation of TaskFactory::currentTaskThreadIndex().
Int32	currentTaskIndex () const final
	Implementation of TaskFactory::currentTaskIndex().
void	printInfos (std::ostream &o) const final
	Prints information about the runtime used.
TaskThreadInfo *	currentTaskThreadInfo () const
	Instance of TaskThreadInfo associated with the current thread.
Public Member Functions inherited from Arcane::ITaskImplementation
Int32	nbAllowedThread () const
	Maximum number of threads used to manage tasks.

Private Member Functions
template<int RankValue>
void	_executeMDParallelFor (const ComplexForLoopRanges< RankValue > &loop_ranges, IMDRangeFunctor< RankValue > *functor, const ForLoopRunInfo &run_info)
	Execution of an N-dimensional loop.
void	_executeParallelFor (const ParallelFor1DLoopInfo &loop_info)

Private Attributes
bool	m_is_active = false
Impl *	m_p = nullptr

Detailed Description

Definition at line 343 of file TBBTaskImplementation.cc.

Constructor & Destructor Documentation

◆ ~TBBTaskImplementation()

Arcane::TBBTaskImplementation::~TBBTaskImplementation ( )

override

Definition at line 1035 of file TBBTaskImplementation.cc.

Member Function Documentation

◆ _executeMDParallelFor()

template<int RankValue>

void Arcane::TBBTaskImplementation::_executeMDParallelFor	(	const ComplexForLoopRanges< RankValue > &	loop_ranges,
		IMDRangeFunctor< RankValue > *	functor,
		const ForLoopRunInfo &	run_info )

private

Execution of an N-dimensional loop.

Warning: The current implementation does not take into account options for loops other than one dimension.

Definition at line 1146 of file TBBTaskImplementation.cc.

References Arcane::TaskFactory::defaultParallelLoopOptions(), Arcane::ParallelLoopOptions::grainSize(), Arcane::makeLoopRanges(), Arcane::ParallelLoopOptions::maxThread(), Arcane::ParallelLoopOptions::mergeUnsetValues(), and Arcane::TaskFactory::verboseLevel().

Referenced by executeParallelFor(), executeParallelFor(), executeParallelFor(), and executeParallelFor().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ _executeParallelFor()

void Arcane::TBBTaskImplementation::_executeParallelFor ( const ParallelFor1DLoopInfo & loop_info )

private

Definition at line 1083 of file TBBTaskImplementation.cc.

◆ ARCCORE_ALIGNAS_PACKED()

class Arcane::TBBTaskImplementation::ARCCORE_ALIGNAS_PACKED ( 64 )

inline

Definition at line 331 of file TBBTaskImplementation.cc.

◆ build()

void Arcane::TBBTaskImplementation::build ( )

inline

Definition at line 412 of file TBBTaskImplementation.cc.

◆ createRootTask()

ITask * Arcane::TBBTaskImplementation::createRootTask ( ITaskFunctor * f )

inlineoverridevirtual

Creates a root task. The implementation must copy the value of f, which is either a TaskFunctor or a TaskFunctorWithContext.

Implements Arcane::ITaskImplementation.

Definition at line 416 of file TBBTaskImplementation.cc.

◆ currentTaskIndex()

Int32 Arcane::TBBTaskImplementation::currentTaskIndex ( ) const

finalvirtual

Implementation of TaskFactory::currentTaskIndex().

Implements Arcane::ITaskImplementation.

Definition at line 1243 of file TBBTaskImplementation.cc.

References currentTaskThreadIndex(), and currentTaskThreadInfo().

Here is the call graph for this function:

◆ currentTaskThreadIndex()

Int32 Arcane::TBBTaskImplementation::currentTaskThreadIndex ( ) const

inlinefinalvirtual

Implementation of TaskFactory::currentTaskThreadIndex().

Implements Arcane::ITaskImplementation.

Definition at line 460 of file TBBTaskImplementation.cc.

References Arcane::ITaskImplementation::nbAllowedThread().

Referenced by currentTaskIndex(), and currentTaskThreadInfo().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ currentTaskThreadInfo()

TBBTaskImplementation::TaskThreadInfo * Arcane::TBBTaskImplementation::currentTaskThreadInfo ( ) const

Instance of TaskThreadInfo associated with the current thread.

May be null if the current thread is not associated with a TBB thread or if it is outside the execution of a task or a parallel loop.

Definition at line 1231 of file TBBTaskImplementation.cc.

References currentTaskThreadIndex().

Referenced by currentTaskIndex().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ executeParallelFor() [1/8]

void Arcane::TBBTaskImplementation::executeParallelFor	(	const ComplexForLoopRanges< 1 > &	loop_ranges,
		const ForLoopRunInfo &	run_info,
		IMDRangeFunctor< 1 > *	functor )

inlinefinalvirtual

Executes a 1D loop in parallel.

Implements Arcane::ITaskImplementation.

Definition at line 430 of file TBBTaskImplementation.cc.

References _executeMDParallelFor().

Here is the call graph for this function:

◆ executeParallelFor() [2/8]

void Arcane::TBBTaskImplementation::executeParallelFor	(	const ComplexForLoopRanges< 2 > &	loop_ranges,
		const ForLoopRunInfo &	run_info,
		IMDRangeFunctor< 2 > *	functor )

inlinefinalvirtual

Executes a 2D loop in parallel.

Implements Arcane::ITaskImplementation.

Definition at line 436 of file TBBTaskImplementation.cc.

References _executeMDParallelFor().

Here is the call graph for this function:

◆ executeParallelFor() [3/8]

void Arcane::TBBTaskImplementation::executeParallelFor	(	const ComplexForLoopRanges< 3 > &	loop_ranges,
		const ForLoopRunInfo &	run_info,
		IMDRangeFunctor< 3 > *	functor )

inlinefinalvirtual

Executes a 3D loop in parallel.

Implements Arcane::ITaskImplementation.

Definition at line 442 of file TBBTaskImplementation.cc.

References _executeMDParallelFor().

Here is the call graph for this function:

◆ executeParallelFor() [4/8]

void Arcane::TBBTaskImplementation::executeParallelFor	(	const ComplexForLoopRanges< 4 > &	loop_ranges,
		const ForLoopRunInfo &	run_info,
		IMDRangeFunctor< 4 > *	functor )

inlinefinalvirtual

Executes a 4D loop in parallel.

Implements Arcane::ITaskImplementation.

Definition at line 448 of file TBBTaskImplementation.cc.

References _executeMDParallelFor().

Here is the call graph for this function:

◆ executeParallelFor() [5/8]

void Arcane::TBBTaskImplementation::executeParallelFor ( const ParallelFor1DLoopInfo & loop_info )

overridevirtual

Executes the loop loop_info in parallel.

Implements Arcane::ITaskImplementation.

Definition at line 1131 of file TBBTaskImplementation.cc.

◆ executeParallelFor() [6/8]

void Arcane::TBBTaskImplementation::executeParallelFor	(	Int32	begin,
		Int32	size,
		const ParallelLoopOptions &	options,
		IRangeFunctor *	f )

finalvirtual

Executes the functor f in parallel.

Implements Arcane::ITaskImplementation.

Definition at line 1222 of file TBBTaskImplementation.cc.

References executeParallelFor().

Referenced by executeParallelFor(), executeParallelFor(), and executeParallelFor().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ executeParallelFor() [7/8]

void Arcane::TBBTaskImplementation::executeParallelFor	(	Int32	begin,
		Int32	size,
		Integer	block_size,
		IRangeFunctor *	f )

finalvirtual

Executes the functor f in parallel.

Implements Arcane::ITaskImplementation.

Definition at line 1210 of file TBBTaskImplementation.cc.

References Arcane::TaskFactory::defaultParallelLoopOptions(), executeParallelFor(), and Arcane::ParallelLoopOptions::setGrainSize().

Here is the call graph for this function:

◆ executeParallelFor() [8/8]

void Arcane::TBBTaskImplementation::executeParallelFor	(	Int32	begin,
		Int32	size,
		IRangeFunctor *	f )

inlinefinalvirtual

Executes the functor f in parallel.

Implements Arcane::ITaskImplementation.

Definition at line 424 of file TBBTaskImplementation.cc.

References Arcane::TaskFactory::defaultParallelLoopOptions(), and executeParallelFor().

Here is the call graph for this function:

◆ initialize()

void Arcane::TBBTaskImplementation::initialize ( Int32 nb_thread )

overridevirtual

. Initializes the implementation with a maximum of nb_thread. If nb_thread is 0, the implementation can choose the number of threads automatically. This method is internal to Arcane and should only be called during the execution initialization.

Implements Arcane::ITaskImplementation.

Definition at line 1044 of file TBBTaskImplementation.cc.

References Arcane::TaskFactory::defaultParallelLoopOptions(), Arcane::ITaskImplementation::nbAllowedThread(), Arcane::TaskFactory::setDefaultParallelLoopOptions(), and Arcane::ParallelLoopOptions::setMaxThread().

Here is the call graph for this function:

◆ isActive()

bool Arcane::TBBTaskImplementation::isActive ( ) const

inlinefinalvirtual

Indicates if the implementation is active.

Implements Arcane::ITaskImplementation.

Definition at line 455 of file TBBTaskImplementation.cc.

◆ printInfos()

void Arcane::TBBTaskImplementation::printInfos ( std::ostream & o ) const

finalvirtual

Prints information about the runtime used.

Implements Arcane::ITaskImplementation.

Definition at line 1071 of file TBBTaskImplementation.cc.

◆ terminate()

void Arcane::TBBTaskImplementation::terminate ( )

overridevirtual

. Terminates the use of the implementation. This method must be called only at the end of the calculation.

Implements Arcane::ITaskImplementation.

Definition at line 1062 of file TBBTaskImplementation.cc.

Member Data Documentation

◆ m_is_active

bool Arcane::TBBTaskImplementation::m_is_active = false

private

Definition at line 481 of file TBBTaskImplementation.cc.

◆ m_p

Impl* Arcane::TBBTaskImplementation::m_p = nullptr

private

Definition at line 482 of file TBBTaskImplementation.cc.

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

TBBTaskImplementation.cc

Classes

Public Member Functions

Private Member Functions

Private Attributes

Detailed Description

Constructor & Destructor Documentation

◆ ~TBBTaskImplementation()

Member Function Documentation

◆ _executeMDParallelFor()

◆ _executeParallelFor()

◆ ARCCORE_ALIGNAS_PACKED()

◆ build()

◆ createRootTask()

◆ currentTaskIndex()

◆ currentTaskThreadIndex()

◆ currentTaskThreadInfo()

◆ executeParallelFor() [1/8]

◆ executeParallelFor() [2/8]

◆ executeParallelFor() [3/8]

◆ executeParallelFor() [4/8]

◆ executeParallelFor() [5/8]

◆ executeParallelFor() [6/8]

◆ executeParallelFor() [7/8]

◆ executeParallelFor() [8/8]

◆ initialize()

◆ isActive()

◆ printInfos()

◆ terminate()

Member Data Documentation

◆ m_is_active

◆ m_p