madness::MacroTaskOperationBase Class Reference

#include <macrotaskq.h>

Inheritance diagram for madness::MacroTaskOperationBase:

Collaboration diagram for madness::MacroTaskOperationBase:

Public Member Functions
	MacroTaskOperationBase ()
virtual	~MacroTaskOperationBase ()
virtual bool	accumulates_own_output () const
virtual void	cleanup ()
	release whatever the task kept across its batches, called once the queue has run
virtual bool	handles_own_data_movement () const
	true if the task moves its operand coefficients into the subworld itself
virtual long	owner_hint (const Batch &, const long) const
	which subworld should run this batch; -1 leaves the choice to the queue

Public Attributes
Batch	batch
Cloud *	cloud_ptr =0
std::string	name ="unknown_task"
std::shared_ptr< MacroTaskPartitioner >	partitioner =0
World *	subworld_ptr =0

Constructor & Destructor Documentation

MacroTaskOperationBase()

madness::MacroTaskOperationBase::MacroTaskOperationBase ( )

inline

~MacroTaskOperationBase()

virtual madness::MacroTaskOperationBase::~MacroTaskOperationBase ( )

inlinevirtual

Member Function Documentation

accumulates_own_output()

virtual bool madness::MacroTaskOperationBase::accumulates_own_output ( ) const

inlinevirtual

true if the task accumulates its own results and drains them in finalize_stage2, instead of the queue moving every batch result into the universe result itself

Reimplemented in madness::Exchange< T, NDIM >::ExchangeImpl< T, NDIM >::MacroTaskExchangeSimple.

cleanup()

virtual void madness::MacroTaskOperationBase::cleanup ( )

inlinevirtual

release whatever the task kept across its batches, called once the queue has run

A task class that caches operands between its batches has to keep them in static state, since each batch is a separate task object. This is where that state must be dropped: it still refers to the subworld the batches were built in, and that subworld is destroyed after the queue finishes. Clearing it later means destroying function implementations whose world is already gone.

Reimplemented in madness::Exchange< T, NDIM >::ExchangeImpl< T, NDIM >::MacroTaskExchangeSimple.

handles_own_data_movement()

virtual bool madness::MacroTaskOperationBase::handles_own_data_movement ( ) const

inlinevirtual

true if the task moves its operand coefficients into the subworld itself

Reimplemented in madness::Exchange< T, NDIM >::ExchangeImpl< T, NDIM >::MacroTaskExchangeSimple.

owner_hint()

virtual long madness::MacroTaskOperationBase::owner_hint ( const Batch &  , const long    ) const

inlinevirtual

which subworld should run this batch; -1 leaves the choice to the queue

Reads whatever assignment prepare_owner_assignment computed, so the two go together: one decides the mapping for the whole partition, this one applies it per batch.

Reimplemented in OwnedTask, and madness::Exchange< T, NDIM >::ExchangeImpl< T, NDIM >::MacroTaskExchangeSimple.

Member Data Documentation

batch

Batch madness::MacroTaskOperationBase::batch

Referenced by madness::MacroTaskMp2ConstantPart::ConstantPartPartitioner::do_partitioning(), madness::MacroTaskConstantPart::ConstantPartPartitioner::do_partitioning(), madness::MacroTaskMp2UpdatePair::UpdatePairPartitioner::do_partitioning(), madness::MacroTaskIteratePair::IteratePairPartitioner::do_partitioning(), madness::MacroTaskMp2UpdatePair::operator()(), madness::MP3::MacroTaskMP3::operator()(), VectorOfScalarTask::operator()(), and OwnedTask::owner_hint().

cloud_ptr

Cloud* madness::MacroTaskOperationBase::cloud_ptr =0

set by MacroTaskInternal::run before the task body runs, so a task that moves its own operands (see handles_own_data_movement) can fetch them from the cloud itself

name

std::string madness::MacroTaskOperationBase::name ="unknown_task"

Referenced by madness::MacroTaskComputeCorrelationEnergy::MacroTaskComputeCorrelationEnergy(), madness::MacroTaskConstantPart::MacroTaskConstantPart(), madness::MacroTaskIteratePair::MacroTaskIteratePair(), madness::MacroTaskMp2UpdatePair::MacroTaskMp2UpdatePair(), madness::MP3::MacroTaskMP3::MacroTaskMP3(), madness::MacroTaskSinglesPotentialEx::MacroTaskSinglesPotentialEx(), madness::MacroTaskSinglesPotentialGs::MacroTaskSinglesPotentialGs(), madness::MacroTaskSinglesPotentialEx::allocator(), madness::MacroTaskSinglesPotentialGs::allocator(), madness::MacroTaskSinglesPotentialEx::operator()(), and madness::MacroTaskSinglesPotentialGs::operator()().

partitioner

std::shared_ptr<MacroTaskPartitioner> madness::MacroTaskOperationBase::partitioner =0

Referenced by madness::MacroTaskConstantPart::MacroTaskConstantPart(), madness::Coulomb< T, NDIM >::MacroTaskCoulomb::MacroTaskCoulomb(), madness::MacroTaskIteratePair::MacroTaskIteratePair(), madness::MacroTaskMp2ConstantPart::MacroTaskMp2ConstantPart(), madness::MacroTaskMp2UpdatePair::MacroTaskMp2UpdatePair(), madness::MP3::MacroTaskMP3::MacroTaskMP3(), madness::MacroTaskSinglesPotentialEx::MacroTaskSinglesPotentialEx(), madness::MP3::MacroTaskMP3::operator()(), and test_2d_partitioning().

subworld_ptr

World* madness::MacroTaskOperationBase::subworld_ptr =0

---

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

• macrotaskq.h