plb::ThreadAttribution Struct Reference

#include <threadAttribution.h>

Inheritance diagram for plb::ThreadAttribution:

[legend]

List of all members.

Public Member Functions
virtual	~ThreadAttribution ()
virtual bool	isLocal (plint blockId) const =0
	Specifies if the block "blockId" is located on the current MPI thread.
virtual bool	allBlocksAreLocal () const =0
virtual int	getMpiProcess (plint blockId) const =0
	Specifies on which MPI process a given block is located.
virtual int	getLocalThreadId (plint blockId) const =0
virtual ThreadAttribution *	merge (ThreadAttribution const &rhs, std::map< plint, std::vector< plint > > const &remappedIds) const =0
virtual ThreadAttribution *	extend (std::vector< plint > const &ids, std::vector< plint > const &mpiProcesses, std::vector< plint > const &localThreads) const =0
virtual bool	hasCoProcessors () const
virtual void	setCoProcessors (std::map< plint, int > processorHandles)
virtual int	getCoProcessorHandle (plint id) const
virtual ThreadAttribution *	clone () const =0

---

Constructor & Destructor Documentation

virtual plb::ThreadAttribution::~ThreadAttribution

(

)

[inline, virtual]

---

Member Function Documentation

virtual bool plb::ThreadAttribution::allBlocksAreLocal

(

)

const [pure virtual]

True means that all blocks are local, false means we don't know if all of them are local or not.

Implemented in plb::SerialThreadAttribution, plb::OneToOneThreadAttribution, and plb::ExplicitThreadAttribution.

virtual ThreadAttribution* plb::ThreadAttribution::clone

(

)

const [pure virtual]

Implemented in plb::SerialThreadAttribution, plb::OneToOneThreadAttribution, and plb::ExplicitThreadAttribution.

Referenced by plb::redistribute(), plb::PowerTwoMultiScaleManager::scaleMultiBlockManagement(), and plb::scaleMultiBlockManagement().

virtual ThreadAttribution* plb::ThreadAttribution::extend	(	std::vector< plint > const &	ids,
		std::vector< plint > const &	mpiProcesses,
		std::vector< plint > const &	localThreads
	)			const [pure virtual]

Extend the current attribution by a selection of new identifiers, with explicitly specified mpiProcess and thread attribution.

Implemented in plb::SerialThreadAttribution, plb::OneToOneThreadAttribution, and plb::ExplicitThreadAttribution.

virtual int plb::ThreadAttribution::getCoProcessorHandle

(

plint

id

)

const [inline, virtual]

Reimplemented in plb::ExplicitThreadAttribution.

Referenced by plb::MultiBlockLattice3D< T, Descriptor >::collideAndStream(), plb::transferFromCoProcessors(), and plb::transferToCoProcessors().

virtual int plb::ThreadAttribution::getLocalThreadId

(

plint

blockId

)

const [pure virtual]

Specifies to which of the local shared-memory threads the blockId belongs.

Implemented in plb::SerialThreadAttribution, plb::OneToOneThreadAttribution, and plb::ExplicitThreadAttribution.

Referenced by plb::ExplicitThreadAttribution::merge(), and plb::OneToOneThreadAttribution::merge().

virtual int plb::ThreadAttribution::getMpiProcess

(

plint

blockId

)

const [pure virtual]

Specifies on which MPI process a given block is located.

Implemented in plb::SerialThreadAttribution, plb::OneToOneThreadAttribution, and plb::ExplicitThreadAttribution.

Referenced by plb::isoSurfaceMarchingCube(), plb::ExplicitThreadAttribution::merge(), plb::OneToOneThreadAttribution::merge(), and plb::RandomRedistribute3D::redistribute().

virtual bool plb::ThreadAttribution::hasCoProcessors

(

)

const [inline, virtual]

Reimplemented in plb::ExplicitThreadAttribution.

Referenced by plb::MultiBlockLattice3D< T, Descriptor >::collideAndStream().

virtual bool plb::ThreadAttribution::isLocal

(

plint

blockId

)

const [pure virtual]

Specifies if the block "blockId" is located on the current MPI thread.

Implemented in plb::SerialThreadAttribution, plb::OneToOneThreadAttribution, and plb::ExplicitThreadAttribution.

Referenced by plb::computeSparseManagement(), plb::SparseBlockStructure3D::getLocalBlocks(), plb::SparseBlockStructure2D::getLocalBlocks(), plb::identifyBlocksWithPureDynamics(), and plb::isoSurfaceMarchingCube().

virtual ThreadAttribution* plb::ThreadAttribution::merge	(	ThreadAttribution const &	rhs,
		std::map< plint, std::vector< plint > > const &	remappedIds
	)			const [pure virtual]

Merge current attribution with rhs. From rhs, select explicitly the specified blocks, and remap them to fit into the new attribution, as specified by the parameter remappedIds.

Implemented in plb::SerialThreadAttribution, plb::OneToOneThreadAttribution, and plb::ExplicitThreadAttribution.

virtual void plb::ThreadAttribution::setCoProcessors

(

std::map< plint, int >

processorHandles

)

[inline, virtual]

Reimplemented in plb::ExplicitThreadAttribution.

Referenced by plb::MultiBlockManagement3D::setCoProcessors().

---

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

• threadAttribution.h