Wrapper for a local plain-copy block communication request. More...

#include <communicator.hh>

Inheritance diagram for olb::ConcreteBlockCommunicator< ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > >::HomogeneousCopyTask:

Collaboration diagram for olb::ConcreteBlockCommunicator< ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > >::HomogeneousCopyTask:

Public Member Functions
	HomogeneousCopyTask (const std::vector< std::type_index > &fields, const std::vector< CellID > &targetCells, ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > &target, const std::vector< CellID > &sourceCells, ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > &source)

	~HomogeneousCopyTask ()

void	copy () override

void	wait () override

Public Member Functions inherited from olb::ConcreteBlockCommunicator< ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > >::CopyTask
virtual	~CopyTask ()

Detailed Description

template<typename T, typename DESCRIPTOR>
class olb::ConcreteBlockCommunicator< ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > >::HomogeneousCopyTask

Wrapper for a local plain-copy block communication request.

Definition at line 530 of file communicator.hh.

Constructor & Destructor Documentation

◆ HomogeneousCopyTask()

template<typename T , typename DESCRIPTOR >

olb::ConcreteBlockCommunicator< ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > >::HomogeneousCopyTask::HomogeneousCopyTask	(	const std::vector< std::type_index > &	fields,
		const std::vector< CellID > &	targetCells,
		ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > &	target,
		const std::vector< CellID > &	sourceCells,
		ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > &	source )

inline

Definition at line 547 of file communicator.hh.

                                                                                                      :
    _sourceFields(source.getDataRegistry().deviceFieldArrays(fields)),
    _targetFields(target.getDataRegistry().deviceFieldArrays(fields)),
    _onlyPopulationField(fields.size() == 1 && fields[0] == typeid(descriptors::POPULATION)),
    _targetCells(targetCells),
    _sourceCells(sourceCells),
    _target(target),
    _source(source),
    _stream(std::make_unique<gpu::cuda::device::Stream>(cudaStreamNonBlocking))
  {
    OLB_ASSERT(_sourceCells.size() == _targetCells.size(),
               "Source cell count must match target cell count");
  }

References OLB_ASSERT.

◆ ~HomogeneousCopyTask()

template<typename T , typename DESCRIPTOR >

olb::ConcreteBlockCommunicator< ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > >::HomogeneousCopyTask::~HomogeneousCopyTask ( )

inline

Definition at line 564 of file communicator.hh.

  {
    wait();
  }

References olb::ConcreteBlockCommunicator< BLOCK >::wait().

Here is the call graph for this function:

Member Function Documentation

◆ copy()

template<typename T , typename DESCRIPTOR >

void olb::ConcreteBlockCommunicator< ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > >::HomogeneousCopyTask::copy ( )

inlineoverridevirtual

Implements olb::ConcreteBlockCommunicator< ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > >::CopyTask.

Definition at line 569 of file communicator.hh.

  {
    if (_onlyPopulationField) {
      gpu::cuda::DeviceContext<T,DESCRIPTOR> sourceLattice(_source);
      gpu::cuda::DeviceContext<T,DESCRIPTOR> targetLattice(_target);
      gpu::cuda::async_copy_field<descriptors::POPULATION>(_stream->get(), sourceLattice, targetLattice, _sourceCells, _targetCells);
    } else {
      gpu::cuda::async_copy_any_fields(_stream->get(), _sourceFields, _targetFields, _sourceCells, _targetCells);
    }
  }