olb::HeterogeneousCopyTaskDataForGpuSource< T, DESCRIPTOR, TARGET > Class Template Reference

Private implementation of heterogeneous copy task between GPU_CUDA source and CPU_* target. More...

#include <communicator.hh>

Inheritance diagram for olb::HeterogeneousCopyTaskDataForGpuSource< T, DESCRIPTOR, TARGET >:

Collaboration diagram for olb::HeterogeneousCopyTaskDataForGpuSource< T, DESCRIPTOR, TARGET >:

Public Member Functions
	HeterogeneousCopyTaskDataForGpuSource (const std::vector< std::type_index > &fields, const std::vector< CellID > &targetCells, ConcreteBlockLattice< T, DESCRIPTOR, TARGET > &target, const std::vector< CellID > &sourceCells, ConcreteBlockLattice< T, DESCRIPTOR, Platform::GPU_CUDA > &source)
void	copy () override
void	wait () override
Public Member Functions inherited from olb::ConcreteHeterogeneousCopyTask
virtual	~ConcreteHeterogeneousCopyTask ()

Detailed Description

template<typename T, typename DESCRIPTOR, Platform TARGET>
class olb::HeterogeneousCopyTaskDataForGpuSource< T, DESCRIPTOR, TARGET >

Private implementation of heterogeneous copy task between GPU_CUDA source and CPU_* target.

Definition at line 642 of file communicator.hh.

Constructor & Destructor Documentation

HeterogeneousCopyTaskDataForGpuSource()

template<typename T , typename DESCRIPTOR , Platform TARGET>

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

inline

Definition at line 659 of file communicator.hh.

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

Member Function Documentation

copy()

template<typename T , typename DESCRIPTOR , Platform TARGET>

void olb::HeterogeneousCopyTaskDataForGpuSource< T, DESCRIPTOR, TARGET >::copy ( )

inlineoverridevirtual

Implements olb::ConcreteHeterogeneousCopyTask.

Definition at line 673 of file communicator.hh.

                       {
    if (_onlyPopulationField) {
      gpu::cuda::DeviceContext<T,DESCRIPTOR> lattice(_source);
      gpu::cuda::async_gather_field<descriptors::POPULATION>(_stream->get(), lattice, _sourceCells, _buffer.deviceData());
    } else {
      gpu::cuda::async_gather_any_fields(_stream->get(), _sourceFields, _sourceCells, _buffer.deviceData());
    }
 
    _buffer.setProcessingContext(ProcessingContext::Evaluation, *_stream);
  }

References olb::gpu::cuda::async_gather_any_fields(), olb::gpu::cuda::Column< T >::deviceData(), olb::Evaluation, and olb::gpu::cuda::Column< T >::setProcessingContext().

Here is the call graph for this function:

wait()

template<typename T , typename DESCRIPTOR , Platform TARGET>

void olb::HeterogeneousCopyTaskDataForGpuSource< T, DESCRIPTOR, TARGET >::wait ( )

inlineoverridevirtual

Implements olb::ConcreteHeterogeneousCopyTask.

Definition at line 684 of file communicator.hh.

                       {
    _stream->synchronize();
    _target.deserialize(_targetCells, _buffer.data());
  }

References olb::gpu::cuda::Column< T >::data(), and olb::MultiConcreteCommunicatable< COMMUNICATEE >::deserialize().

Here is the call graph for this function:

---

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

• src/core/platform/gpu/cuda/communicator.hh