*
* This class is very specific to the requirement of parallel code computations.
*/
- class ParaUMesh
+ class ParaUMesh : public RefCountObject
{
public:
- ParaUMesh(MEDCouplingUMesh *mesh, DataArrayIdType *globalCellIds, DataArrayIdType *globalNodeIds);
+ static ParaUMesh *New(MEDCouplingUMesh *mesh, DataArrayIdType *globalCellIds, DataArrayIdType *globalNodeIds);
MCAuto<DataArrayIdType> getCellIdsLyingOnNodes(const DataArrayIdType *globalNodeIds, bool fullyIn) const;
- MCAuto<ParaUMesh> redistributeCells(const DataArrayIdType *globalCellIds) const;
+ ParaUMesh *redistributeCells(const DataArrayIdType *globalCellIds) const;
+ DataArrayDouble *redistributeCellField(const DataArrayIdType *globalCellIds, const DataArrayDouble *fieldValueToRed) const;
+ DataArrayIdType *redistributeCellField(const DataArrayIdType *globalCellIds, const DataArrayIdType *fieldValueToRed) const;
+ DataArrayDouble *redistributeNodeField(const DataArrayIdType *globalCellIds, const DataArrayDouble *fieldValueToRed) const;
+ DataArrayIdType *redistributeNodeField(const DataArrayIdType *globalCellIds, const DataArrayIdType *fieldValueToRed) const;
+ MEDCouplingUMesh *getMesh() { return _mesh; }
+ DataArrayIdType *getGlobalCellIds() { return _cell_global; }
+ DataArrayIdType *getGlobalNodeIds() { return _node_global; }
+ protected:
virtual ~ParaUMesh() { }
+ ParaUMesh(MEDCouplingUMesh *mesh, DataArrayIdType *globalCellIds, DataArrayIdType *globalNodeIds);
+ std::string getClassName() const override { return "ParaUMesh"; }
+ std::size_t getHeapMemorySizeWithoutChildren() const override;
+ std::vector<const BigMemoryObject *> getDirectChildrenWithNull() const override;
private:
MCAuto<MEDCouplingUMesh> _mesh;
MCAuto<DataArrayIdType> _cell_global;
MCAuto<DataArrayIdType> _node_global;
+ private:
+ MCAuto<DataArrayIdType> getCellIdsLyingOnNodesFalse(const DataArrayIdType *globalNodeIds) const;
+ MCAuto<DataArrayIdType> getCellIdsLyingOnNodesTrue(const DataArrayIdType *globalNodeIds) const;
+ template<class T>
+ typename Traits<T>::ArrayType *redistributeCellFieldT(const DataArrayIdType *globalCellIds, const typename Traits<T>::ArrayType *fieldValueToRed) const
+ {
+ using DataArrayT = typename Traits<T>::ArrayType;
+ MPI_Comm comm(MPI_COMM_WORLD);
+ CommInterface ci;
+ if( _cell_global->getNumberOfTuples() != fieldValueToRed->getNumberOfTuples() )
+ throw INTERP_KERNEL::Exception("PAraUMesh::redistributeCellFieldT : invalid input length of array !");
+ std::unique_ptr<mcIdType[]> allGlobalCellIds,allGlobalCellIdsIndex;
+ int size(ci.allGatherArrays(comm,globalCellIds,allGlobalCellIds,allGlobalCellIdsIndex));
+ // Prepare ParaUMesh parts to be sent : compute for each proc the contribution of current rank.
+ std::vector< MCAuto<DataArrayIdType> > globalCellIdsToBeSent(size);
+ std::vector< MCAuto<DataArrayT> > fieldToBeSent(size);
+ for(int curRk = 0 ; curRk < size ; ++curRk)
+ {
+ mcIdType offset(allGlobalCellIdsIndex[curRk]);
+ MCAuto<DataArrayIdType> globalCellIdsOfCurProc(DataArrayIdType::New());
+ globalCellIdsOfCurProc->useArray(allGlobalCellIds.get()+offset,false,DeallocType::CPP_DEALLOC,allGlobalCellIdsIndex[curRk+1]-offset,1);
+ // the key call is here : compute for rank curRk the cells to be sent
+ MCAuto<DataArrayIdType> globalCellIdsCaptured(_cell_global->buildIntersection(globalCellIdsOfCurProc));// OK for the global cellIds
+ MCAuto<DataArrayIdType> localCellIdsCaptured(_cell_global->findIdForEach(globalCellIdsCaptured->begin(),globalCellIdsCaptured->end()));
+ globalCellIdsToBeSent[curRk] = globalCellIdsCaptured;
+ fieldToBeSent[curRk] = fieldValueToRed->selectByTupleIdSafe(localCellIdsCaptured->begin(),localCellIdsCaptured->end());
+ }
+ // Receive
+ std::vector< MCAuto<DataArrayIdType> > globalCellIdsReceived;
+ ci.allToAllArrays(comm,globalCellIdsToBeSent,globalCellIdsReceived);
+ std::vector< MCAuto<DataArrayT> > fieldValueReceived;
+ ci.allToAllArrays(comm,fieldToBeSent,fieldValueReceived);
+ // use globalCellIdsReceived to reorganize everything
+ MCAuto<DataArrayIdType> aggregatedCellIds( DataArrayIdType::Aggregate(FromVecAutoToVecOfConst<DataArrayIdType>(globalCellIdsReceived)) );
+ MCAuto<DataArrayIdType> aggregatedCellIdsSorted(aggregatedCellIds->copySorted());
+ MCAuto<DataArrayIdType> idsIntoAggregatedIds(DataArrayIdType::FindPermutationFromFirstToSecondDuplicate(aggregatedCellIdsSorted,aggregatedCellIds));
+ MCAuto<DataArrayIdType> cellIdsOfSameNodeIds(aggregatedCellIdsSorted->indexOfSameConsecutiveValueGroups());
+ MCAuto<DataArrayIdType> n2o_cells(idsIntoAggregatedIds->selectByTupleIdSafe(cellIdsOfSameNodeIds->begin(),cellIdsOfSameNodeIds->end()-1));//new == new ordering so that global cell ids are sorted . old == coarse ordering implied by the aggregation
+ //
+ MCAuto<DataArrayT> fieldAggregated(DataArrayT::Aggregate(FromVecAutoToVecOfConst<DataArrayT>(fieldValueReceived)));
+ MCAuto<DataArrayT> ret(fieldAggregated->selectByTupleIdSafe(n2o_cells->begin(),n2o_cells->end()));
+ return ret.retn();
+ }
+
+ template<class T>
+ typename Traits<T>::ArrayType *redistributeNodeFieldT(const DataArrayIdType *globalCellIds, const typename Traits<T>::ArrayType *fieldValueToRed) const
+ {
+ using DataArrayT = typename Traits<T>::ArrayType;
+ MPI_Comm comm(MPI_COMM_WORLD);
+ CommInterface ci;
+ if( _node_global->getNumberOfTuples() != fieldValueToRed->getNumberOfTuples() )
+ throw INTERP_KERNEL::Exception("PAraUMesh::redistributeNodeFieldT : invalid input length of array !");
+ std::unique_ptr<mcIdType[]> allGlobalCellIds,allGlobalCellIdsIndex;
+ int size(ci.allGatherArrays(comm,globalCellIds,allGlobalCellIds,allGlobalCellIdsIndex));
+ // Prepare ParaUMesh parts to be sent : compute for each proc the contribution of current rank.
+ std::vector< MCAuto<DataArrayIdType> > globalNodeIdsToBeSent(size);
+ std::vector< MCAuto<DataArrayT> > fieldToBeSent(size);
+ for(int curRk = 0 ; curRk < size ; ++curRk)
+ {
+ mcIdType offset(allGlobalCellIdsIndex[curRk]);
+ MCAuto<DataArrayIdType> globalCellIdsOfCurProc(DataArrayIdType::New());
+ globalCellIdsOfCurProc->useArray(allGlobalCellIds.get()+offset,false,DeallocType::CPP_DEALLOC,allGlobalCellIdsIndex[curRk+1]-offset,1);
+ // the key call is here : compute for rank curRk the cells to be sent
+ MCAuto<DataArrayIdType> globalCellIdsCaptured(_cell_global->buildIntersection(globalCellIdsOfCurProc));// OK for the global cellIds
+ MCAuto<DataArrayIdType> localCellIdsCaptured(_cell_global->findIdForEach(globalCellIdsCaptured->begin(),globalCellIdsCaptured->end()));
+ MCAuto<MEDCouplingUMesh> meshPart(_mesh->buildPartOfMySelf(localCellIdsCaptured->begin(),localCellIdsCaptured->end(),true));
+ MCAuto<DataArrayIdType> o2n(meshPart->zipCoordsTraducer());// OK for the mesh
+ MCAuto<DataArrayIdType> n2o(o2n->invertArrayO2N2N2O(meshPart->getNumberOfNodes()));
+ MCAuto<DataArrayIdType> globalNodeIdsPart(_node_global->selectByTupleIdSafe(n2o->begin(),n2o->end())); // OK for the global nodeIds
+ globalNodeIdsToBeSent[curRk] = globalNodeIdsPart;
+ fieldToBeSent[curRk] = fieldValueToRed->selectByTupleIdSafe(n2o->begin(),n2o->end());
+ }
+ // Receive
+ std::vector< MCAuto<DataArrayIdType> > globalNodeIdsReceived;
+ ci.allToAllArrays(comm,globalNodeIdsToBeSent,globalNodeIdsReceived);
+ std::vector< MCAuto<DataArrayT> > fieldValueReceived;
+ ci.allToAllArrays(comm,fieldToBeSent,fieldValueReceived);
+ // firstly deal with nodes.
+ MCAuto<DataArrayIdType> aggregatedNodeIds( DataArrayIdType::Aggregate(FromVecAutoToVecOfConst<DataArrayIdType>(globalNodeIdsReceived)) );
+ MCAuto<DataArrayIdType> aggregatedNodeIdsSorted(aggregatedNodeIds->copySorted());
+ MCAuto<DataArrayIdType> nodeIdsIntoAggregatedIds(DataArrayIdType::FindPermutationFromFirstToSecondDuplicate(aggregatedNodeIdsSorted,aggregatedNodeIds));
+ MCAuto<DataArrayIdType> idxOfSameNodeIds(aggregatedNodeIdsSorted->indexOfSameConsecutiveValueGroups());
+ MCAuto<DataArrayIdType> n2o_nodes(nodeIdsIntoAggregatedIds->selectByTupleIdSafe(idxOfSameNodeIds->begin(),idxOfSameNodeIds->end()-1));//new == new ordering so that global node ids are sorted . old == coarse ordering implied by the aggregation
+ //
+ MCAuto<DataArrayT> fieldAggregated(DataArrayT::Aggregate(FromVecAutoToVecOfConst<DataArrayT>(fieldValueReceived)));
+ MCAuto<DataArrayT> ret(fieldAggregated->selectByTupleIdSafe(n2o_nodes->begin(),n2o_nodes->end()));
+ //
+ return ret.retn();
+ }
};
}