f97f6326f0
When snappyHexMesh is run in parallel it re-balances the mesh during refinement
and layer addition by redistribution which requires a decomposition method
that operates in parallel, e.g. hierachical or ptscotch. decomposePar uses a
decomposition method which operates in serial e.g. hierachical but NOT
ptscotch. In order to run decomposePar followed by snappyHexMesh in parallel it
has been necessary to change the method specified in decomposeParDict but now
this is avoided by separately specifying the decomposition and distribution
methods, e.g. in the incompressible/simpleFoam/motorBike case:
numberOfSubdomains 6;
decomposer hierarchical;
distributor ptscotch;
hierarchicalCoeffs
{
n (3 2 1);
order xyz;
}
The distributor entry is also used for run-time mesh redistribution, e.g. in the
multiphase/interFoam/RAS/floatingObject case re-distribution for load-balancing
is enabled in constant/dynamicMeshDict:
distributor
{
type distributor;
libs ("libfvMeshDistributors.so");
redistributionInterval 10;
}
which uses the distributor specified in system/decomposeParDict:
distributor hierarchical;
This rationalisation provides the structure for development of mesh
redistribution and load-balancing.
483 lines
15 KiB
C++
483 lines
15 KiB
C++
/*---------------------------------------------------------------------------*\
|
|
========= |
|
|
\\ / F ield | OpenFOAM: The Open Source CFD Toolbox
|
|
\\ / O peration | Website: https://openfoam.org
|
|
\\ / A nd | Copyright (C) 2011-2021 OpenFOAM Foundation
|
|
\\/ M anipulation |
|
|
-------------------------------------------------------------------------------
|
|
License
|
|
This file is part of OpenFOAM.
|
|
|
|
OpenFOAM is free software: you can redistribute it and/or modify it
|
|
under the terms of the GNU General Public License as published by
|
|
the Free Software Foundation, either version 3 of the License, or
|
|
(at your option) any later version.
|
|
|
|
OpenFOAM is distributed in the hope that it will be useful, but WITHOUT
|
|
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
|
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
|
|
for more details.
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
along with OpenFOAM. If not, see <http://www.gnu.org/licenses/>.
|
|
|
|
InClass
|
|
domainDecomposition
|
|
|
|
Description
|
|
Private member of domainDecomposition.
|
|
Decomposes the mesh into bits
|
|
|
|
\*---------------------------------------------------------------------------*/
|
|
|
|
#include "domainDecomposition.H"
|
|
|
|
// * * * * * * * * * * * * * Private Member Functions * * * * * * * * * * * //
|
|
|
|
void Foam::domainDecomposition::append(labelList& lst, const label elem)
|
|
{
|
|
label sz = lst.size();
|
|
lst.setSize(sz+1);
|
|
lst[sz] = elem;
|
|
}
|
|
|
|
|
|
void Foam::domainDecomposition::addInterProcFace
|
|
(
|
|
const label facei,
|
|
const label ownerProc,
|
|
const label nbrProc,
|
|
|
|
List<Map<label>>& nbrToInterPatch,
|
|
List<DynamicList<DynamicList<label>>>& interPatchFaces
|
|
) const
|
|
{
|
|
Map<label>::iterator patchiter = nbrToInterPatch[ownerProc].find(nbrProc);
|
|
|
|
// Introduce turning index only for internal faces (are duplicated).
|
|
label ownerIndex = facei+1;
|
|
label nbrIndex = -(facei+1);
|
|
|
|
if (patchiter != nbrToInterPatch[ownerProc].end())
|
|
{
|
|
// Existing interproc patch. Add to both sides.
|
|
label toNbrProcPatchi = patchiter();
|
|
interPatchFaces[ownerProc][toNbrProcPatchi].append(ownerIndex);
|
|
|
|
if (isInternalFace(facei))
|
|
{
|
|
label toOwnerProcPatchi = nbrToInterPatch[nbrProc][ownerProc];
|
|
interPatchFaces[nbrProc][toOwnerProcPatchi].append(nbrIndex);
|
|
}
|
|
}
|
|
else
|
|
{
|
|
// Create new interproc patches.
|
|
label toNbrProcPatchi = nbrToInterPatch[ownerProc].size();
|
|
nbrToInterPatch[ownerProc].insert(nbrProc, toNbrProcPatchi);
|
|
DynamicList<label> oneFace;
|
|
oneFace.append(ownerIndex);
|
|
interPatchFaces[ownerProc].append(oneFace);
|
|
|
|
if (isInternalFace(facei))
|
|
{
|
|
label toOwnerProcPatchi = nbrToInterPatch[nbrProc].size();
|
|
nbrToInterPatch[nbrProc].insert(ownerProc, toOwnerProcPatchi);
|
|
oneFace.clear();
|
|
oneFace.append(nbrIndex);
|
|
interPatchFaces[nbrProc].append(oneFace);
|
|
}
|
|
}
|
|
}
|
|
|
|
|
|
void Foam::domainDecomposition::decomposeMesh()
|
|
{
|
|
// Decide which cell goes to which processor
|
|
distributeCells();
|
|
|
|
// Distribute the cells according to the given processor label
|
|
|
|
// calculate the addressing information for the original mesh
|
|
Info<< "\nCalculating original mesh data" << endl;
|
|
|
|
// set references to the original mesh
|
|
const polyBoundaryMesh& patches = boundaryMesh();
|
|
const faceList& fcs = faces();
|
|
const labelList& owner = faceOwner();
|
|
const labelList& neighbour = faceNeighbour();
|
|
|
|
// loop through the list of processor labels for the cell and add the
|
|
// cell shape to the list of cells for the appropriate processor
|
|
|
|
Info<< "\nDistributing cells to processors" << endl;
|
|
|
|
// Cells per processor
|
|
procCellAddressing_ = invertOneToMany(nProcs_, cellToProc_);
|
|
|
|
Info<< "\nDistributing faces to processors" << endl;
|
|
|
|
// Loop through all internal faces and decide which processor they belong to
|
|
// First visit all internal faces. If cells at both sides belong to the
|
|
// same processor, the face is an internal face. If they are different,
|
|
// it belongs to both processors.
|
|
|
|
procFaceAddressing_.setSize(nProcs_);
|
|
|
|
// Internal faces
|
|
forAll(neighbour, facei)
|
|
{
|
|
if (cellToProc_[owner[facei]] == cellToProc_[neighbour[facei]])
|
|
{
|
|
// Face internal to processor. Notice no turning index.
|
|
procFaceAddressing_[cellToProc_[owner[facei]]].append(facei+1);
|
|
}
|
|
}
|
|
|
|
// for all processors, set the size of start index and patch size
|
|
// lists to the number of patches in the mesh
|
|
forAll(procPatchSize_, proci)
|
|
{
|
|
procPatchSize_[proci].setSize(patches.size());
|
|
procPatchStartIndex_[proci].setSize(patches.size());
|
|
}
|
|
|
|
forAll(patches, patchi)
|
|
{
|
|
// Reset size and start index for all processors
|
|
forAll(procPatchSize_, proci)
|
|
{
|
|
procPatchSize_[proci][patchi] = 0;
|
|
procPatchStartIndex_[proci][patchi] =
|
|
procFaceAddressing_[proci].size();
|
|
}
|
|
|
|
const label patchStart = patches[patchi].start();
|
|
|
|
if (!isA<cyclicPolyPatch>(patches[patchi]))
|
|
{
|
|
// Normal patch. Add faces to processor where the cell
|
|
// next to the face lives
|
|
|
|
const labelUList& patchFaceCells =
|
|
patches[patchi].faceCells();
|
|
|
|
forAll(patchFaceCells, facei)
|
|
{
|
|
const label curProc = cellToProc_[patchFaceCells[facei]];
|
|
|
|
// add the face without turning index
|
|
procFaceAddressing_[curProc].append(patchStart+facei+1);
|
|
|
|
// increment the number of faces for this patch
|
|
procPatchSize_[curProc][patchi]++;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
const cyclicPolyPatch& pp = refCast<const cyclicPolyPatch>
|
|
(
|
|
patches[patchi]
|
|
);
|
|
// cyclic: check opposite side on this processor
|
|
const labelUList& patchFaceCells = pp.faceCells();
|
|
|
|
const labelUList& nbrPatchFaceCells =
|
|
pp.nbrPatch().faceCells();
|
|
|
|
forAll(patchFaceCells, facei)
|
|
{
|
|
const label curProc = cellToProc_[patchFaceCells[facei]];
|
|
const label nbrProc = cellToProc_[nbrPatchFaceCells[facei]];
|
|
if (curProc == nbrProc)
|
|
{
|
|
// add the face without turning index
|
|
procFaceAddressing_[curProc].append(patchStart+facei+1);
|
|
// increment the number of faces for this patch
|
|
procPatchSize_[curProc][patchi]++;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
|
|
// Done internal bits of the new mesh and the ordinary patches.
|
|
|
|
|
|
// Per processor, from neighbour processor to the inter-processor patch
|
|
// that communicates with that neighbour
|
|
List<Map<label>> procNbrToInterPatch(nProcs_);
|
|
|
|
// Per processor the faces per inter-processor patch
|
|
List<DynamicList<DynamicList<label>>> interPatchFaces(nProcs_);
|
|
|
|
// Processor boundaries from internal faces
|
|
forAll(neighbour, facei)
|
|
{
|
|
label ownerProc = cellToProc_[owner[facei]];
|
|
label nbrProc = cellToProc_[neighbour[facei]];
|
|
|
|
if (ownerProc != nbrProc)
|
|
{
|
|
// inter - processor patch face found.
|
|
addInterProcFace
|
|
(
|
|
facei,
|
|
ownerProc,
|
|
nbrProc,
|
|
|
|
procNbrToInterPatch,
|
|
interPatchFaces
|
|
);
|
|
}
|
|
}
|
|
|
|
// Add the proper processor faces to the sub information. For faces
|
|
// originating from internal faces this is always -1.
|
|
List<labelListList> subPatchIDs(nProcs_);
|
|
List<labelListList> subPatchStarts(nProcs_);
|
|
forAll(interPatchFaces, proci)
|
|
{
|
|
label nInterfaces = interPatchFaces[proci].size();
|
|
|
|
subPatchIDs[proci].setSize(nInterfaces, labelList(1, label(-1)));
|
|
subPatchStarts[proci].setSize(nInterfaces, labelList(1, label(0)));
|
|
}
|
|
|
|
|
|
// Special handling needed for the case that multiple processor cyclic
|
|
// patches are created on each local processor domain, e.g. if a 3x3 case
|
|
// is decomposed using the decomposition:
|
|
//
|
|
// | 1 | 0 | 2 |
|
|
// cyclic left | 2 | 0 | 1 | cyclic right
|
|
// | 2 | 0 | 1 |
|
|
//
|
|
// - processors 1 and 2 will both have pieces of both cyclic left- and
|
|
// right sub-patches present
|
|
// - the interface patch faces are stored in a single list, where each
|
|
// sub-patch is referenced into the list using a patch start index and
|
|
// size
|
|
// - if the patches are in order (in the boundary file) of left, right
|
|
// - processor 1 will send: left, right
|
|
// - processor 1 will need to receive in reverse order: right, left
|
|
// - similarly for processor 2
|
|
// - the sub-patches are therefore generated in 4 passes of the patch lists
|
|
// 1. add faces from owner patch where local proc i < nbr proc i
|
|
// 2. add faces from nbr patch where local proc i < nbr proc i
|
|
// 3. add faces from owner patch where local proc i > nbr proc i
|
|
// 4. add faces from nbr patch where local proc i > nbr proc i
|
|
|
|
processInterCyclics
|
|
(
|
|
patches,
|
|
interPatchFaces,
|
|
procNbrToInterPatch,
|
|
subPatchIDs,
|
|
subPatchStarts,
|
|
true,
|
|
lessOp<label>()
|
|
);
|
|
|
|
processInterCyclics
|
|
(
|
|
patches,
|
|
interPatchFaces,
|
|
procNbrToInterPatch,
|
|
subPatchIDs,
|
|
subPatchStarts,
|
|
false,
|
|
lessOp<label>()
|
|
);
|
|
|
|
processInterCyclics
|
|
(
|
|
patches,
|
|
interPatchFaces,
|
|
procNbrToInterPatch,
|
|
subPatchIDs,
|
|
subPatchStarts,
|
|
false,
|
|
greaterOp<label>()
|
|
);
|
|
|
|
processInterCyclics
|
|
(
|
|
patches,
|
|
interPatchFaces,
|
|
procNbrToInterPatch,
|
|
subPatchIDs,
|
|
subPatchStarts,
|
|
true,
|
|
greaterOp<label>()
|
|
);
|
|
|
|
|
|
// Sort inter-proc patch by neighbour
|
|
labelList order;
|
|
forAll(procNbrToInterPatch, proci)
|
|
{
|
|
label nInterfaces = procNbrToInterPatch[proci].size();
|
|
|
|
procNeighbourProcessors_[proci].setSize(nInterfaces);
|
|
procProcessorPatchSize_[proci].setSize(nInterfaces);
|
|
procProcessorPatchStartIndex_[proci].setSize(nInterfaces);
|
|
procProcessorPatchSubPatchIDs_[proci].setSize(nInterfaces);
|
|
procProcessorPatchSubPatchStarts_[proci].setSize(nInterfaces);
|
|
|
|
// Info<< "Processor " << proci << endl;
|
|
|
|
// Get sorted neighbour processors
|
|
const Map<label>& curNbrToInterPatch = procNbrToInterPatch[proci];
|
|
labelList nbrs = curNbrToInterPatch.toc();
|
|
|
|
sortedOrder(nbrs, order);
|
|
|
|
DynamicList<DynamicList<label>>& curInterPatchFaces =
|
|
interPatchFaces[proci];
|
|
|
|
forAll(nbrs, i)
|
|
{
|
|
const label nbrProc = nbrs[i];
|
|
const label interPatch = curNbrToInterPatch[nbrProc];
|
|
|
|
procNeighbourProcessors_[proci][i] = nbrProc;
|
|
procProcessorPatchSize_[proci][i] =
|
|
curInterPatchFaces[interPatch].size();
|
|
procProcessorPatchStartIndex_[proci][i] =
|
|
procFaceAddressing_[proci].size();
|
|
|
|
// Add size as last element to substarts and transfer
|
|
append
|
|
(
|
|
subPatchStarts[proci][interPatch],
|
|
curInterPatchFaces[interPatch].size()
|
|
);
|
|
procProcessorPatchSubPatchIDs_[proci][i].transfer
|
|
(
|
|
subPatchIDs[proci][interPatch]
|
|
);
|
|
procProcessorPatchSubPatchStarts_[proci][i].transfer
|
|
(
|
|
subPatchStarts[proci][interPatch]
|
|
);
|
|
|
|
// Info<< " nbr:" << nbrProc << endl;
|
|
// Info<< " interpatch:" << interPatch << endl;
|
|
// Info<< " size:" << procProcessorPatchSize_[proci][i] << endl;
|
|
// Info<< " start:" << procProcessorPatchStartIndex_[proci][i]
|
|
// << endl;
|
|
// Info<< " subPatches:"
|
|
// << procProcessorPatchSubPatchIDs_[proci][i]
|
|
// << endl;
|
|
// Info<< " subStarts:"
|
|
// << procProcessorPatchSubPatchStarts_[proci][i] << endl;
|
|
|
|
// And add all the face labels for interPatch
|
|
DynamicList<label>& interPatchFaces =
|
|
curInterPatchFaces[interPatch];
|
|
|
|
forAll(interPatchFaces, j)
|
|
{
|
|
procFaceAddressing_[proci].append(interPatchFaces[j]);
|
|
}
|
|
interPatchFaces.clearStorage();
|
|
}
|
|
curInterPatchFaces.clearStorage();
|
|
procFaceAddressing_[proci].shrink();
|
|
}
|
|
|
|
|
|
////XXXXXXX
|
|
//// Print a bit
|
|
// forAll(procPatchStartIndex_, proci)
|
|
// {
|
|
// Info<< "Processor:" << proci << endl;
|
|
//
|
|
// Info<< " total faces:" << procFaceAddressing_[proci].size()
|
|
// << endl;
|
|
//
|
|
// const labelList& curProcPatchStartIndex = procPatchStartIndex_[proci];
|
|
//
|
|
// forAll(curProcPatchStartIndex, patchi)
|
|
// {
|
|
// Info<< " patch:" << patchi
|
|
// << "\tstart:" << curProcPatchStartIndex[patchi]
|
|
// << "\tsize:" << procPatchSize_[proci][patchi]
|
|
// << endl;
|
|
// }
|
|
// }
|
|
// Info<< endl;
|
|
//
|
|
// forAll(procNeighbourProcessors_, proci)
|
|
// {
|
|
// Info<< "Processor " << proci << endl;
|
|
//
|
|
// forAll(procNeighbourProcessors_[proci], i)
|
|
// {
|
|
// Info<< " nbr:" << procNeighbourProcessors_[proci][i] << endl;
|
|
// Info<< " size:" << procProcessorPatchSize_[proci][i] << endl;
|
|
// Info<< " start:" << procProcessorPatchStartIndex_[proci][i]
|
|
// << endl;
|
|
// }
|
|
// }
|
|
// Info<< endl;
|
|
//
|
|
// forAll(procFaceAddressing_, proci)
|
|
// {
|
|
// Info<< "Processor:" << proci << endl;
|
|
//
|
|
// Info<< " faces:" << procFaceAddressing_[proci] << endl;
|
|
// }
|
|
|
|
|
|
|
|
Info<< "\nDistributing points to processors" << endl;
|
|
// For every processor, loop through the list of faces for the processor.
|
|
// For every face, loop through the list of points and mark the point as
|
|
// used for the processor. Collect the list of used points for the
|
|
// processor.
|
|
|
|
forAll(procPointAddressing_, proci)
|
|
{
|
|
boolList pointLabels(nPoints(), false);
|
|
|
|
// Get reference to list of used faces
|
|
const labelList& procFaceLabels = procFaceAddressing_[proci];
|
|
|
|
forAll(procFaceLabels, facei)
|
|
{
|
|
// Because of the turning index, some labels may be negative
|
|
const labelList& facePoints = fcs[mag(procFaceLabels[facei]) - 1];
|
|
|
|
forAll(facePoints, pointi)
|
|
{
|
|
// Mark the point as used
|
|
pointLabels[facePoints[pointi]] = true;
|
|
}
|
|
}
|
|
|
|
// Collect the used points
|
|
labelList& procPointLabels = procPointAddressing_[proci];
|
|
|
|
procPointLabels.setSize(pointLabels.size());
|
|
|
|
label nUsedPoints = 0;
|
|
|
|
forAll(pointLabels, pointi)
|
|
{
|
|
if (pointLabels[pointi])
|
|
{
|
|
procPointLabels[nUsedPoints] = pointi;
|
|
|
|
nUsedPoints++;
|
|
}
|
|
}
|
|
|
|
// Reset the size of used points
|
|
procPointLabels.setSize(nUsedPoints);
|
|
}
|
|
}
|
|
|
|
// ************************************************************************* //
|