openfoam/applications/utilities/parallelProcessing/redistributePar/redistributePar.C

815 lines
23 KiB
C

/*---------------------------------------------------------------------------*\
========= |
\\ / F ield | OpenFOAM: The Open Source CFD Toolbox
\\ / O peration |
\\ / A nd | Copyright (C) 2011-2014 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/>.
Application
redistributePar
Description
Redistributes existing decomposed mesh and fields according to the current
settings in the decomposeParDict file.
Must be run on maximum number of source and destination processors.
Balances mesh and writes new mesh to new time directory.
Can also work like decomposePar:
\verbatim
# Create empty processor directories (have to exist for argList)
mkdir processor0
..
mkdir processorN
# Copy undecomposed polyMesh
cp -r constant processor0
# Distribute
mpirun -np ddd redistributePar -parallel
\endverbatim
\*---------------------------------------------------------------------------*/
#include "fvMesh.H"
#include "decompositionMethod.H"
#include "PstreamReduceOps.H"
#include "fvCFD.H"
#include "fvMeshDistribute.H"
#include "mapDistributePolyMesh.H"
#include "IOobjectList.H"
#include "globalIndex.H"
#include "loadOrCreateMesh.H"
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
// Tolerance (as fraction of the bounding box). Needs to be fairly lax since
// usually meshes get written with limited precision (6 digits)
static const scalar defaultMergeTol = 1e-6;
// Get merging distance when matching face centres
scalar getMergeDistance
(
const argList& args,
const Time& runTime,
const boundBox& bb
)
{
scalar mergeTol = defaultMergeTol;
args.optionReadIfPresent("mergeTol", mergeTol);
scalar writeTol =
Foam::pow(scalar(10.0), -scalar(IOstream::defaultPrecision()));
Info<< "Merge tolerance : " << mergeTol << nl
<< "Write tolerance : " << writeTol << endl;
if (runTime.writeFormat() == IOstream::ASCII && mergeTol < writeTol)
{
FatalErrorIn("getMergeDistance")
<< "Your current settings specify ASCII writing with "
<< IOstream::defaultPrecision() << " digits precision." << endl
<< "Your merging tolerance (" << mergeTol << ") is finer than this."
<< endl
<< "Please change your writeFormat to binary"
<< " or increase the writePrecision" << endl
<< "or adjust the merge tolerance (-mergeTol)."
<< exit(FatalError);
}
scalar mergeDist = mergeTol * bb.mag();
Info<< "Overall meshes bounding box : " << bb << nl
<< "Relative tolerance : " << mergeTol << nl
<< "Absolute matching distance : " << mergeDist << nl
<< endl;
return mergeDist;
}
//void printMeshData(Ostream& os, const polyMesh& mesh)
//{
// os << "Number of points: " << mesh.points().size() << nl
// << " faces: " << mesh.faces().size() << nl
// << " internal faces: " << mesh.faceNeighbour().size() << nl
// << " cells: " << mesh.cells().size() << nl
// << " boundary patches: " << mesh.boundaryMesh().size() << nl
// << " point zones: " << mesh.pointZones().size() << nl
// << " face zones: " << mesh.faceZones().size() << nl
// << " cell zones: " << mesh.cellZones().size() << nl;
//}
void printMeshData(const polyMesh& mesh)
{
// Collect all data on master
globalIndex globalCells(mesh.nCells());
labelListList patchNeiProcNo(Pstream::nProcs());
labelListList patchSize(Pstream::nProcs());
const labelList& pPatches = mesh.globalData().processorPatches();
patchNeiProcNo[Pstream::myProcNo()].setSize(pPatches.size());
patchSize[Pstream::myProcNo()].setSize(pPatches.size());
forAll(pPatches, i)
{
const processorPolyPatch& ppp = refCast<const processorPolyPatch>
(
mesh.boundaryMesh()[pPatches[i]]
);
patchNeiProcNo[Pstream::myProcNo()][i] = ppp.neighbProcNo();
patchSize[Pstream::myProcNo()][i] = ppp.size();
}
Pstream::gatherList(patchNeiProcNo);
Pstream::gatherList(patchSize);
// Print stats
globalIndex globalBoundaryFaces(mesh.nFaces()-mesh.nInternalFaces());
label maxProcCells = 0;
label totProcFaces = 0;
label maxProcPatches = 0;
label totProcPatches = 0;
label maxProcFaces = 0;
for (label procI = 0; procI < Pstream::nProcs(); procI++)
{
Info<< endl
<< "Processor " << procI << nl
<< " Number of cells = " << globalCells.localSize(procI)
<< endl;
label nProcFaces = 0;
const labelList& nei = patchNeiProcNo[procI];
forAll(patchNeiProcNo[procI], i)
{
Info<< " Number of faces shared with processor "
<< patchNeiProcNo[procI][i] << " = " << patchSize[procI][i]
<< endl;
nProcFaces += patchSize[procI][i];
}
Info<< " Number of processor patches = " << nei.size() << nl
<< " Number of processor faces = " << nProcFaces << nl
<< " Number of boundary faces = "
<< globalBoundaryFaces.localSize(procI) << endl;
maxProcCells = max(maxProcCells, globalCells.localSize(procI));
totProcFaces += nProcFaces;
totProcPatches += nei.size();
maxProcPatches = max(maxProcPatches, nei.size());
maxProcFaces = max(maxProcFaces, nProcFaces);
}
// Stats
scalar avgProcCells = scalar(globalCells.size())/Pstream::nProcs();
scalar avgProcPatches = scalar(totProcPatches)/Pstream::nProcs();
scalar avgProcFaces = scalar(totProcFaces)/Pstream::nProcs();
// In case of all faces on one processor. Just to avoid division by 0.
if (totProcPatches == 0)
{
avgProcPatches = 1;
}
if (totProcFaces == 0)
{
avgProcFaces = 1;
}
Info<< nl
<< "Number of processor faces = " << totProcFaces/2 << nl
<< "Max number of cells = " << maxProcCells
<< " (" << 100.0*(maxProcCells-avgProcCells)/avgProcCells
<< "% above average " << avgProcCells << ")" << nl
<< "Max number of processor patches = " << maxProcPatches
<< " (" << 100.0*(maxProcPatches-avgProcPatches)/avgProcPatches
<< "% above average " << avgProcPatches << ")" << nl
<< "Max number of faces between processors = " << maxProcFaces
<< " (" << 100.0*(maxProcFaces-avgProcFaces)/avgProcFaces
<< "% above average " << avgProcFaces << ")" << nl
<< endl;
}
// Debugging: write volScalarField with decomposition for post processing.
void writeDecomposition
(
const word& name,
const fvMesh& mesh,
const labelList& decomp
)
{
Info<< "Writing wanted cell distribution to volScalarField " << name
<< " for postprocessing purposes." << nl << endl;
volScalarField procCells
(
IOobject
(
name,
mesh.time().timeName(),
mesh,
IOobject::NO_READ,
IOobject::AUTO_WRITE,
false // do not register
),
mesh,
dimensionedScalar(name, dimless, -1),
zeroGradientFvPatchScalarField::typeName
);
forAll(procCells, cI)
{
procCells[cI] = decomp[cI];
}
procCells.write();
}
// Read vol or surface fields
//template<class T, class Mesh>
template<class GeoField>
void readFields
(
const boolList& haveMesh,
const fvMesh& mesh,
const autoPtr<fvMeshSubset>& subsetterPtr,
IOobjectList& allObjects,
PtrList<GeoField>& fields
)
{
//typedef GeometricField<T, fvPatchField, Mesh> fldType;
// Get my objects of type
IOobjectList objects(allObjects.lookupClass(GeoField::typeName));
// Check that we all have all objects
wordList objectNames = objects.toc();
// Get master names
wordList masterNames(objectNames);
Pstream::scatter(masterNames);
if (haveMesh[Pstream::myProcNo()] && objectNames != masterNames)
{
FatalErrorIn("readFields()")
<< "differing fields of type " << GeoField::typeName
<< " on processors." << endl
<< "Master has:" << masterNames << endl
<< Pstream::myProcNo() << " has:" << objectNames
<< abort(FatalError);
}
fields.setSize(masterNames.size());
// Have master send all fields to processors that don't have a mesh
if (Pstream::master())
{
forAll(masterNames, i)
{
const word& name = masterNames[i];
IOobject& io = *objects[name];
io.writeOpt() = IOobject::AUTO_WRITE;
// Load field
fields.set(i, new GeoField(io, mesh));
// Create zero sized field and send
if (subsetterPtr.valid())
{
tmp<GeoField> tsubfld = subsetterPtr().interpolate(fields[i]);
// Send to all processors that don't have a mesh
for (label procI = 1; procI < Pstream::nProcs(); procI++)
{
if (!haveMesh[procI])
{
OPstream toProc(Pstream::blocking, procI);
toProc<< tsubfld();
}
}
}
}
}
else if (!haveMesh[Pstream::myProcNo()])
{
// Don't have mesh (nor fields). Receive empty field from master.
forAll(masterNames, i)
{
const word& name = masterNames[i];
// Receive field
IPstream fromMaster(Pstream::blocking, Pstream::masterNo());
dictionary fieldDict(fromMaster);
fields.set
(
i,
new GeoField
(
IOobject
(
name,
mesh.time().timeName(),
mesh,
IOobject::NO_READ,
IOobject::AUTO_WRITE
),
mesh,
fieldDict
)
);
//// Write it for next time round (since mesh gets written as well)
//fields[i].write();
}
}
else
{
// Have mesh so just try to load
forAll(masterNames, i)
{
const word& name = masterNames[i];
IOobject& io = *objects[name];
io.writeOpt() = IOobject::AUTO_WRITE;
// Load field
fields.set(i, new GeoField(io, mesh));
}
}
}
// Debugging: compare two fields.
void compareFields
(
const scalar tolDim,
const volVectorField& a,
const volVectorField& b
)
{
forAll(a, cellI)
{
if (mag(b[cellI] - a[cellI]) > tolDim)
{
FatalErrorIn
(
"compareFields"
"(const scalar, const volVectorField&, const volVectorField&)"
) << "Did not map volVectorField correctly:" << nl
<< "cell:" << cellI
<< " transfer b:" << b[cellI]
<< " real cc:" << a[cellI]
<< abort(FatalError);
}
}
forAll(a.boundaryField(), patchI)
{
// We have real mesh cellcentre and
// mapped original cell centre.
const fvPatchVectorField& aBoundary =
a.boundaryField()[patchI];
const fvPatchVectorField& bBoundary =
b.boundaryField()[patchI];
if (!bBoundary.coupled())
{
forAll(aBoundary, i)
{
if (mag(aBoundary[i] - bBoundary[i]) > tolDim)
{
WarningIn
(
"compareFields"
"(const scalar, const volVectorField&"
", const volVectorField&)"
) << "Did not map volVectorField correctly:"
<< endl
<< "patch:" << patchI << " patchFace:" << i
<< " cc:" << endl
<< " real :" << aBoundary[i] << endl
<< " mapped :" << bBoundary[i] << endl
<< "This might be just a precision entry"
<< " on writing the mesh." << endl;
//<< abort(FatalError);
}
}
}
}
}
int main(int argc, char *argv[])
{
# include "addRegionOption.H"
# include "addOverwriteOption.H"
argList::addOption
(
"mergeTol",
"scalar",
"specify the merge distance relative to the bounding box size "
"(default 1e-6)"
);
# include "setRootCase.H"
if (env("FOAM_SIGFPE"))
{
WarningIn(args.executable())
<< "Detected floating point exception trapping (FOAM_SIGFPE)."
<< " This might give" << nl
<< " problems when mapping fields. Switch it off in case"
<< " of problems." << endl;
}
// Create processor directory if non-existing
if (!Pstream::master() && !isDir(args.path()))
{
Pout<< "Creating case directory " << args.path() << endl;
mkDir(args.path());
}
// Make sure we do not use the master-only reading.
regIOobject::fileModificationChecking = regIOobject::timeStamp;
# include "createTime.H"
runTime.functionObjects().off();
word regionName = polyMesh::defaultRegion;
fileName meshSubDir;
if (args.optionReadIfPresent("region", regionName))
{
meshSubDir = regionName/polyMesh::meshSubDir;
}
else
{
meshSubDir = polyMesh::meshSubDir;
}
Info<< "Using mesh subdirectory " << meshSubDir << nl << endl;
const bool overwrite = args.optionFound("overwrite");
// Get time instance directory. Since not all processors have meshes
// just use the master one everywhere.
fileName masterInstDir;
if (Pstream::master())
{
masterInstDir = runTime.findInstance(meshSubDir, "points");
}
Pstream::scatter(masterInstDir);
// Check who has a mesh
const fileName meshPath = runTime.path()/masterInstDir/meshSubDir;
Info<< "Found points in " << meshPath << nl << endl;
boolList haveMesh(Pstream::nProcs(), false);
haveMesh[Pstream::myProcNo()] = isDir(meshPath);
Pstream::gatherList(haveMesh);
Pstream::scatterList(haveMesh);
Info<< "Per processor mesh availability : " << haveMesh << endl;
const bool allHaveMesh = (findIndex(haveMesh, false) == -1);
autoPtr<fvMesh> meshPtr = loadOrCreateMesh
(
IOobject
(
regionName,
masterInstDir,
runTime,
Foam::IOobject::MUST_READ
)
);
fvMesh& mesh = meshPtr();
// Print some statistics
Info<< "Before distribution:" << endl;
printMeshData(mesh);
IOdictionary decompositionDict
(
IOobject
(
"decomposeParDict",
runTime.system(),
mesh,
IOobject::MUST_READ_IF_MODIFIED,
IOobject::NO_WRITE
)
);
labelList finalDecomp;
// Create decompositionMethod and new decomposition
{
autoPtr<decompositionMethod> decomposer
(
decompositionMethod::New
(
decompositionDict
)
);
if (!decomposer().parallelAware())
{
WarningIn(args.executable())
<< "You have selected decomposition method "
<< decomposer().typeName
<< " which does" << endl
<< "not synchronise the decomposition across"
<< " processor patches." << endl
<< " You might want to select a decomposition method which"
<< " is aware of this. Continuing."
<< endl;
}
finalDecomp = decomposer().decompose(mesh, mesh.cellCentres());
}
// Dump decomposition to volScalarField
if (!overwrite)
{
writeDecomposition("decomposition", mesh, finalDecomp);
}
// Create 0 sized mesh to do all the generation of zero sized
// fields on processors that have zero sized meshes. Note that this is
// only nessecary on master but since polyMesh construction with
// Pstream::parRun does parallel comms we have to do it on all
// processors
autoPtr<fvMeshSubset> subsetterPtr;
if (!allHaveMesh)
{
// Find last non-processor patch.
const polyBoundaryMesh& patches = mesh.boundaryMesh();
label nonProcI = -1;
forAll(patches, patchI)
{
if (isA<processorPolyPatch>(patches[patchI]))
{
break;
}
nonProcI++;
}
if (nonProcI == -1)
{
FatalErrorIn(args.executable())
<< "Cannot find non-processor patch on processor "
<< Pstream::myProcNo() << endl
<< " Current patches:" << patches.names() << abort(FatalError);
}
// Subset 0 cells, no parallel comms. This is used to create zero-sized
// fields.
subsetterPtr.reset(new fvMeshSubset(mesh));
subsetterPtr().setLargeCellSubset(labelHashSet(0), nonProcI, false);
}
// Get original objects (before incrementing time!)
IOobjectList objects(mesh, runTime.timeName());
// We don't want to map the decomposition (mapping already tested when
// mapping the cell centre field)
IOobjectList::iterator iter = objects.find("decomposition");
if (iter != objects.end())
{
objects.erase(iter);
}
// volFields
PtrList<volScalarField> volScalarFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
volScalarFields
);
PtrList<volVectorField> volVectorFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
volVectorFields
);
PtrList<volSphericalTensorField> volSphereTensorFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
volSphereTensorFields
);
PtrList<volSymmTensorField> volSymmTensorFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
volSymmTensorFields
);
PtrList<volTensorField> volTensorFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
volTensorFields
);
// surfaceFields
PtrList<surfaceScalarField> surfScalarFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
surfScalarFields
);
PtrList<surfaceVectorField> surfVectorFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
surfVectorFields
);
PtrList<surfaceSphericalTensorField> surfSphereTensorFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
surfSphereTensorFields
);
PtrList<surfaceSymmTensorField> surfSymmTensorFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
surfSymmTensorFields
);
PtrList<surfaceTensorField> surfTensorFields;
readFields
(
haveMesh,
mesh,
subsetterPtr,
objects,
surfTensorFields
);
// Debugging: Create additional volField that will be mapped.
// Used to test correctness of mapping
//volVectorField mapCc("mapCc", 1*mesh.C());
// Global matching tolerance
const scalar tolDim = getMergeDistance
(
args,
runTime,
mesh.bounds()
);
// Mesh distribution engine
fvMeshDistribute distributor(mesh, tolDim);
//Pout<< "Wanted distribution:"
// << distributor.countCells(finalDecomp) << nl << endl;
// Do actual sending/receiving of mesh
autoPtr<mapDistributePolyMesh> map = distributor.distribute(finalDecomp);
//// Distribute any non-registered data accordingly
//map().distributeFaceData(faceCc);
// Print some statistics
Info<< "After distribution:" << endl;
printMeshData(mesh);
if (!overwrite)
{
runTime++;
}
else
{
mesh.setInstance(masterInstDir);
}
Info<< "Writing redistributed mesh to " << runTime.timeName() << nl << endl;
mesh.write();
// Debugging: test mapped cellcentre field.
//compareFields(tolDim, mesh.C(), mapCc);
// Print nice message
// ~~~~~~~~~~~~~~~~~~
// Determine which processors remain so we can print nice final message.
labelList nFaces(Pstream::nProcs());
nFaces[Pstream::myProcNo()] = mesh.nFaces();
Pstream::gatherList(nFaces);
Pstream::scatterList(nFaces);
Info<< nl
<< "You can pick up the redecomposed mesh from the polyMesh directory"
<< " in " << runTime.timeName() << "." << nl
<< "If you redecomposed the mesh to less processors you can delete"
<< nl
<< "the processor directories with 0 sized meshes in them." << nl
<< "Below is a sample set of commands to do this."
<< " Take care when issuing these" << nl
<< "commands." << nl << endl;
forAll(nFaces, procI)
{
fileName procDir = "processor" + name(procI);
if (nFaces[procI] == 0)
{
Info<< " rm -r " << procDir.c_str() << nl;
}
else
{
fileName timeDir = procDir/runTime.timeName()/meshSubDir;
fileName constDir = procDir/runTime.constant()/meshSubDir;
Info<< " rm -r " << constDir.c_str() << nl
<< " mv " << timeDir.c_str()
<< ' ' << constDir.c_str() << nl;
}
}
Info<< endl;
Info<< "End\n" << endl;
return 0;
}
// ************************************************************************* //