ForkMaster/netgen
1
0
Fork 0
mirror of https://github.com/NGSolve/netgen.git synced 2024-11-14 10:08:32 +05:00
Code Issues Packages Projects Releases Wiki Activity
777f9c65f2
netgen/libsrc/meshing/parallelmesh.cpp

1911 lines
47 KiB
C++
Raw Normal View History

move parallel files
2009-01-26 00:37:14 +05:00
#ifdef PARALLEL
#include <meshing.hpp>
#include "paralleltop.hpp"
#ifdef METIS
namespace metis { extern "C" {
#include <metis.h>
} }
#endif
using namespace metis;
namespace netgen
{
// slaves receive the mesh from the master
void Mesh :: ReceiveParallelMesh ( )
{
int timer = NgProfiler::CreateTimer ("ReceiveParallelMesh");
NgProfiler::RegionTimer reg(timer);
int timer_pts = NgProfiler::CreateTimer ("Receive Points");
int timer_els = NgProfiler::CreateTimer ("Receive Elements");
int timer_sels = NgProfiler::CreateTimer ("Receive Surface Elements");
int timer_edges = NgProfiler::CreateTimer ("Receive Edges");
#ifdef SCALASCA
#pragma pomp inst begin(loadmesh)
#endif
// PrintMessage (1, "LOAD PARALLEL MESH");
// MPI_Barrier (MPI_COMM_WORLD);
string st;
double doublebuf[100];
int i, n;
int tag_dim = 10, tag_token = 100, tag_n=11, tag_pnum=12, tag_point=13;
int tag_index = 101, tag_facedescr = 102;
MPI_Status status;
bool endmesh = false;
int dim;
int nelglob, nelloc, nvglob, nedglob, nfaglob;
// receive global values
MPI_Bcast( &nelglob, 1, MPI_INT, 0, MPI_COMM_WORLD );
MPI_Bcast( &nvglob, 1, MPI_INT, 0, MPI_COMM_WORLD );
MPI_Bcast( &nedglob, 1, MPI_INT, 0, MPI_COMM_WORLD );
MPI_Bcast( &nfaglob, 1, MPI_INT, 0, MPI_COMM_WORLD );
MPI_Bcast( &dimension, 1, MPI_INT, 0, MPI_COMM_WORLD );
MyMPI_Recv ( nelloc, 0 );
paralleltop -> SetNVGlob ( nvglob );
paralleltop -> SetNEGlob ( nelglob );
INDEX_CLOSED_HASHTABLE<int> glob2loc_vert_ht (1);
PrintMessage (1, "Receive mesh");
int ve = 0;
while (!endmesh)
{
MyMPI_Recv ( st, 0 );
// receive vertices
if (st == "vertex")
{
NgProfiler::RegionTimer reg(timer_pts);
Array<double> pointarray;
MyMPI_Recv ( pointarray, 0 );
int numvert = pointarray.Size() / 5;
paralleltop -> SetNV (numvert);
glob2loc_vert_ht.SetSize (3*numvert+1);
for ( int vert=0; vert<numvert; vert++ )
{
int globvert = int (pointarray[ vert*5 ]);
paralleltop->SetLoc2Glob_Vert ( vert+1, globvert );
glob2loc_vert_ht.Set (globvert, vert+1);
netgen::Point<3> p;
p(0) = pointarray[vert*5+1];
p(1) = pointarray[vert*5+2];
p(2) = pointarray[vert*5+3];
AddPoint (p);
(*this)[PointIndex(vert+1)] .Singularity ( pointarray[vert*5+4] );
}
Array<int> dist_pnums;
MyMPI_Recv ( dist_pnums, 0);
for (int hi = 0; hi < dist_pnums.Size(); hi += 3)
{
paralleltop ->
SetDistantPNum ( dist_pnums[hi+1], dist_pnums[hi], dist_pnums[hi+2]);
}
}
if ( strcmp (st.c_str(), "volumeelements" ) == 0 )
{
NgProfiler::RegionTimer reg(timer_els);
*testout << "receiving elements" << endl;
Element el;
Array<int> elarray;
MyMPI_Recv ( elarray, 0);
int ind = 0;
int elnum = 1;
int nelloc = elarray[ind++];
paralleltop -> SetNE (nelloc);
while ( ind < elarray.Size() )
{
paralleltop->SetLoc2Glob_VolEl ( elnum, elarray[ind++]);
el.SetIndex(elarray[ind++]);
el.SetNP(elarray[ind++]);
for ( int j = 0; j < el.GetNP(); j++)
el[j] = glob2loc_vert_ht.Get (elarray[ind++]);
AddVolumeElement (el);
elnum++;
}
}
if (strcmp (st.c_str(), "facedescriptor") == 0)
{
Array<double> doublebuf;
MyMPI_Recv( doublebuf, 0 );
int faceind = AddFaceDescriptor (FaceDescriptor(int(doublebuf[0]), int(doublebuf[1]), int(doublebuf[2]), 0));
GetFaceDescriptor(faceind).SetBCProperty (int(doublebuf[3]));
GetFaceDescriptor(faceind).domin_singular = doublebuf[4];
GetFaceDescriptor(faceind).domout_singular = doublebuf[5];
}
if (strcmp (st.c_str(), "surfaceelementsgi") == 0)
{
NgProfiler::RegionTimer reg(timer_sels);
int j;
int surfnr, bcp, domin, domout, nep, faceind = 0;
int globsel;
int * selbuf;
selbuf = 0;
int bufsize;
// receive:
// faceindex
// nep
// tri.pnum
// tri.geominfopi.trignum
int nlocsel;
MyMPI_Recv ( selbuf, bufsize, 0);
int ii= 0;
int sel = 0;
nlocsel = selbuf[ii++];
paralleltop -> SetNSE ( nlocsel );
while ( ii < bufsize-1 )
{
globsel = selbuf[ii++];
faceind = selbuf[ii++];
bool isghost = selbuf[ii++];
nep = selbuf[ii++];
Element2d tri(nep);
tri.SetIndex(faceind);
for( j=1; j<=nep; j++)
{
tri.PNum(j) = glob2loc_vert_ht.Get (selbuf[ii++]);
// tri.GeomInfoPi(j).trignum = paralleltop->Glob2Loc_SurfEl(selbuf[ii++]);
tri.GeomInfoPi(j).trignum = selbuf[ii++];
// Frage JS->AS: Brauchst du die trignum irgendwo ?
// sie sonst nur bei STL - Geometrien benötigt
// die Umrechnung war ein bottleneck !
}
tri.SetGhost(isghost);
paralleltop->SetLoc2Glob_SurfEl ( sel+1, globsel );
AddSurfaceElement (tri);
sel ++;
}
delete [] selbuf ;
}
if (strcmp (st.c_str(), "edgesegmentsgi") == 0)
{
NgProfiler::RegionTimer reg(timer_edges);
int endtime, starttime;
starttime = clock();
double * segmbuf = 0;
int bufsize;
MyMPI_Recv ( segmbuf, bufsize, 0);
Segment seg;
int hi;
int globsegi;
int ii = 0;
int segi = 1;
int nsegloc = int ( bufsize / 14 ) ;
paralleltop -> SetNSegm ( nsegloc );
while ( ii < bufsize )
{
globsegi = int (segmbuf[ii++]);
seg.si = int (segmbuf[ii++]);
seg.p1 = glob2loc_vert_ht.Get (int(segmbuf[ii++]));
seg.p2 = glob2loc_vert_ht.Get (int(segmbuf[ii++]));
seg.geominfo[0].trignum = int( segmbuf[ii++] );
seg.geominfo[1].trignum = int ( segmbuf[ii++]);
seg.surfnr1 = int ( segmbuf[ii++]);
seg.surfnr2 = int ( segmbuf[ii++]);
seg.edgenr = int ( segmbuf[ii++]);
seg.epgeominfo[0].dist = segmbuf[ii++];
seg.epgeominfo[1].edgenr = int (segmbuf[ii++]);
seg.epgeominfo[1].dist = segmbuf[ii++];
seg.singedge_left = segmbuf[ii++];
seg.singedge_right = segmbuf[ii++];
seg.epgeominfo[0].edgenr = seg.epgeominfo[1].edgenr;
seg.domin = seg.surfnr1;
seg.domout = seg.surfnr2;
if ( seg.p1 >0 && seg.p2 > 0 )
{
paralleltop-> SetLoc2Glob_Segm ( segi, globsegi );
AddSegment (seg);
segi++;
}
}
delete [] segmbuf;
endtime = clock();
(*testout) << "Receiving Time fde = " << double(endtime - starttime)/CLOCKS_PER_SEC << endl;
}
/*
for ( int eli = 1; eli < GetNE(); eli++ )
{
Element & el = VolumeElement(eli);
}
*/
if (strcmp (st.c_str(), "endmesh") == 0)
{
endmesh = true;
}
}
// ohne diesem Zusammenwarten gibts Abstürze, und ich weiß nicht warum ???
// MPI_Barrier (MPI_COMM_WORLD);
// PrintMessage (1, "Have recevied the mesh");
// MPI_Barrier (MPI_COMM_WORLD);
// paralleltop -> SetNV ( this -> GetNV() );
// for ( int i = 0; i < GetNV(); i++ )
// paralleltop -> SetLoc2Glob_Vert ( i+1, (*loc2globvert)[i] );
int timerloc = NgProfiler::CreateTimer ("Update local mesh");
int timerloc2 = NgProfiler::CreateTimer ("CalcSurfacesOfNode");
NgProfiler::RegionTimer regloc(timerloc);
PrintMessage (2, "Got ", GetNE(), " elements");
NgProfiler::StartTimer (timerloc2);
CalcSurfacesOfNode ();
NgProfiler::StopTimer (timerloc2);
// BuildConnectedNodes ();
topology -> Update();
// UpdateOverlap();
clusters -> Update();
SetNextMajorTimeStamp();
// paralleltop->Print();
#ifdef SCALASCA
#pragma pomp inst end(loadmesh)
#endif
}
// distribute the mesh to the slave processors
// call it only for the master !
void Mesh :: Distribute ()
{
if ( id != 0 || ntasks == 1 ) return;
// metis partition of mesh, only if more than one proc
#ifdef SCALASCA
#pragma pomp inst begin(metis)
#endif
// partition mesh
ParallelMetis ();
#ifdef SCALASCA
#pragma pomp inst end(metis)
#endif
// send partition
SendMesh ();
paralleltop -> UpdateCoarseGrid();
#ifdef SCALASCA
#pragma pomp inst end(loadmesh_seq)
#endif
// paralleltop -> Print();
}
void Mesh :: ParallelMetis ( )
{
int timer = NgProfiler::CreateTimer ("Mesh::Partition");
NgProfiler::RegionTimer reg(timer);
PrintMessage (3, "Metis called");
if (GetDimension() == 2)
{
PartDualHybridMesh2D ( ); // neloc );
return;
}
int ne = GetNE();
int nn = GetNP();
if (ntasks <= 2)
{
if (ntasks == 1) return;
for (int i=1; i<=ne; i++)
VolumeElement(i).SetPartition(1);
return;
}
bool uniform_els = true;
ELEMENT_TYPE elementtype = TET; // VolumeElement(1).GetType();
// metis works only for uniform tet/hex meshes
for ( int el = 2; el <= GetNE(); el++ )
if ( VolumeElement(el).GetType() != elementtype )
{
// int nelperproc = ne / (ntasks-1);
// for (int i=1; i<=ne; i++)
// {
// int partition = i / nelperproc + 1;
// if ( partition >= ntasks ) partition = ntasks-1;
// VolumeElement(i).SetPartition(partition);
// }
uniform_els = false;
break;
}
// uniform_els = false;
if (!uniform_els)
{
PartHybridMesh ( ); // neloc );
return;
}
// uniform (TET) mesh, JS
int npe = VolumeElement(1).GetNP();
idxtype *elmnts;
elmnts = new idxtype[ne*npe];
int etype;
if (elementtype == TET)
etype = 2;
else if (elementtype == HEX)
etype = 3;
for (int i=1; i<=ne; i++)
for (int j=1; j<=npe; j++)
elmnts[(i-1)*npe+(j-1)] = VolumeElement(i).PNum(j)-1;
int numflag = 0;
int nparts = ntasks-1;
int edgecut;
idxtype *epart, *npart;
epart = new idxtype[ne];
npart = new idxtype[nn];
// if ( ntasks == 1 )
// {
// (*this) = *mastermesh;
// nparts = 4;
// metis :: METIS_PartMeshDual (&ne, &nn, elmnts, &etype, &numflag, &nparts,
// &edgecut, epart, npart);
// cout << "done" << endl;
// cout << "edge-cut: " << edgecut << ", balance: " << metis :: ComputeElementBalance(ne, nparts, epart) << endl;
// for (int i=1; i<=ne; i++)
// {
// mastermesh->VolumeElement(i).SetPartition(epart[i-1]);
// }
// return;
// }
cout << "call metis ... " << flush;
int timermetis = NgProfiler::CreateTimer ("Metis itself");
NgProfiler::StartTimer (timermetis);
metis :: METIS_PartMeshDual (&ne, &nn, elmnts, &etype, &numflag, &nparts,
&edgecut, epart, npart);
NgProfiler::StopTimer (timermetis);
cout << "complete" << endl;
cout << "edge-cut: " << edgecut << ", balance: " << metis :: ComputeElementBalance(ne, nparts, epart) << endl;
// partition numbering by metis : 0 ... ntasks - 1
// we want: 1 ... ntasks
for (int i=1; i<=ne; i++)
VolumeElement(i).SetPartition(epart[i-1] + 1);
delete [] elmnts;
delete [] epart;
delete [] npart;
}
void Mesh :: PartHybridMesh () // Array<int> & neloc )
{
int ne = GetNE();
int nn = GetNP();
int nedges = topology->GetNEdges();
idxtype *xadj, * adjacency, *v_weights = NULL, *e_weights = NULL;
int weightflag = 0;
int numflag = 0;
int nparts = ntasks - 1;
int options[5];
options[0] = 0;
int edgecut;
idxtype * part;
xadj = new idxtype[nn+1];
part = new idxtype[nn];
Array<int> cnt(nn+1);
cnt = 0;
for ( int edge = 1; edge <= nedges; edge++ )
{
int v1, v2;
topology->GetEdgeVertices ( edge, v1, v2);
cnt[v1-1] ++;
cnt[v2-1] ++;
}
xadj[0] = 0;
for ( int n = 1; n <= nn; n++ )
{
xadj[n] = idxtype(xadj[n-1] + cnt[n-1]);
}
adjacency = new idxtype[xadj[nn]];
cnt = 0;
for ( int edge = 1; edge <= nedges; edge++ )
{
int v1, v2;
topology->GetEdgeVertices ( edge, v1, v2);
adjacency[ xadj[v1-1] + cnt[v1-1] ] = v2-1;
adjacency[ xadj[v2-1] + cnt[v2-1] ] = v1-1;
cnt[v1-1]++;
cnt[v2-1]++;
}
for ( int vert = 0; vert < nn; vert++ )
{
FlatArray<int> array ( cnt[vert], &adjacency[ xadj[vert] ] );
BubbleSort(array);
}
metis :: METIS_PartGraphKway ( &nn, xadj, adjacency, v_weights, e_weights, &weightflag,
&numflag, &nparts, options, &edgecut, part );
Array<int> nodesinpart(ntasks);
for ( int el = 1; el <= ne; el++ )
{
Element & volel = VolumeElement(el);
nodesinpart = 0;
// VolumeElement(el).SetPartition(part[ volel[1] ] + 1);
int el_np = volel.GetNP();
int partition = 0;
for ( int i = 0; i < el_np; i++ )
nodesinpart[ part[volel[i]-1]+1 ] ++;
for ( int i = 1; i < ntasks; i++ )
if ( nodesinpart[i] > nodesinpart[partition] )
partition = i;
volel.SetPartition(partition);
}
/*
for ( int i=1; i<=ne; i++)
{
neloc[ VolumeElement(i).GetPartition() ] ++;
}
*/
delete [] xadj;
delete [] part;
delete [] adjacency;
}
void Mesh :: PartDualHybridMesh ( ) // Array<int> & neloc )
{
int ne = GetNE();
int nn = GetNP();
int nedges = topology->GetNEdges();
int nfaces = topology->GetNFaces();
idxtype *xadj, * adjacency, *v_weights = NULL, *e_weights = NULL;
int weightflag = 0;
int numflag = 0;
int nparts = ntasks - 1;
int options[5];
options[0] = 0;
int edgecut;
idxtype * part;
Array<int, 0> facevolels1(nfaces), facevolels2(nfaces);
facevolels1 = -1;
facevolels2 = -1;
Array<int, 0> elfaces;
xadj = new idxtype[ne+1];
part = new idxtype[ne];
Array<int, 0> cnt(ne+1);
cnt = 0;
for ( int el=1; el <= ne; el++ )
{
Element volel = VolumeElement(el);
topology->GetElementFaces(el, elfaces);
for ( int i = 0; i < elfaces.Size(); i++ )
{
if ( facevolels1[elfaces[i]-1] == -1 )
facevolels1[elfaces[i]-1] = el;
else
{
facevolels2[elfaces[i]-1] = el;
cnt[facevolels1[elfaces[i]-1]-1]++;
cnt[facevolels2[elfaces[i]-1]-1]++;
}
}
}
xadj[0] = 0;
for ( int n = 1; n <= ne; n++ )
{
xadj[n] = idxtype(xadj[n-1] + cnt[n-1]);
}
adjacency = new idxtype[xadj[ne]];
cnt = 0;
for ( int face = 1; face <= nfaces; face++ )
{
int e1, e2;
e1 = facevolels1[face-1];
e2 = facevolels2[face-1];
if ( e2 == -1 ) continue;
adjacency[ xadj[e1-1] + cnt[e1-1] ] = e2-1;
adjacency[ xadj[e2-1] + cnt[e2-1] ] = e1-1;
cnt[e1-1]++;
cnt[e2-1]++;
}
for ( int el = 0; el < ne; el++ )
{
FlatArray<int> array ( cnt[el], &adjacency[ xadj[el] ] );
BubbleSort(array);
}
int timermetis = NgProfiler::CreateTimer ("Metis itself");
NgProfiler::StartTimer (timermetis);
metis :: METIS_PartGraphKway ( &ne, xadj, adjacency, v_weights, e_weights, &weightflag,
&numflag, &nparts, options, &edgecut, part );
NgProfiler::StopTimer (timermetis);
Array<int> nodesinpart(ntasks);
for ( int el = 1; el <= ne; el++ )
{
Element & volel = VolumeElement(el);
nodesinpart = 0;
VolumeElement(el).SetPartition(part[el-1 ] + 1);
}
/*
for ( int i=1; i<=ne; i++)
{
neloc[ VolumeElement(i).GetPartition() ] ++;
}
*/
delete [] xadj;
delete [] part;
delete [] adjacency;
}
void Mesh :: PartDualHybridMesh2D ( )
{
int ne = GetNSE();
int nv = GetNV();
Array<idxtype> xadj(ne+1);
Array<idxtype> adjacency(ne*4);
// first, build the vertex 2 element table:
Array<int, PointIndex::BASE> cnt(nv);
cnt = 0;
for (SurfaceElementIndex sei = 0; sei < GetNSE(); sei++)
for (int j = 0; j < (*this)[sei].GetNP(); j++)
cnt[ (*this)[sei][j] ] ++;
TABLE<SurfaceElementIndex, PointIndex::BASE> vert2els(cnt);
for (SurfaceElementIndex sei = 0; sei < GetNSE(); sei++)
for (int j = 0; j < (*this)[sei].GetNP(); j++)
vert2els.Add ((*this)[sei][j], sei);
// find all neighbour elements
int cntnb = 0;
Array<int> marks(ne); // to visit each neighbour just once
marks = -1;
for (SurfaceElementIndex sei = 0; sei < ne; sei++)
{
xadj[sei] = cntnb;
for (int j = 0; j < (*this)[sei].GetNP(); j++)
{
PointIndex vnr = (*this)[sei][j];
// all elements with at least one common vertex
for (int k = 0; k < vert2els[vnr].Size(); k++)
{
SurfaceElementIndex sei2 = vert2els[vnr][k];
if (sei == sei2) continue;
if (marks[sei2] == sei) continue;
// neighbour, if two common vertices
int common = 0;
for (int m1 = 0; m1 < (*this)[sei].GetNP(); m1++)
for (int m2 = 0; m2 < (*this)[sei2].GetNP(); m2++)
if ( (*this)[sei][m1] == (*this)[sei2][m2])
common++;
if (common >= 2)
{
marks[sei2] = sei; // mark as visited
adjacency[cntnb++] = sei2;
}
}
}
}
xadj[ne] = cntnb;
idxtype *v_weights = NULL, *e_weights = NULL;
int weightflag = 0;
int numflag = 0;
int nparts = ntasks - 1;
int options[5];
options[0] = 0;
int edgecut;
Array<idxtype> part(ne);
for ( int el = 0; el < ne; el++ )
BubbleSort (adjacency.Range (xadj[el], xadj[el+1]));
metis :: METIS_PartGraphKway ( &ne, &xadj[0], &adjacency[0], v_weights, e_weights, &weightflag,
&numflag, &nparts, options, &edgecut, &part[0] );
for (SurfaceElementIndex sei = 0; sei < ne; sei++)
(*this) [sei].SetPartition (part[sei]+1);
}
void Mesh :: SendMesh () const // Mesh * mastermesh, Array<int> & neloc ) const
{
const Mesh * mastermesh = this; // the original plan was different
PrintMessage ( 1, "Sending Mesh to local processors" );
int timer = NgProfiler::CreateTimer ("SendMesh");
int timer2 = NgProfiler::CreateTimer ("SM::Prepare Points");
int timer3 = NgProfiler::CreateTimer ("SM::Send Points");
int timer4 = NgProfiler::CreateTimer ("SM::Send Elements");
// int timer5 = NgProfiler::CreateTimer ("SM::Prepare Poins");
NgProfiler::RegionTimer reg(timer);
#ifdef SCALASCA
#pragma pomp inst begin(sendmesh)
#endif
NgProfiler::StartTimer (timer2);
clock_t starttime, endtime, soltime;
starttime = clock();
paralleltop -> SetNV ( GetNV() );
paralleltop -> SetNE ( GetNE() );
paralleltop -> SetNSegm ( GetNSeg() );
paralleltop -> SetNSE ( GetNSE() );
MPI_Request sendrequest[ntasks];
for ( int dest = 1; dest < ntasks; dest++)
MyMPI_Send ("mesh", dest);
// MPI_Barrier (MPI_COMM_WORLD);
Array<int> num_els_on_proc(ntasks);
num_els_on_proc = 0;
for (ElementIndex ei = 0; ei < mastermesh->GetNE(); ei++)
num_els_on_proc[(*this)[ei].GetPartition()]++;
int nel = GetNE();
int nv = GetNV();
int nedges = (GetTopology().GetNEdges());
int nfaces = GetTopology().GetNFaces();
int dim = dimension;
MPI_Bcast( &nel, 1, MPI_INT, 0, MPI_COMM_WORLD );
MPI_Bcast( &nv, 1, MPI_INT, 0, MPI_COMM_WORLD );
MPI_Bcast( &nedges, 1, MPI_INT, 0, MPI_COMM_WORLD );
MPI_Bcast( &nfaces, 1, MPI_INT, 0, MPI_COMM_WORLD );
MPI_Bcast( &dim, 1, MPI_INT, 0, MPI_COMM_WORLD );
for ( int dest = 1; dest < ntasks; dest++ )
MyMPI_Send (num_els_on_proc[dest], dest);
// get number of vertices for each processor
Array<int> elarraysize(ntasks);
Array<int> nelloc ( ntasks );
nelloc = 0;
elarraysize = 1;
PrintMessage ( 3, "Sending vertices");
TABLE<ElementIndex> els_of_proc (num_els_on_proc);
for (ElementIndex ei = 0; ei < mastermesh->GetNE(); ei++)
els_of_proc.Add ( (*this)[ei].GetPartition(), ei);
Array<int, PointIndex::BASE> vert_flag ( GetNV() );
Array<int> num_verts_on_proc (ntasks);
Array<int, PointIndex::BASE> num_procs_on_vert ( GetNV() );
num_verts_on_proc = 0;
num_procs_on_vert = 0;
vert_flag = -1;
for (int dest = 1; dest < ntasks; dest++)
{
FlatArray<ElementIndex> els = els_of_proc[dest];
for (int hi = 0; hi < els.Size(); hi++)
{
const Element & el = (*this) [ els[hi] ];
for (int i = 0; i < el.GetNP(); i++)
{
PointIndex epi = el[i];
if (vert_flag[epi] < dest)
{
vert_flag[epi] = dest;
num_verts_on_proc[dest]++;
num_procs_on_vert[epi]++;
paralleltop -> SetDistantPNum ( dest, epi, num_verts_on_proc[dest]);
}
}
elarraysize[dest] += 3 + el.GetNP();
nelloc[dest] ++;
paralleltop -> SetDistantEl ( dest, els[hi]+1, nelloc[dest] );
}
}
TABLE<PointIndex> verts_of_proc (num_verts_on_proc);
TABLE<int, PointIndex::BASE> procs_of_vert (num_procs_on_vert);
TABLE<int, PointIndex::BASE> loc_num_of_vert (num_procs_on_vert);
vert_flag = -1;
for (int dest = 1; dest < ntasks; dest++)
{
FlatArray<ElementIndex> els = els_of_proc[dest];
for (int hi = 0; hi < els.Size(); hi++)
{
const Element & el = (*mastermesh) [ els[hi] ];
for (int i = 0; i < el.GetNP(); i++)
{
PointIndex epi = el.PNum(i+1);
if (vert_flag[epi] < dest)
{
vert_flag[epi] = dest;
procs_of_vert.Add (epi, dest);
}
}
}
}
for (int vert = 1; vert <= mastermesh->GetNP(); vert++ )
{
FlatArray<int> procs = procs_of_vert[vert];
for (int j = 0; j < procs.Size(); j++)
{
int dest = procs[j];
verts_of_proc.Add (dest, vert);
loc_num_of_vert.Add (vert, verts_of_proc[dest].Size());
}
}
Array<int> nvi5(ntasks);
for (int i = 0; i < ntasks; i++)
nvi5[i] = 5 * num_verts_on_proc[i];
TABLE<double> pointarrays(nvi5);
NgProfiler::StopTimer (timer2);
NgProfiler::StartTimer (timer3);
for (int dest = 1; dest < ntasks; dest++)
{
FlatArray<PointIndex> verts = verts_of_proc[dest];
for ( int j = 0, ii = 0; j < verts.Size(); j++)
{
const MeshPoint & hp = mastermesh -> Point (verts[j]);
pointarrays.Add (dest, double(verts[j]));
pointarrays.Add (dest, hp(0));
pointarrays.Add (dest, hp(1));
pointarrays.Add (dest, hp(2));
pointarrays.Add (dest, hp.Singularity());
}
MyMPI_Send ( "vertex", dest );
MyMPI_ISend ( pointarrays[dest], dest, sendrequest[dest] );
MPI_Request_free (&sendrequest[dest]);
}
NgProfiler::StopTimer (timer3);
NgProfiler::StartTimer (timer4);
Array<int> num_distpnums(ntasks);
num_distpnums = 0;
for (int vert = 1; vert <= mastermesh -> GetNP(); vert++)
{
FlatArray<int> procs = procs_of_vert[vert];
for (int j = 0; j < procs.Size(); j++)
num_distpnums[procs[j]] += 3 * (procs.Size()-1);
}
TABLE<int> distpnums (num_distpnums);
for (int vert = 1; vert <= mastermesh -> GetNP(); vert++)
{
FlatArray<int> procs = procs_of_vert[vert];
for (int j = 0; j < procs.Size(); j++)
for (int k = 0; k < procs.Size(); k++)
if (j != k)
{
distpnums.Add (procs[j], loc_num_of_vert[vert][j]);
distpnums.Add (procs[j], procs_of_vert[vert][k]);
distpnums.Add (procs[j], loc_num_of_vert[vert][k]);
}
}
for ( int dest = 1; dest < ntasks; dest ++ )
{
MyMPI_ISend ( distpnums[dest], dest, sendrequest[dest] );
MPI_Request_free (&sendrequest[dest]);
}
endtime = clock();
(*testout) << "Sending Time verts = " << double(endtime - starttime)/CLOCKS_PER_SEC << endl;
PrintMessage ( 3, "Sending elements" );
TABLE<int> elementarrays(elarraysize);
starttime = clock();
// for ( int dest = 1; dest < ntasks; dest ++ )
// MyMPI_Send ( "volumeelements", dest);
for ( int dest = 1; dest < ntasks; dest++ )
elementarrays.Add (dest, nelloc[dest]);
for ( int ei = 1; ei <= mastermesh->GetNE(); ei++)
{
const Element & el = mastermesh -> VolumeElement (ei);
int dest = el.GetPartition();
if ( dest > 0 )
{
// send volume element
elementarrays.Add (dest, ei); //
elementarrays.Add (dest, el.GetIndex());
elementarrays.Add (dest, el.GetNP());
for ( int ii=0; ii<el.GetNP(); ii++)
elementarrays.Add (dest, el[ii]);
}
}
for ( int dest = 1; dest < ntasks; dest ++ )
{
MyMPI_Send ( "volumeelements", dest);
MyMPI_ISend ( elementarrays[dest], dest, sendrequest[dest] );
}
// for (int dest = 1; dest < ntasks; dest++)
// MPI_Request_free (&sendrequest[dest]);
NgProfiler::StopTimer (timer4);
endtime = clock();
(*testout) << "Sending Time els = " << double(endtime - starttime)/CLOCKS_PER_SEC << endl;
starttime = clock();
PrintMessage ( 3, "Sending Face Descriptors" );
Array<double> double6(6);
for ( int dest = 1; dest < ntasks; dest++)
for ( int fdi = 1; fdi <= mastermesh->GetNFD(); fdi++)
{
MyMPI_Send("facedescriptor", dest);
double6[0] = GetFaceDescriptor(fdi).SurfNr();
double6[1] = GetFaceDescriptor(fdi).DomainIn();
double6[2] = GetFaceDescriptor(fdi).DomainOut();
double6[3] = GetFaceDescriptor(fdi).BCProperty();
double6[4] = GetFaceDescriptor(fdi).domin_singular;
double6[5] = GetFaceDescriptor(fdi).domout_singular;
MyMPI_Send ( double6, dest);
}
endtime = clock();
(*testout) << "Sending Time fdi = " << double(endtime - starttime)/CLOCKS_PER_SEC << endl;
starttime = clock();
// hasglobalsurf( edgenr * ntasks + dest ) .... global edge edgenr is in mesh at dest
// BitArray hasglobaledge;
// hasglobaledge.SetSize(ntasks * nedglob);
// hasglobaledge.Clear();
// cout << "mmf " << mastermesh->GetTopology().GetNFaces() << endl;
// determine sizes of local surface element arrays
PrintMessage ( 3, "Sending Surface elements" );
Array <int> nlocsel(ntasks), bufsize ( ntasks), seli(ntasks);
for ( int i = 0; i < ntasks; i++)
{
nlocsel[i] = 0;
bufsize[i] = 1;
seli[i] = 1;
}
for ( int sei = 1; sei <= mastermesh -> GetNSE(); sei ++ )
{
int ei1, ei2;
mastermesh -> GetTopology().GetSurface2VolumeElement (sei, ei1, ei2);
const Element2d & sel = mastermesh -> SurfaceElement (sei);
int dest;
// first element
for (int j = 0; j < 2; j++)
{
int ei = (j == 0) ? ei1 : ei2;
if ( ei > 0 && ei <= mastermesh->GetNE() )
{
const Element & el = mastermesh -> VolumeElement (ei);
dest = el.GetPartition();
nlocsel[dest] ++;
bufsize[dest] += 4 + 2*sel.GetNP();
}
}
}
int ** selbuf = 0;
selbuf = new int*[ntasks];
for ( int i = 0; i < ntasks; i++)
if ( bufsize[i] > 0 )
{*(selbuf+i) = new int[bufsize[i]];}
else
selbuf[i] = 0;
Array<int> nselloc (ntasks);
nselloc = 0;
for ( int dest = 1; dest < ntasks; dest++ )
{
MyMPI_Send ( "surfaceelementsgi", dest);
selbuf[dest][0] = nlocsel[dest];
}
for ( int sei = 1; sei <= mastermesh -> GetNSE(); sei ++ )
{
int ei1, ei2;
mastermesh -> GetTopology().GetSurface2VolumeElement (sei, ei1, ei2);
const Element2d & sel = mastermesh -> SurfaceElement (sei);
int dest;
int isghost = 0;
for (int j = 0; j < 2; j++)
{
int ei = (j == 0) ? ei1 : ei2;
if ( ei > 0 && ei <= mastermesh->GetNE() )
{
const Element & el = mastermesh -> VolumeElement (ei);
dest = el.GetPartition();
if (dest > 0)
{
// send:
// sei
// faceind
// nep
// tri.pnums, tri.geominfopi.trignums
selbuf[dest][seli[dest]++] = sei;
selbuf[dest][seli[dest]++] = sel.GetIndex();
selbuf[dest][seli[dest]++] = isghost;
selbuf[dest][seli[dest]++] = sel.GetNP();
for ( int ii = 1; ii <= sel.GetNP(); ii++)
{
selbuf[dest][seli[dest]++] = sel.PNum(ii);
selbuf[dest][seli[dest]++] = sel.GeomInfoPi(ii).trignum;
}
nselloc[dest] ++;
paralleltop -> SetDistantSurfEl ( dest, sei, nselloc[dest] );
isghost = 1;
}
}
}
}
for ( int dest = 1; dest < ntasks; dest++)
MyMPI_Send( selbuf[dest], bufsize[dest], dest);
for ( int dest = 0; dest < ntasks; dest++ )
{
if (selbuf[dest])
delete [] *(selbuf+dest);
}
delete [] selbuf;
endtime = clock();
(*testout) << "Sending Time surfels = " << double(endtime - starttime)/CLOCKS_PER_SEC << endl;
starttime = clock();
PrintMessage ( 3, "Sending Edge Segments");
for ( int dest = 1; dest < ntasks; dest++ )
MyMPI_Send ( "edgesegmentsgi", dest);
Array <int> nlocseg(ntasks), segi(ntasks);
for ( int i = 0; i < ntasks; i++)
{
nlocseg[i] = 0;
bufsize[i] = 0;
segi[i] = 0;
}
for ( int segi = 1; segi <= mastermesh -> GetNSeg(); segi ++ )
{
Array<int> volels;
const MeshTopology & topol = mastermesh -> GetTopology();
topol . GetSegmentVolumeElements ( segi, volels );
const Segment & segm = mastermesh -> LineSegment (segi);
int dest;
for (int j = 0; j < volels.Size(); j++)
{
int ei = volels[j];
int dest;
if ( ei > 0 && ei <= mastermesh->GetNE() )
{
const Element & el = mastermesh -> VolumeElement (ei);
dest = el.GetPartition();
nlocseg[dest] ++;
bufsize[dest] += 14;
}
}
}
double ** segmbuf;
segmbuf = new double*[ntasks];
for ( int i = 0; i < ntasks; i++)
if ( bufsize[i] > 0 )
segmbuf[i] = new double[bufsize[i]];
else
segmbuf[i] = 0;
// cout << "mastermesh " << mastermesh -> GetNSeg() << " lineseg " << mastermesh -> LineSegment (1) << endl;
for ( int ls=1; ls <= mastermesh -> GetNSeg(); ls++)
{
Array<int> volels;
mastermesh -> GetTopology().GetSegmentVolumeElements ( ls, volels );
const Segment & seg = mastermesh -> LineSegment (ls);
int dest;
for (int j = 0; j < volels.Size(); j++)
{
int ei = volels[j];
int dest;
if ( ei > 0 && ei <= mastermesh->GetNE() )
{
const Element & el = mastermesh -> VolumeElement (ei);
dest = el.GetPartition();
if ( dest > 0 )
{
segmbuf[dest][segi[dest]++] = ls;
segmbuf[dest][segi[dest]++] = seg.si;
segmbuf[dest][segi[dest]++] = seg.p1;
segmbuf[dest][segi[dest]++] = seg.p2;
segmbuf[dest][segi[dest]++] = seg.geominfo[0].trignum;
segmbuf[dest][segi[dest]++] = seg.geominfo[1].trignum;
segmbuf[dest][segi[dest]++] = seg.surfnr1;
segmbuf[dest][segi[dest]++] = seg.surfnr2;
segmbuf[dest][segi[dest]++] = seg.edgenr;
segmbuf[dest][segi[dest]++] = seg.epgeominfo[0].dist;
segmbuf[dest][segi[dest]++] = seg.epgeominfo[1].edgenr;
segmbuf[dest][segi[dest]++] = seg.epgeominfo[1].dist;
segmbuf[dest][segi[dest]++] = seg.singedge_right;
segmbuf[dest][segi[dest]++] = seg.singedge_left;
}
paralleltop -> SetDistantSegm ( dest, ls, int ( segi[dest] / 14 ) );
}
}
}
PrintMessage ( 3, "sending segments" );
for ( int dest = 1; dest < ntasks; dest++)
{
MyMPI_Send( segmbuf[dest], bufsize[dest], dest);
}
for ( int dest = 0; dest < ntasks; dest++ )
{
if ( segmbuf[dest] )
delete [] segmbuf[dest];
}
delete [] segmbuf;
PrintMessage ( 3, "segments sent");
endtime = clock();
(*testout) << "Sending Time segments = " << double(endtime - starttime)/CLOCKS_PER_SEC << endl;
starttime = clock();
for ( int dest = 1; dest < ntasks; dest++ )
MyMPI_Send("endmesh", dest);
for ( int dest = 1; dest < ntasks; dest ++ )
{
MPI_Status status;
MPI_Wait (&sendrequest[dest], &status);
}
#ifdef SCALASCA
#pragma pomp inst end(sendmesh)
#endif
}
void Mesh :: UpdateOverlap()
{
(*testout) << "UPDATE OVERLAP" << endl;
Array<int> * globelnums;
#ifdef SCALASCA
#pragma pomp inst begin(updateoverlap)
#endif
paralleltop->IncreaseOverlap();
if ( id > 0 )
{
int nvglob = paralleltop->GetNVGlob ();
int nelglob = paralleltop->GetNEGlob();
int nv = GetNV();
int ned = topology -> GetNEdges();
int nfa = topology -> GetNFaces();
int nel = GetNE();
Array<int,PointIndex::BASE> glob2loc_vert(nvglob);
glob2loc_vert = -1;
for ( int locv = 1; locv <= GetNV(); locv++)
{
int globv = paralleltop->GetLoc2Glob_Vert(locv);
glob2loc_vert[globv] = locv;
}
BitArray addedpoint ( paralleltop -> GetNVGlob () );
BitArray addedel ( paralleltop -> GetNEGlob () );
addedpoint.Clear();
addedel.Clear();
Array<int> distvert(ntasks), distel(ntasks), nsenddistel(ntasks);
nsenddistel = 0;
MyMPI_Allgather (GetNV(), distvert.Range(1, ntasks), MPI_HIGHORDER_COMM);
MyMPI_Allgather (GetNE(), distel.Range(1, ntasks), MPI_HIGHORDER_COMM);
BitArray appendedpoint ( GetNP() * ntasks );
appendedpoint.Clear();
TABLE<double> sendpoints(ntasks);
TABLE<int> sendelements(ntasks);
TABLE<int> sendsel(ntasks);
/*
TABLE<int> cluster_neighbors(nv+ned+nfa+nel);
for ( int node = 1; node <= nv+ned+nfa+nel; node++ )
{
int cluster_rep;
cluster_rep = clusters->GetVertexRepresentant(node);
if ( node == cluster_rep ) continue;
Array<int> dests;
int nneigh = 0;
if ( node - GetNV() <= 0 ) // cluster representant is vertex
{
int vert = node;
if ( paralleltop -> IsExchangeVert( vert ) )
{
nneigh = paralleltop -> GetNDistantPNums(vert);
dests.SetSize(2*nneigh);
paralleltop -> GetDistantPNums ( vert, &dests[0] );
}
}
else if ( node - GetNV() - ned <= 0 ) // cluster representant is edge
{
int edge = node - GetNV();
if ( paralleltop -> IsExchangeEdge( edge ) )
{
nneigh = paralleltop -> GetNDistantEdgeNums(edge);
dests.SetSize(2*nneigh);
paralleltop -> GetDistantEdgeNums ( edge, &dests[0] );
}
}
else if ( node - GetNV() - ned - nfa <= 0 ) // cluster representant is face
{
int face = node - GetNV() - ned;
if ( paralleltop -> IsExchangeFace( face ) )
{
nneigh = paralleltop -> GetNDistantFaceNums(face);
dests.SetSize(2*nneigh);
paralleltop -> GetDistantFaceNums ( face, &dests[0] );
}
}
else // cluster representant is element
{
int el = node - GetNV() - ned - nfa;
if ( paralleltop -> IsExchangeElement( el ) )
{
nneigh = paralleltop -> GetNDistantElNums(el);
dests.SetSize(2*nneigh);
paralleltop -> GetDistantElNums ( el, &dests[0] );
}
}
for ( int j = 1; j < nneigh; j++ )
if ( !cluster_neighbors[cluster_rep].Contains ( dests[2*j] ) )
{
cluster_neighbors.Add( cluster_rep-1, dests[2*j] );
}
}
*/
for ( int i = 0; i < ntasks; i++ )
{
sendelements.Add (i, 0);
sendsel.Add(i, 0);
}
Array<int> nsentsel (ntasks);
nsentsel = 0;
for ( int seli = 1; seli <= GetNSE(); seli++ )
{
const Element2d & sel = SurfaceElement(seli);
int selnp = sel.GetNP();
Array<int> vert (selnp);
Array<int> alldests (0), dests;
bool isparsel = false;
for ( int i = 0; i < selnp; i++ )
{
vert[i] = sel.PNum(i+1);
if ( paralleltop -> IsExchangeVert ( vert[i] ) )
{
isparsel = true;
paralleltop -> GetVertNeighbours ( vert[i], dests );
for ( int j = 0; j < dests.Size(); j++ )
if ( !alldests.Contains ( dests[j] ) )
alldests.Append( dests[j] );
}
}
/*
int face = topology->GetSurfaceElementFace(seli);
int cluster_rep = clusters->GetFaceRepresentant(face);
if ( cluster_neighbors[cluster_rep-1].Size() > 0 )
{
isparsel = true;
for ( int j = 0; j < cluster_neighbors[cluster_rep-1].Size(); j++ )
if ( !alldests.Contains ( cluster_neighbors[cluster_rep-1][j] ) )
alldests.Append( cluster_neighbors[cluster_rep-1][j] );
}
*/
if ( !isparsel ) continue;
for ( int i = 0; i < alldests.Size(); i ++ )
{
// send the surface element to all distant procs:
// loc number,
// number of points
// global vert numbers
// surface_element_index
int dest = alldests[i];
// ***************** MISSING id = 0
if ( dest == 0 ) continue;
sendsel.Add (dest, seli);
sendsel.Add (dest, selnp);
for ( int ii=0; ii<selnp; ii++)
{
sendsel.Add (dest, paralleltop -> GetLoc2Glob_Vert (vert[ii]) );
}
sendsel.Add (dest, sel.GetIndex() );
nsentsel[dest] ++;
}
}
for ( int dest = 1; dest < ntasks; dest++ )
sendsel[dest][0] = nsentsel[dest];
for ( int eli = 1; eli <= GetNE(); eli++ )
{
const Element & el = VolumeElement(eli);
int elnp = el.GetNP();
Array<int> vert (elnp);
Array<int> alldests (0), dests;
for ( int i = 0; i < elnp; i++ )
{
vert[i] = el.PNum(i+1);
if ( paralleltop -> IsExchangeVert ( vert[i] ) )
{
paralleltop -> GetVertNeighbours ( vert[i], dests );
for ( int j = 0; j < dests.Size(); j++ )
if ( !alldests.Contains ( dests[j] ) )
{
alldests.Append( dests[j] );
paralleltop->SetExchangeElement ( dests[j], eli );
}
paralleltop->SetExchangeElement ( eli );
}
}
/*
int cluster_rep = clusters->GetElementRepresentant(eli);
if ( cluster_neighbors[cluster_rep-1].Size() > 0 )
{
for ( int j = 0; j < cluster_neighbors[cluster_rep-1].Size(); j++ )
if ( !alldests.Contains ( cluster_neighbors[cluster_rep-1][j] ) )
{
alldests.Append( cluster_neighbors[cluster_rep-1][j] );
paralleltop->SetExchangeElement ( cluster_neighbors[cluster_rep-1][j], eli );
}
paralleltop->SetExchangeElement ( eli );
}
*/
}
for ( int eli = 1; eli <= GetNE(); eli++ )
{
const Element & el = VolumeElement(eli);
int elnp = el.GetNP();
Array<int> vert (elnp);
// append to point list:
// local pnum
// global pnum
// point coordinates
Array<Point3d> points(elnp);
for ( int i = 0; i < elnp; i++ )
{
vert[i] = el.PNum(i+1);
points[i] = Point(vert[i]);
Array<int> knowndests;
// send point to all dests which get the volume element
for ( int dest = 0; dest < ntasks; dest ++ )
{
// nur die neuen verts
if ( !paralleltop -> IsExchangeElement ( dest, eli ) ) continue;
// jeder vertex nur ein mal
if ( appendedpoint.Test( (vert[i]-1) * ntasks + dest ) ) continue;
appendedpoint.Set( (vert[i]-1) * ntasks + dest );
paralleltop -> SetExchangeVert (dest, vert[i]);
paralleltop -> SetExchangeVert ( vert[i] );
// append vertex to be sent
// loc pnum
// glob pnum
// coords
// local pnum
sendpoints.Add (dest, vert[i]);
// global pnum
sendpoints.Add (dest, paralleltop -> GetLoc2Glob_Vert ( vert[i] ) );
// coordinates
sendpoints.Add (dest, points[i].X() );
sendpoints.Add (dest, points[i].Y() );
sendpoints.Add (dest, points[i].Z() );
}
}
for ( int dest = 1; dest < ntasks; dest ++ )
{
// send the volume element to all distant procs:
// loc number,
// glob number
// number of points
// glob vertices
// element_index
if ( !paralleltop -> IsExchangeElement ( dest, eli ) ) continue;
// loc number
sendelements.Add (dest, eli);
// glob number
sendelements.Add (dest, paralleltop -> GetLoc2Glob_VolEl(eli));
sendelements.Add (dest, elnp);
for ( int j = 0; j < elnp; j++ )
sendelements.Add (dest, paralleltop -> GetLoc2Glob_Vert(vert[j]) );
sendelements.Add (dest, el.GetIndex() );
distel[dest]++;
nsenddistel[dest] ++;
paralleltop -> SetDistantEl ( dest, eli, -1);// distel[dest] );
}
}
for ( int dest = 1; dest < ntasks; dest++ )
if ( dest != id )
sendelements[dest][0] = nsenddistel[dest];
// find parallel surface elements, if there, append to sendsel - list
distel = 0;
// sizes of sendpoints, sendelements, sendsels
Array<int> sendsize_pts(ntasks), recvsize_pts(ntasks);
Array<int> sendsize_els(ntasks), recvsize_els(ntasks);
Array<int> sendsize_sels(ntasks), recvsize_sels(ntasks);
for (int i = 0; i < ntasks; i++)
{
sendsize_pts[i] = sendpoints[i].Size();
sendsize_els[i] = sendelements[i].Size();
sendsize_sels[i] = sendsel[i].Size();
}
MyMPI_Alltoall (sendsize_pts.Range(1, ntasks), recvsize_pts.Range(1, ntasks), MPI_HIGHORDER_COMM);
MyMPI_Alltoall (sendsize_els.Range(1, ntasks), recvsize_els.Range(1, ntasks), MPI_HIGHORDER_COMM);
MyMPI_Alltoall (sendsize_sels.Range(1, ntasks), recvsize_sels.Range(1, ntasks), MPI_HIGHORDER_COMM);
recvsize_pts[0] = 0;
recvsize_els[0] = 0;
recvsize_sels[0] = 0;
TABLE<double> recvpoints(recvsize_pts);
TABLE<int> recvelements(recvsize_els);
TABLE<int> recvsel(recvsize_sels);
recvpoints.SetElementSizesToMaxSizes ();
recvelements.SetElementSizesToMaxSizes ();
recvsel.SetElementSizesToMaxSizes ();
/*
Array<MPI_Request> sendrequest(3*ntasks), recvrequest(3*ntasks);
Array<MPI_Status> status(3*ntasks);
for ( int proc = 1; proc < ntasks; proc++)
{
MyMPI_ISend ( sendpoints[proc], proc, sendrequest[proc] );
MyMPI_IRecv ( recvpoints[proc], proc, recvrequest[proc] );
}
for ( int proc = 1; proc < ntasks; proc++)
{
MPI_Wait(&sendrequest[proc], &status[proc]);
MPI_Wait(&recvrequest[proc], &status[proc]);
MyMPI_ISend ( sendelements[proc], proc, sendrequest[proc+1]);
MyMPI_IRecv ( recvelements[proc], proc, recvrequest[proc+1]);
}
for ( int proc = 1; proc < ntasks; proc++)
{
MPI_Wait(&sendrequest[proc+1], &status[proc+1]);
MPI_Wait(&recvrequest[proc+1], &status[proc+1]);
MyMPI_ISend ( sendsel[proc], proc, sendrequest[proc+2] );
MyMPI_IRecv ( recvsel[proc], proc, recvrequest[proc+2] );
}
for ( int proc = 1; proc < ntasks; proc++)
{
MPI_Wait(&sendrequest[proc+2], &status[proc+2]);
MPI_Wait(&recvrequest[proc+2], &status[proc+2]);
}
*/
Array<MPI_Request> requests;
for ( int proc = 1; proc < ntasks; proc++)
{
requests.Append (MyMPI_ISend ( sendpoints[proc], proc ));
requests.Append (MyMPI_IRecv ( recvpoints[proc], proc ));
}
for ( int proc = 1; proc < ntasks; proc++)
{
requests.Append (MyMPI_ISend ( sendelements[proc], proc ));
requests.Append (MyMPI_IRecv ( recvelements[proc], proc ));
}
for ( int proc = 1; proc < ntasks; proc++)
{
requests.Append (MyMPI_ISend ( sendsel[proc], proc ));
requests.Append (MyMPI_IRecv ( recvsel[proc], proc ));
}
MPI_Status stat;
for (int i = 0; i < requests.Size(); i++)
MPI_Wait (&requests[i], &stat);
Array<int> * distpnum2parpnum;
distpnum2parpnum = new Array<int> [2];
distpnum2parpnum[0].SetSize(0);
distpnum2parpnum[1].SetSize(0);
Array<int> firstdistpnum (ntasks);
for ( int sender = 1; sender < ntasks; sender++)
{
firstdistpnum[sender] = distpnum2parpnum[0].Size();
if ( sender == id ) continue;
int ii = 0;
// receiving points
// dist pnum
// glob pnum
// coords
int numrecvpts = int ( recvpoints[sender].Size() / 5 );
paralleltop -> SetNV ( GetNV() + numrecvpts );
int expectnp = GetNV () + numrecvpts;
// received points
while ( ii < recvpoints[sender].Size() )
{
int distpnum = int ( (recvpoints[sender])[ii++] );
int globpnum = int ( (recvpoints[sender])[ii++] );
Point3d point;
point.X() = (recvpoints[sender])[ii++];
point.Y() = (recvpoints[sender])[ii++];
point.Z() = (recvpoints[sender])[ii++];
// append point as ghost
// if not already there
int pnum= glob2loc_vert[globpnum];//paralleltop -> Glob2Loc_Vert ( globpnum );
if ( pnum <= 0 )
{
pnum = AddPoint ( point, true );
}
paralleltop -> SetDistantPNum ( 0, pnum, globpnum );
glob2loc_vert[globpnum] = pnum;
paralleltop -> SetDistantPNum ( sender, pnum, distpnum );
paralleltop -> SetExchangeVert ( pnum );
}
ii = 0;
int recvnel = (recvelements[sender])[ii++];
paralleltop -> SetNE ( recvnel + GetNE() );
while ( ii < recvelements[sender].Size() )
{
// receive list:
// distant number,
// glob number
// number of points
// glob vertices
// element_index
int distelnum = (recvelements[sender])[ii++];
int globelnum = (recvelements[sender])[ii++] ;
int elnp = (recvelements[sender])[ii++] ;
Array<int> pnums(elnp), globpnums(elnp);
// append volel
ELEMENT_TYPE eltype;
switch ( elnp )
{
case 4: eltype = TET; break;
case 5: eltype = PYRAMID; break;
case 6: eltype = PRISM; break;
case 8: eltype = HEX; break;
}
Element el ( eltype ) ;
for ( int i = 0; i < elnp; i++ )
{
globpnums[i] = int ( (recvelements[sender])[ii++] );
pnums[i] = glob2loc_vert[globpnums[i]]; //paralleltop -> Glob2Loc_Vert(globpnums[i]);
}
el.SetIndex ( (recvelements[sender])[ii++] );
el.SetGhost ( 1 );
for ( int i = 0; i < elnp; i++)
{
(int&) el[i] = pnums[i];
}
int eli = AddVolumeElement (el) + 1;
paralleltop -> SetDistantEl ( sender, eli, distelnum);
paralleltop -> SetDistantEl ( 0, eli, globelnum );
paralleltop -> SetExchangeElement ( eli );
}
ii = 0;
int nrecvsendsel = 0;
if ( recvsel[sender].Size() > 0 )
nrecvsendsel = (recvsel[sender])[ii++];
paralleltop -> SetNSE ( nrecvsendsel + GetNSE() );
while ( ii < recvsel[sender] . Size() )
{
// receive list:
// distant number,
// number of points
// global vert numbers
// surface_element_index
int distselnum = (recvsel[sender])[ii++];
int selnp = (recvsel[sender])[ii++] ;
Array<int> globpnums(selnp);
Array<int> pnums(selnp);
// append volel
ELEMENT_TYPE eltype;
switch ( selnp )
{
case 4: eltype = QUAD; break;
case 3: eltype = TRIG; break;
}
Element2d sel ( eltype ) ;
for ( int i = 0; i < selnp; i++ )
{
globpnums[i] = int ( (recvsel[sender])[ii++] );
pnums[i] = glob2loc_vert[globpnums[i]];
}
sel.SetIndex ( (recvsel[sender])[ii++] );
sel.SetGhost ( 1 );
for ( int i = 0; i < selnp; i++)
{
(int&) sel[i] = pnums[i];
}
int seli = AddSurfaceElement (sel);
}
}
delete [] distpnum2parpnum;
}
globelnums = new Array<int>;
if ( id == 0 )
{
for ( int dest = 1; dest < ntasks; dest++)
{
MyMPI_Recv ( *globelnums, dest );
for ( int i = 0; i < globelnums->Size(); i++ )
{
paralleltop -> SetDistantEl ( dest, (*globelnums)[i],i+1 );
paralleltop -> SetExchangeElement ( dest, (*globelnums)[i] );
}
}
}
else
{
globelnums -> SetSize(GetNE());
for ( int i = 0; i < GetNE(); i++ )
{
(*globelnums)[i] = paralleltop -> GetLoc2Glob_VolEl ( i+1 );
}
MyMPI_Send ( *globelnums, 0 );
}
delete globelnums;
// send which elements are where
topology -> Update();
// edge, facenums have probably changed as elements were added
// paralleltop has to be updated
paralleltop -> UpdateExchangeElements();
paralleltop -> UpdateCoarseGridOverlap();
//paralleltop -> UpdateTopology();
// *testout << "############################################" << endl << endl;
// paralleltop -> Print();
clusters -> Update();
;
#ifdef SCALASCA
#pragma pomp inst end(updateoverlap)
#endif
}
}
#endif
Reference in New Issue Copy Permalink