portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-10 07:55:35 +00:00
Code Releases Activity

Staggered multinode block cg debugged. Missing global sum.

Browse Source 
Code stalls and resumes on KNL at cambridge. Curious.

CG iterations 23ms each, then 3200 ms pauses. Mean bandwidth reports
as 200MB/s. Comms dominant in the report. However, the time behaviour suggests it
is *bursty*.... Could be swap to disk?
This commit is contained in:
Azusa Yamaguchi 2017-08-23 15:07:18 +01:00
parent 383ca7d392
commit d9cd4f0273
5 changed files with 110 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
lib/algorithms/iterative/BlockConjugateGradient.h
View File

@ -199,7 +199,12 @@ void BlockCGrQsolve(LinearOperatorBase<Field> &Linop, const Field &B, Field &X)
Linop.HermOp(X, AD);
tmp = B - AD;
//std::cout << GridLogMessage << " initial tmp " << norm2(tmp)<< std::endl;
ThinQRfact (m_rr, m_C, m_Cinv, Q, tmp);
//std::cout << GridLogMessage << " initial Q " << norm2(Q)<< std::endl;
//std::cout << GridLogMessage << " m_rr " << m_rr<<std::endl;
//std::cout << GridLogMessage << " m_C " << m_C<<std::endl;
//std::cout << GridLogMessage << " m_Cinv " << m_Cinv<<std::endl;
D=Q;
std::cout << GridLogMessage<<"BlockCGrQ computed initial residual and QR fact " <<std::endl;
@ -221,13 +226,15 @@ void BlockCGrQsolve(LinearOperatorBase<Field> &Linop, const Field &B, Field &X)
MatrixTimer.Start();
Linop.HermOp(D, Z);
MatrixTimer.Stop();
//std::cout << GridLogMessage << " norm2 Z " <<norm2(Z)<<std::endl;
//4. M = [D^dag Z]^{-1}
sliceInnerTimer.Start();
sliceInnerProductMatrix(m_DZ,D,Z,Orthog);
sliceInnerTimer.Stop();
m_M = m_DZ.inverse();
//std::cout << GridLogMessage << " m_DZ " <<m_DZ<<std::endl;
//5. X = X + D MC
m_tmp = m_M * m_C;
sliceMaddTimer.Start();

38
lib/lattice/Lattice_reduction.h
View File

@ -369,6 +369,7 @@ static void sliceMaddVector(Lattice<vobj> &R,std::vector<RealD> &a,const Lattice
}
};
/*
inline GridBase *makeSubSliceGrid(const GridBase *BlockSolverGrid,int Orthog)
{
int NN = BlockSolverGrid->_ndimension;
@ -387,6 +388,7 @@ inline GridBase *makeSubSliceGrid(const GridBase *BlockSolverGrid,int Or
}
return (GridBase *)new GridCartesian(latt_phys,simd_phys,mpi_phys);
}
*/
template<class vobj>
static void sliceMaddMatrix (Lattice<vobj> &R,Eigen::MatrixXcd &aa,const Lattice<vobj> &X,const Lattice<vobj> &Y,int Orthog,RealD scale=1.0)
@ -398,14 +400,15 @@ static void sliceMaddMatrix (Lattice<vobj> &R,Eigen::MatrixXcd &aa,const Lattice
int Nblock = X._grid->GlobalDimensions()[Orthog];
GridBase *FullGrid = X._grid;
GridBase *SliceGrid = makeSubSliceGrid(FullGrid,Orthog);
// GridBase *SliceGrid = makeSubSliceGrid(FullGrid,Orthog);
Lattice<vobj> Xslice(SliceGrid);
Lattice<vobj> Rslice(SliceGrid);
// Lattice<vobj> Xslice(SliceGrid);
// Lattice<vobj> Rslice(SliceGrid);
assert( FullGrid->_simd_layout[Orthog]==1);
int nh = FullGrid->_ndimension;
int nl = SliceGrid->_ndimension;
// int nl = SliceGrid->_ndimension;
int nl = nh-1;
//FIXME package in a convenient iterator
//Should loop over a plane orthogonal to direction "Orthog"
@ -448,14 +451,14 @@ static void sliceMulMatrix (Lattice<vobj> &R,Eigen::MatrixXcd &aa,const Lattice<
int Nblock = X._grid->GlobalDimensions()[Orthog];
GridBase *FullGrid = X._grid;
GridBase *SliceGrid = makeSubSliceGrid(FullGrid,Orthog);
Lattice<vobj> Xslice(SliceGrid);
Lattice<vobj> Rslice(SliceGrid);
// GridBase *SliceGrid = makeSubSliceGrid(FullGrid,Orthog);
// Lattice<vobj> Xslice(SliceGrid);
// Lattice<vobj> Rslice(SliceGrid);
assert( FullGrid->_simd_layout[Orthog]==1);
int nh = FullGrid->_ndimension;
int nl = SliceGrid->_ndimension;
// int nl = SliceGrid->_ndimension;
int nl=1;
//FIXME package in a convenient iterator
//Should loop over a plane orthogonal to direction "Orthog"
@ -498,18 +501,19 @@ static void sliceInnerProductMatrix( Eigen::MatrixXcd &mat, const Lattice<vobj>
typedef typename vobj::vector_type vector_type;
GridBase *FullGrid = lhs._grid;
GridBase *SliceGrid = makeSubSliceGrid(FullGrid,Orthog);
// GridBase *SliceGrid = makeSubSliceGrid(FullGrid,Orthog);
int Nblock = FullGrid->GlobalDimensions()[Orthog];
Lattice<vobj> Lslice(SliceGrid);
Lattice<vobj> Rslice(SliceGrid);
// Lattice<vobj> Lslice(SliceGrid);
// Lattice<vobj> Rslice(SliceGrid);
mat = Eigen::MatrixXcd::Zero(Nblock,Nblock);
assert( FullGrid->_simd_layout[Orthog]==1);
int nh = FullGrid->_ndimension;
int nl = SliceGrid->_ndimension;
// int nl = SliceGrid->_ndimension;
int nl = nh-1;
//FIXME package in a convenient iterator
//Should loop over a plane orthogonal to direction "Orthog"
@ -550,6 +554,14 @@ static void sliceInnerProductMatrix( Eigen::MatrixXcd &mat, const Lattice<vobj>
mat += mat_thread;
}
}
for(int i=0;i<Nblock;i++){
for(int j=0;j<Nblock;j++){
ComplexD sum = mat(i,j);
FullGrid->GlobalSum(sum);
mat(i,j)=sum;
}}
return;
}

60
lib/qcd/action/fermion/ImprovedStaggeredFermion5D.cc
View File

@ -230,8 +230,15 @@ void ImprovedStaggeredFermion5D<Impl>::DhopInternal(StencilImpl & st, LebesgueOr
{
Compressor compressor;
int LLs = in._grid->_rdimensions[0];
DhopTotalTime -= usecond();
DhopCommTime -= usecond();
st.HaloExchange(in,compressor);
DhopCommTime += usecond();
DhopComputeTime -= usecond();
// Dhop takes the 4d grid from U, and makes a 5d index for fermion
if (dag == DaggerYes) {
parallel_for (int ss = 0; ss < U._grid->oSites(); ss++) {
@ -244,12 +251,15 @@ void ImprovedStaggeredFermion5D<Impl>::DhopInternal(StencilImpl & st, LebesgueOr
Kernels::DhopSite(st,lo,U,UUU,st.CommBuf(),LLs,sU,in,out);
}
}
DhopComputeTime += usecond();
DhopTotalTime += usecond();
}
template<class Impl>
void ImprovedStaggeredFermion5D<Impl>::DhopOE(const FermionField &in, FermionField &out,int dag)
{
DhopCalls+=1;
conformable(in._grid,FermionRedBlackGrid()); // verifies half grid
conformable(in._grid,out._grid); // drops the cb check
@ -261,6 +271,7 @@ void ImprovedStaggeredFermion5D<Impl>::DhopOE(const FermionField &in, FermionFie
template<class Impl>
void ImprovedStaggeredFermion5D<Impl>::DhopEO(const FermionField &in, FermionField &out,int dag)
{
DhopCalls+=1;
conformable(in._grid,FermionRedBlackGrid()); // verifies half grid
conformable(in._grid,out._grid); // drops the cb check
@ -272,6 +283,7 @@ void ImprovedStaggeredFermion5D<Impl>::DhopEO(const FermionField &in, FermionFie
template<class Impl>
void ImprovedStaggeredFermion5D<Impl>::Dhop(const FermionField &in, FermionField &out,int dag)
{
DhopCalls+=2;
conformable(in._grid,FermionGrid()); // verifies full grid
conformable(in._grid,out._grid);
@ -280,6 +292,54 @@ void ImprovedStaggeredFermion5D<Impl>::Dhop(const FermionField &in, FermionField
DhopInternal(Stencil,Lebesgue,Umu,UUUmu,in,out,dag);
}
template<class Impl>
void ImprovedStaggeredFermion5D<Impl>::Report(void)
{
std::vector<int> latt = GridDefaultLatt();
RealD volume = Ls; for(int mu=0;mu<Nd;mu++) volume=volume*latt[mu];
RealD NP = _FourDimGrid->_Nprocessors;
RealD NN = _FourDimGrid->NodeCount();
std::cout << GridLogMessage << "#### Dhop calls report " << std::endl;
std::cout << GridLogMessage << "ImprovedStaggeredFermion5D Number of DhopEO Calls : "
<< DhopCalls << std::endl;
std::cout << GridLogMessage << "ImprovedStaggeredFermion5D TotalTime /Calls : "
<< DhopTotalTime / DhopCalls << " us" << std::endl;
std::cout << GridLogMessage << "ImprovedStaggeredFermion5D CommTime /Calls : "
<< DhopCommTime / DhopCalls << " us" << std::endl;
std::cout << GridLogMessage << "ImprovedStaggeredFermion5D ComputeTime/Calls : "
<< DhopComputeTime / DhopCalls << " us" << std::endl;
// Average the compute time
_FourDimGrid->GlobalSum(DhopComputeTime);
DhopComputeTime/=NP;
RealD mflops = 1154*volume*DhopCalls/DhopComputeTime/2; // 2 for red black counting
std::cout << GridLogMessage << "Average mflops/s per call : " << mflops << std::endl;
std::cout << GridLogMessage << "Average mflops/s per call per rank : " << mflops/NP << std::endl;
std::cout << GridLogMessage << "Average mflops/s per call per node : " << mflops/NN << std::endl;
RealD Fullmflops = 1154*volume*DhopCalls/(DhopTotalTime)/2; // 2 for red black counting
std::cout << GridLogMessage << "Average mflops/s per call (full) : " << Fullmflops << std::endl;
std::cout << GridLogMessage << "Average mflops/s per call per rank (full): " << Fullmflops/NP << std::endl;
std::cout << GridLogMessage << "Average mflops/s per call per node (full): " << Fullmflops/NN << std::endl;
std::cout << GridLogMessage << "ImprovedStaggeredFermion5D Stencil" <<std::endl; Stencil.Report();
std::cout << GridLogMessage << "ImprovedStaggeredFermion5D StencilEven"<<std::endl; StencilEven.Report();
std::cout << GridLogMessage << "ImprovedStaggeredFermion5D StencilOdd" <<std::endl; StencilOdd.Report();
}
template<class Impl>
void ImprovedStaggeredFermion5D<Impl>::ZeroCounters(void)
{
DhopCalls = 0;
DhopTotalTime = 0;
DhopCommTime = 0;
DhopComputeTime = 0;
Stencil.ZeroCounters();
StencilEven.ZeroCounters();
StencilOdd.ZeroCounters();
}
/////////////////////////////////////////////////////////////////////////
// Implement the general interface. Here we use SAME mass on all slices

10
lib/qcd/action/fermion/ImprovedStaggeredFermion5D.h
View File

@ -55,6 +55,16 @@ namespace QCD {
FermionField _tmp;
FermionField &tmp(void) { return _tmp; }
////////////////////////////////////////
// Performance monitoring
////////////////////////////////////////
void Report(void);
void ZeroCounters(void);
double DhopTotalTime;
double DhopCalls;
double DhopCommTime;
double DhopComputeTime;
///////////////////////////////////////////////////////////////
// Implement the abstract base
///////////////////////////////////////////////////////////////

8
tests/solver/Test_staggered_block_cg_unprec.cc
View File

@ -75,7 +75,7 @@ int main (int argc, char ** argv)
LatticeGaugeField Umu(UGrid); SU3::HotConfiguration(pRNG,Umu);
RealD mass=0.003;
ImprovedStaggeredFermion5DR Ds(Umu,Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass);
ImprovedStaggeredFermion5DR Ds(Umu,Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass);
MdagMLinearOperator<ImprovedStaggeredFermion5DR,FermionField> HermOp(Ds);
ConjugateGradient<FermionField> CG(1.0e-8,10000);
@ -99,21 +99,27 @@ int main (int argc, char ** argv)
std::cout << GridLogMessage << " Calling 5d CG for "<<Ls <<" right hand sides" <<std::endl;
std::cout << GridLogMessage << "************************************************************************ "<<std::endl;
result=zero;
Ds.ZeroCounters();
CG(HermOp,src,result);
Ds.Report();
std::cout << GridLogMessage << "************************************************************************ "<<std::endl;
std::cout << GridLogMessage << "************************************************************************ "<<std::endl;
std::cout << GridLogMessage << " Calling multiRHS CG for "<<Ls <<" right hand sides" <<std::endl;
std::cout << GridLogMessage << "************************************************************************ "<<std::endl;
result=zero;
Ds.ZeroCounters();
mCG(HermOp,src,result);
Ds.Report();
std::cout << GridLogMessage << "************************************************************************ "<<std::endl;
std::cout << GridLogMessage << "************************************************************************ "<<std::endl;
std::cout << GridLogMessage << " Calling Block CG for "<<Ls <<" right hand sides" <<std::endl;
std::cout << GridLogMessage << "************************************************************************ "<<std::endl;
result=zero;
Ds.ZeroCounters();
BCGrQ(HermOp,src,result);
Ds.Report();
std::cout << GridLogMessage << "************************************************************************ "<<std::endl;