portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-14 05:07:05 +01:00
Code Releases Activity

Sizable improvement in multigrid for unsquared.

Browse Source 
6000 matmuls CG unprec
2000 matmuls CG prec (4000 eo muls)
1050 matmuls PGCR on 16^3 x 32 x 8 m=.01

Substantial effort on timing and logging infrastructure
This commit is contained in:
Peter Boyle
2015-07-24 01:31:13 +09:00
parent d6a2d734d3
commit 28bdc90908
67 changed files with 945 additions and 753 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
lib/Grid.h
View File

@ -35,6 +35,9 @@
#include <GridConfig.h>
#include <GridTime.h>
#include <GridLog.h>
////////////////////////////////////////////////////////////
// Tunable header includes
////////////////////////////////////////////////////////////

123
lib/GridInit.cc
View File

@ -25,17 +25,19 @@
namespace Grid {
//////////////////////////////////////////////////////
// Convenience functions to access stadard command line arg
// driven parallelism controls
//////////////////////////////////////////////////////
static std::vector<int> Grid_default_latt;
static std::vector<int> Grid_default_mpi;
//////////////////////////////////////////////////////
// Convenience functions to access stadard command line arg
// driven parallelism controls
//////////////////////////////////////////////////////
static std::vector<int> Grid_default_latt;
static std::vector<int> Grid_default_mpi;
int GridThread::_threads;
int GridThread::_threads;
const std::vector<int> GridDefaultSimd(int dims,int nsimd)
{
const std::vector<int> &GridDefaultLatt(void) {return Grid_default_latt;};
const std::vector<int> &GridDefaultMpi(void) {return Grid_default_mpi;};
const std::vector<int> GridDefaultSimd(int dims,int nsimd)
{
std::vector<int> layout(dims);
int nn=nsimd;
for(int d=dims-1;d>=0;d--){
@ -48,15 +50,11 @@ namespace Grid {
}
assert(nn==1);
return layout;
}
}
const std::vector<int> &GridDefaultLatt(void) {return Grid_default_latt;};
const std::vector<int> &GridDefaultMpi(void) {return Grid_default_mpi;};
////////////////////////////////////////////////////////////
// Command line parsing assist for stock controls
////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////
// Command line parsing assist for stock controls
////////////////////////////////////////////////////////////
std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option)
{
char ** itr = std::find(begin, end, option);
@ -70,6 +68,23 @@ bool GridCmdOptionExists(char** begin, char** end, const std::string& option)
{
return std::find(begin, end, option) != end;
}
// Comma separated list
void GridCmdOptionCSL(std::string str,std::vector<std::string> & vec)
{
size_t pos = 0;
std::string token;
std::string delimiter(",");
vec.resize(0);
while ((pos = str.find(delimiter)) != std::string::npos) {
token = str.substr(0, pos);
vec.push_back(token);
str.erase(0, pos + delimiter.length());
}
token = str;
vec.push_back(token);
return;
}
void GridCmdOptionIntVector(std::string &str,std::vector<int> & vec)
{
@ -84,6 +99,7 @@ void GridCmdOptionIntVector(std::string &str,std::vector<int> & vec)
return;
}
void GridParseLayout(char **argv,int argc,
std::vector<int> &latt,
std::vector<int> &mpi)
@ -117,8 +133,9 @@ std::string GridCmdVectorIntToString(const std::vector<int> & vec){
std::copy(vec.begin(), vec.end(),std::ostream_iterator<int>(oss, " "));
return oss.str();
}
/////////////////////////////////////////////////////////
/////////////////////////////////////////////////////////
/////////////////////////////////////////////////////////
//
/////////////////////////////////////////////////////////
void Grid_init(int *argc,char ***argv)
{
#ifdef GRID_COMMS_MPI
@ -126,15 +143,33 @@ void Grid_init(int *argc,char ***argv)
#endif
// Parse command line args.
GridLogger::StopWatch.Start();
std::string arg;
std::vector<std::string> logstreams;
std::string defaultLog("Error,Warning,Message,Performance");
GridCmdOptionCSL(defaultLog,logstreams);
GridLogConfigure(logstreams);
if( GridCmdOptionExists(*argv,*argv+*argc,"--help") ){
std::cout<<"--help : this message"<<std::endl;
std::cout<<"--debug-signals : catch sigsegv and print a blame report"<<std::endl;
std::cout<<"--debug-stdout : print stdout from EVERY node"<<std::endl;
std::cout<<"--decomposition : report on default omp,mpi and simd decomposition"<<std::endl;
std::cout<<"--mpi n.n.n.n : default MPI decomposition"<<std::endl;
std::cout<<"--omp n : default number of OMP threads"<<std::endl;
std::cout<<"--grid n.n.n.n : default Grid size"<<std::endl;
std::cout<<GridLogMessage<<"--help : this message"<<std::endl;
std::cout<<GridLogMessage<<"--debug-signals : catch sigsegv and print a blame report"<<std::endl;
std::cout<<GridLogMessage<<"--debug-stdout : print stdout from EVERY node"<<std::endl;
std::cout<<GridLogMessage<<"--decomposition : report on default omp,mpi and simd decomposition"<<std::endl;
std::cout<<GridLogMessage<<"--mpi n.n.n.n : default MPI decomposition"<<std::endl;
std::cout<<GridLogMessage<<"--omp n : default number of OMP threads"<<std::endl;
std::cout<<GridLogMessage<<"--grid n.n.n.n : default Grid size"<<std::endl;
std::cout<<GridLogMessage<<"--log list : comma separted list of streams from Error,Warning,Message,Performance,Iterative,Debug"<<std::endl;
}
if( GridCmdOptionExists(*argv,*argv+*argc,"--log") ){
arg = GridCmdOptionPayload(*argv,*argv+*argc,"--log");
GridCmdOptionCSL(arg,logstreams);
GridLogConfigure(logstreams);
}
if( GridCmdOptionExists(*argv,*argv+*argc,"--debug-signals") ){
Grid_debug_handler_init();
}
@ -152,38 +187,18 @@ void Grid_init(int *argc,char ***argv)
Grid_default_latt,
Grid_default_mpi);
if( GridCmdOptionExists(*argv,*argv+*argc,"--decomposition") ){
std::cout<<"Grid Decomposition\n";
std::cout<<"\tOpenMP threads : "<<GridThread::GetThreads()<<std::endl;
std::cout<<"\tMPI tasks : "<<GridCmdVectorIntToString(GridDefaultMpi())<<std::endl;
std::cout<<"\tvRealF : "<<sizeof(vRealF)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealF::Nsimd()))<<std::endl;
std::cout<<"\tvRealD : "<<sizeof(vRealD)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealD::Nsimd()))<<std::endl;
std::cout<<"\tvComplexF : "<<sizeof(vComplexF)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexF::Nsimd()))<<std::endl;
std::cout<<"\tvComplexD : "<<sizeof(vComplexD)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexD::Nsimd()))<<std::endl;
std::cout<<GridLogMessage<<"Grid Decomposition\n";
std::cout<<GridLogMessage<<"\tOpenMP threads : "<<GridThread::GetThreads()<<std::endl;
std::cout<<GridLogMessage<<"\tMPI tasks : "<<GridCmdVectorIntToString(GridDefaultMpi())<<std::endl;
std::cout<<GridLogMessage<<"\tvRealF : "<<sizeof(vRealF)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealF::Nsimd()))<<std::endl;
std::cout<<GridLogMessage<<"\tvRealD : "<<sizeof(vRealD)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealD::Nsimd()))<<std::endl;
std::cout<<GridLogMessage<<"\tvComplexF : "<<sizeof(vComplexF)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexF::Nsimd()))<<std::endl;
std::cout<<GridLogMessage<<"\tvComplexD : "<<sizeof(vComplexD)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexD::Nsimd()))<<std::endl;
}
}
////////////////////////////////////////////////////////////
// Verbose limiter on MPI tasks
////////////////////////////////////////////////////////////
void Grid_quiesce_nodes(void)
{
#ifdef GRID_COMMS_MPI
int me;
MPI_Comm_rank(MPI_COMM_WORLD,&me);
if ( me ) {
std::cout.setstate(std::ios::badbit);
}
#endif
}
void Grid_unquiesce_nodes(void)
{
#ifdef GRID_COMMS_MPI
std::cout.clear();
#endif
}
void Grid_finalize(void)
{

62
lib/GridLog.cc Normal file
View File

@ -0,0 +1,62 @@
#include <Grid.h>
namespace Grid {
GridStopWatch GridLogger::StopWatch;
std::ostream GridLogger::devnull(0);
GridLogger GridLogError (1,"Error");
GridLogger GridLogWarning (1,"Warning");
GridLogger GridLogMessage (1,"Message");
GridLogger GridLogDebug (1,"Debug");
GridLogger GridLogPerformance(1,"Performance");
GridLogger GridLogIterative (1,"Iterative");
void GridLogConfigure(std::vector<std::string> &logstreams)
{
GridLogError.Active(0);
GridLogWarning.Active(0);
GridLogMessage.Active(0);
GridLogIterative.Active(0);
GridLogDebug.Active(0);
GridLogPerformance.Active(0);
for(int i=0;i<logstreams.size();i++){
if ( logstreams[i]== std::string("Error") ) GridLogError.Active(1);
if ( logstreams[i]== std::string("Warning") ) GridLogWarning.Active(1);
if ( logstreams[i]== std::string("Message") ) GridLogMessage.Active(1);
if ( logstreams[i]== std::string("Iterative") ) GridLogIterative.Active(1);
if ( logstreams[i]== std::string("Debug") ) GridLogDebug.Active(1);
if ( logstreams[i]== std::string("Performance") ) GridLogPerformance.Active(1);
}
}
////////////////////////////////////////////////////////////
// Verbose limiter on MPI tasks
////////////////////////////////////////////////////////////
void Grid_quiesce_nodes(void)
{
#ifdef GRID_COMMS_MPI
int me;
MPI_Comm_rank(MPI_COMM_WORLD,&me);
if ( me ) {
std::cout<<GridLogMessage.setstate(std::ios::badbit);
}
#endif
}
void Grid_unquiesce_nodes(void)
{
#ifdef GRID_COMMS_MPI
std::cout<<GridLogMessage.clear();
#endif
}
std::ostream& operator<< (std::ostream& stream, const GridTime& time)
{
stream << time.count()<<" ms";
return stream;
}
}

46
lib/GridLog.h Normal file
View File

@ -0,0 +1,46 @@
#ifndef GRID_LOG_H
#define GRID_LOG_H
namespace Grid {
// Dress the output; use std::chrono for time stamping via the StopWatch class
std::ostream& operator<< (std::ostream& stream, const GridTime& time);
class GridLogger {
int active;
std::string name;
public:
static GridStopWatch StopWatch;
static std::ostream devnull;
GridLogger(int on, std::string nm): active(on), name(nm) {
};
void Active(int on) {active = on;};
friend std::ostream& operator<< (std::ostream& stream, const GridLogger& log){
if ( log.active ) {
StopWatch.Stop();
GridTime now = StopWatch.Elapsed();
StopWatch.Start();
stream << "Grid : "<<log.name << " : " << now << " : ";
return stream;
} else {
return devnull;
}
}
};
void GridLogConfigure(std::vector<std::string> &logstreams);
extern GridLogger GridLogError;
extern GridLogger GridLogWarning;
extern GridLogger GridLogMessage;
extern GridLogger GridLogDebug ;
extern GridLogger GridLogPerformance;
extern GridLogger GridLogIterative ;
}
#endif

44
lib/GridTime.h Normal file
View File

@ -0,0 +1,44 @@
#ifndef GRID_TIME_H
#define GRID_TIME_H
namespace Grid {
// Dress the output; use std::chrono
#include <chrono>
#include <ctime>
typedef std::chrono::system_clock GridClock;
typedef std::chrono::time_point<GridClock> GridTimePoint;
typedef std::chrono::milliseconds GridTime;
class GridStopWatch {
private:
bool running;
GridTimePoint start;
GridTime accumulator;
public:
GridStopWatch () {
Reset();
}
void Start(void) {
assert(running == false);
start = GridClock::now();
running = true;
}
void Stop(void) {
assert(running == true);
accumulator+= std::chrono::duration_cast<GridTime>(GridClock::now()-start);
running = false;
};
void Reset(void){
running = false;
start = GridClock::now();
accumulator = std::chrono::duration_cast<GridTime>(start-start);
}
GridTime Elapsed(void) {
assert(running == false);
return accumulator;
}
};
}
#endif

4
lib/Make.inc
View File

@ -1,4 +1,4 @@
HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_exp.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_logical.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_index.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_determinant.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/HMC.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/ActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Old/Tensor_poke.h ./Old/Tensor_peek.h ./Threads.h ./Grid.h ./algorithms/Preconditioner.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/AdefGeneric.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/GeneralisedMinimumResidual.h ./algorithms/iterative/gmres.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./GridLog.h ./GridTime.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
CCFILES=./qcd/hmc/integrators/Integrator.cc ./qcd/hmc/HMC.cc ./qcd/utils/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/spin/Dirac.cc ./GridInit.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./GridLog.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/hmc/integrators/Integrator.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc

53
lib/algorithms/CoarsenedMatrix.h
View File

@ -32,12 +32,12 @@ namespace Grid {
displacements[2*_d]=0;
//// report back
std::cout<<"directions :";
std::cout<<GridLogMessage<<"directions :";
for(int d=0;d<npoint;d++) std::cout<< directions[d]<< " ";
std::cout <<std::endl;
std::cout<<"displacements :";
std::cout<<GridLogMessage<<"displacements :";
for(int d=0;d<npoint;d++) std::cout<< displacements[d]<< " ";
std::cout <<std::endl;
std::cout<<std::endl;
}
/*
@ -100,9 +100,9 @@ namespace Grid {
eProj._odata[ss](i)=CComplex(1.0);
}
eProj=eProj - iProj;
std::cout<<"Orthog check error "<<i<<" " << norm2(eProj)<<std::endl;
std::cout<<GridLogMessage<<"Orthog check error "<<i<<" " << norm2(eProj)<<std::endl;
}
std::cout <<"CheckOrthog done"<<std::endl;
std::cout<<GridLogMessage <<"CheckOrthog done"<<std::endl;
}
void ProjectToSubspace(CoarseVector &CoarseVec,const FineField &FineVec){
blockProject(CoarseVec,FineVec,subspace);
@ -113,7 +113,7 @@ namespace Grid {
void CreateSubspaceRandom(GridParallelRNG &RNG){
for(int i=0;i<nbasis;i++){
random(RNG,subspace[i]);
std::cout<<" norm subspace["<<i<<"] "<<norm2(subspace[i])<<std::endl;
std::cout<<GridLogMessage<<" norm subspace["<<i<<"] "<<norm2(subspace[i])<<std::endl;
}
Orthogonalise();
}
@ -121,7 +121,7 @@ namespace Grid {
RealD scale;
ConjugateGradient<FineField> CG(2.0e-3,10000);
ConjugateGradient<FineField> CG(1.0e-2,10000);
FineField noise(FineGrid);
FineField Mn(FineGrid);
@ -131,7 +131,7 @@ namespace Grid {
scale = std::pow(norm2(noise),-0.5);
noise=noise*scale;
hermop.Op(noise,Mn); std::cout << "noise ["<<b<<"] <n|MdagM|n> "<<norm2(Mn)<<std::endl;
hermop.Op(noise,Mn); std::cout<<GridLogMessage << "noise ["<<b<<"] <n|MdagM|n> "<<norm2(Mn)<<std::endl;
for(int i=0;i<1;i++){
@ -143,7 +143,7 @@ namespace Grid {
}
hermop.Op(noise,Mn); std::cout << "filtered["<<b<<"] <f|MdagM|f> "<<norm2(Mn)<<std::endl;
hermop.Op(noise,Mn); std::cout<<GridLogMessage << "filtered["<<b<<"] <f|MdagM|f> "<<norm2(Mn)<<std::endl;
subspace[b] = noise;
}
@ -189,7 +189,7 @@ namespace Grid {
SimpleCompressor<siteVector> compressor;
Stencil.HaloExchange(in,comm_buf,compressor);
//PARALLEL_FOR_LOOP
PARALLEL_FOR_LOOP
for(int ss=0;ss<Grid()->oSites();ss++){
siteVector res = zero;
siteVector nbr;
@ -252,10 +252,6 @@ namespace Grid {
// Orthogonalise the subblocks over the basis
blockOrthogonalise(InnerProd,Subspace.subspace);
//Subspace.Orthogonalise();
// Subspace.CheckOrthogonal();
//Subspace.Orthogonalise();
// Subspace.CheckOrthogonal();
// Compute the matrix elements of linop between this orthonormal
// set of vectors.
@ -306,6 +302,7 @@ namespace Grid {
Subspace.ProjectToSubspace(oProj,oblock);
// blockProject(iProj,iblock,Subspace.subspace);
// blockProject(oProj,oblock,Subspace.subspace);
PARALLEL_FOR_LOOP
for(int ss=0;ss<Grid()->oSites();ss++){
for(int j=0;j<nbasis;j++){
if( disp!= 0 ) {
@ -321,12 +318,12 @@ namespace Grid {
///////////////////////////
// test code worth preserving in if block
///////////////////////////
std::cout<< " Computed matrix elements "<< self_stencil <<std::endl;
std::cout<<GridLogMessage<< " Computed matrix elements "<< self_stencil <<std::endl;
for(int p=0;p<geom.npoint;p++){
std::cout<< "A["<<p<<"]" << std::endl;
std::cout<< A[p] << std::endl;
std::cout<<GridLogMessage<< "A["<<p<<"]" << std::endl;
std::cout<<GridLogMessage<< A[p] << std::endl;
}
std::cout<< " picking by block0 "<< self_stencil <<std::endl;
std::cout<<GridLogMessage<< " picking by block0 "<< self_stencil <<std::endl;
phi=Subspace.subspace[0];
std::vector<int> bc(FineGrid->_ndimension,0);
@ -334,9 +331,9 @@ namespace Grid {
blockPick(Grid(),phi,tmp,bc); // Pick out a block
linop.Op(tmp,Mphi); // Apply big dop
blockProject(iProj,Mphi,Subspace.subspace); // project it and print it
std::cout<< " Computed matrix elements from block zero only "<<std::endl;
std::cout<< iProj <<std::endl;
std::cout<<"Computed Coarse Operator"<<std::endl;
std::cout<<GridLogMessage<< " Computed matrix elements from block zero only "<<std::endl;
std::cout<<GridLogMessage<< iProj <<std::endl;
std::cout<<GridLogMessage<<"Computed Coarse Operator"<<std::endl;
#endif
// ForceHermitian();
AssertHermitian();
@ -345,9 +342,9 @@ namespace Grid {
void ForceDiagonal(void) {
std::cout<<"**************************************************"<<std::endl;
std::cout<<"**** Forcing coarse operator to be diagonal ****"<<std::endl;
std::cout<<"**************************************************"<<std::endl;
std::cout<<GridLogMessage<<"**************************************************"<<std::endl;
std::cout<<GridLogMessage<<"**** Forcing coarse operator to be diagonal ****"<<std::endl;
std::cout<<GridLogMessage<<"**************************************************"<<std::endl;
for(int p=0;p<8;p++){
A[p]=zero;
}
@ -387,13 +384,13 @@ namespace Grid {
Diff = AA - adj(AAc);
std::cout<<"Norm diff dim "<<d<<" "<< norm2(Diff)<<std::endl;
std::cout<<"Norm dim "<<d<<" "<< norm2(AA)<<std::endl;
std::cout<<GridLogMessage<<"Norm diff dim "<<d<<" "<< norm2(Diff)<<std::endl;
std::cout<<GridLogMessage<<"Norm dim "<<d<<" "<< norm2(AA)<<std::endl;
}
Diff = A[8] - adj(A[8]);
std::cout<<"Norm diff local "<< norm2(Diff)<<std::endl;
std::cout<<"Norm local "<< norm2(A[8])<<std::endl;
std::cout<<GridLogMessage<<"Norm diff local "<< norm2(Diff)<<std::endl;
std::cout<<GridLogMessage<<"Norm local "<< norm2(A[8])<<std::endl;
}
};

2
lib/algorithms/approx/Chebyshev.h
View File

@ -120,7 +120,7 @@ namespace Grid {
Field *Tn = &T1;
Field *Tnp = &T2;
std::cout << "Chebyshev ["<<lo<<","<<hi<<"]"<< " order "<<order <<std::endl;
std::cout<<GridLogMessage << "Chebyshev ["<<lo<<","<<hi<<"]"<< " order "<<order <<std::endl;
// Tn=T1 = (xscale M + mscale)in
double xscale = 2.0/(hi-lo);
double mscale = -(hi+lo)/(hi-lo);

1
lib/algorithms/approx/Remez.cc
View File

@ -757,3 +757,4 @@ void AlgRemez::csv(std::ostream & os)
}
return;
}

1
lib/algorithms/approx/Remez.h
View File

@ -28,6 +28,7 @@
remez.getIPFE(res,pole,&norm);
remez.csv(ostream &os);
*/
class AlgRemez
{
private:

6
lib/algorithms/iterative/AdefGeneric.h
View File

@ -149,7 +149,7 @@ class TwoLevelFlexiblePcg : public LinearFunction<Field>
}
RealD rrn=sqrt(rn/ssq);
std::cout<<"TwoLevelfPcg: k= "<<k<<" residual = "<<rrn<<std::endl;
std::cout<<GridLogMessage<<"TwoLevelfPcg: k= "<<k<<" residual = "<<rrn<<std::endl;
// Stopping condition
if ( rn <= rsq ) {
@ -161,8 +161,8 @@ class TwoLevelFlexiblePcg : public LinearFunction<Field>
RealD srcnorm = sqrt(norm2(src));
RealD tmpnorm = sqrt(norm2(tmp));
RealD true_residual = tmpnorm/srcnorm;
std::cout<<"TwoLevelfPcg: true residual is "<<true_residual<<std::endl;
std::cout<<"TwoLevelfPcg: target residual was"<<Tolerance<<std::endl;
std::cout<<GridLogMessage<<"TwoLevelfPcg: true residual is "<<true_residual<<std::endl;
std::cout<<GridLogMessage<<"TwoLevelfPcg: target residual was"<<Tolerance<<std::endl;
return k;
}
}

20
lib/algorithms/iterative/ConjugateGradient.h
View File

@ -43,12 +43,12 @@ public:
ssq=norm2(src);
if ( verbose ) {
std::cout <<std::setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
std::cout <<std::setprecision(4)<< "ConjugateGradient: src "<<ssq <<std::endl;
std::cout <<std::setprecision(4)<< "ConjugateGradient: mp "<<d <<std::endl;
std::cout <<std::setprecision(4)<< "ConjugateGradient: mmp "<<b <<std::endl;
std::cout <<std::setprecision(4)<< "ConjugateGradient: cp,r "<<cp <<std::endl;
std::cout <<std::setprecision(4)<< "ConjugateGradient: p "<<a <<std::endl;
std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient: src "<<ssq <<std::endl;
std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient: mp "<<d <<std::endl;
std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient: mmp "<<b <<std::endl;
std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient: cp,r "<<cp <<std::endl;
std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient: p "<<a <<std::endl;
}
RealD rsq = Tolerance* Tolerance*ssq;
@ -58,7 +58,7 @@ public:
return;
}
if(verbose) std::cout << std::setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
if(verbose) std::cout<<GridLogMessage << std::setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
int k;
for (k=1;k<=MaxIterations;k++){
@ -80,7 +80,7 @@ public:
psi= a*p+psi;
p = p*b+r;
if (verbose) std::cout<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
if (verbose) std::cout<<GridLogMessage<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
// Stopping condition
if ( cp <= rsq ) {
@ -94,14 +94,14 @@ public:
RealD resnorm = sqrt(norm2(p));
RealD true_residual = resnorm/srcnorm;
std::cout<<"ConjugateGradient: Converged on iteration " <<k
std::cout<<GridLogMessage<<"ConjugateGradient: Converged on iteration " <<k
<<" computed residual "<<sqrt(cp/ssq)
<<" true residual "<<true_residual
<<" target "<<Tolerance<<std::endl;
return;
}
}
std::cout<<"ConjugateGradient did NOT converge"<<std::endl;
std::cout<<GridLogMessage<<"ConjugateGradient did NOT converge"<<std::endl;
assert(0);
}
};

14
lib/algorithms/iterative/ConjugateGradientMultiShift.h
View File

@ -91,7 +91,7 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
cp = norm2(src);
for(int s=0;s<nshift;s++){
rsq[s] = cp * mresidual[s] * mresidual[s];
std::cout<<"ConjugateGradientMultiShift: shift "<<s
std::cout<<GridLogMessage<<"ConjugateGradientMultiShift: shift "<<s
<<" target resid "<<rsq[s]<<std::endl;
ps[s] = src;
}
@ -109,7 +109,7 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
// p and mmp is equal to d after this since
// the d computation is tricky
// qq = real(innerProduct(p,mmp));
// std::cout << "debug equal ? qq "<<qq<<" d "<< d<<std::endl;
// std::cout<<GridLogMessage << "debug equal ? qq "<<qq<<" d "<< d<<std::endl;
b = -cp /d;
@ -214,7 +214,7 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
if(css<rsq[s]){
if ( ! converged[s] )
std::cout<<"ConjugateGradientMultiShift k="<<k<<" Shift "<<s<<" has converged"<<std::endl;
std::cout<<GridLogMessage<<"ConjugateGradientMultiShift k="<<k<<" Shift "<<s<<" has converged"<<std::endl;
converged[s]=1;
} else {
all_converged=0;
@ -225,8 +225,8 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
if ( all_converged ){
std::cout<< "CGMultiShift: All shifts have converged iteration "<<k<<std::endl;
std::cout<< "CGMultiShift: Checking solutions"<<std::endl;
std::cout<<GridLogMessage<< "CGMultiShift: All shifts have converged iteration "<<k<<std::endl;
std::cout<<GridLogMessage<< "CGMultiShift: Checking solutions"<<std::endl;
// Check answers
for(int s=0; s < nshift; s++) {
@ -235,13 +235,13 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
axpy(r,-alpha[s],src,tmp);
RealD rn = norm2(r);
RealD cn = norm2(src);
std::cout<<"CGMultiShift: shift["<<s<<"] true residual "<<std::sqrt(rn/cn)<<std::endl;
std::cout<<GridLogMessage<<"CGMultiShift: shift["<<s<<"] true residual "<<std::sqrt(rn/cn)<<std::endl;
}
return;
}
}
// ugly hack
std::cout<<"CG multi shift did not converge"<<std::endl;
std::cout<<GridLogMessage<<"CG multi shift did not converge"<<std::endl;
assert(0);
}

8
lib/algorithms/iterative/ConjugateResidual.h
View File

@ -41,7 +41,7 @@ namespace Grid {
ssq=norm2(src);
rsq=Tolerance*Tolerance*ssq;
if (verbose) std::cout<<"ConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
if (verbose) std::cout<<GridLogMessage<<"ConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
for(int k=1;k<MaxIterations;k++){
@ -60,13 +60,13 @@ namespace Grid {
axpy(p,b,p,r);
pAAp=axpy_norm(Ap,b,Ap,Ar);
if(verbose) std::cout<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
if(verbose) std::cout<<GridLogMessage<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
if(cp<rsq) {
Linop.HermOp(psi,Ap);
axpy(r,-1.0,src,Ap);
RealD true_resid = norm2(r)/ssq;
std::cout<<"ConjugateResidual: Converged on iteration " <<k
std::cout<<GridLogMessage<<"ConjugateResidual: Converged on iteration " <<k
<< " computed residual "<<sqrt(cp/ssq)
<< " true residual "<<sqrt(true_resid)
<< " target " <<Tolerance <<std::endl;
@ -75,7 +75,7 @@ namespace Grid {
}
std::cout<<"ConjugateResidual did NOT converge"<<std::endl;
std::cout<<GridLogMessage<<"ConjugateResidual did NOT converge"<<std::endl;
assert(0);
}
};

26
lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
View File

@ -45,13 +45,13 @@ namespace Grid {
cp=GCRnStep(Linop,src,psi,rsq);
if ( verbose ) std::cout<<"VPGCR("<<mmax<<","<<nstep<<") "<< steps <<" steps cp = "<<cp<<std::endl;
if ( verbose ) std::cout<<GridLogMessage<<"VPGCR("<<mmax<<","<<nstep<<") "<< steps <<" steps cp = "<<cp<<std::endl;
if(cp<rsq) {
Linop.HermOp(psi,r);
axpy(r,-1.0,src,r);
RealD tr = norm2(r);
std::cout<<"PrecGeneralisedConjugateResidual: Converged on iteration " <<steps
std::cout<<GridLogMessage<<"PrecGeneralisedConjugateResidual: Converged on iteration " <<steps
<< " computed residual "<<sqrt(cp/ssq)
<< " true residual " <<sqrt(tr/ssq)
<< " target " <<Tolerance <<std::endl;
@ -59,7 +59,7 @@ namespace Grid {
}
}
std::cout<<"Variable Preconditioned GCR did not converge"<<std::endl;
std::cout<<GridLogMessage<<"Variable Preconditioned GCR did not converge"<<std::endl;
assert(0);
}
RealD GCRnStep(LinearOperatorBase<Field> &Linop,const Field &src, Field &psi,RealD rsq){
@ -96,21 +96,21 @@ namespace Grid {
/////////////////////
Preconditioner(r,z);
std::cout<< " Preconditioner in " << norm2(r)<<std::endl;
std::cout<< " Preconditioner out " << norm2(z)<<std::endl;
std::cout<<GridLogMessage<< " Preconditioner in " << norm2(r)<<std::endl;
std::cout<<GridLogMessage<< " Preconditioner out " << norm2(z)<<std::endl;
Linop.HermOp(z,tmp);
std::cout<< " Preconditioner Aout " << norm2(tmp)<<std::endl;
std::cout<<GridLogMessage<< " Preconditioner Aout " << norm2(tmp)<<std::endl;
ttmp=tmp;
tmp=tmp-r;
std::cout<< " Preconditioner resid " << std::sqrt(norm2(tmp)/norm2(r))<<std::endl;
std::cout<<GridLogMessage<< " Preconditioner resid " << std::sqrt(norm2(tmp)/norm2(r))<<std::endl;
/*
std::cout<<r<<std::endl;
std::cout<<z<<std::endl;
std::cout<<ttmp<<std::endl;
std::cout<<tmp<<std::endl;
std::cout<<GridLogMessage<<r<<std::endl;
std::cout<<GridLogMessage<<z<<std::endl;
std::cout<<GridLogMessage<<ttmp<<std::endl;
std::cout<<GridLogMessage<<tmp<<std::endl;
*/
Linop.HermOpAndNorm(z,Az,zAz,zAAz);
@ -137,7 +137,7 @@ namespace Grid {
cp = axpy_norm(r,-a,q[peri_k],r);
std::cout<< " VPGCR_step resid" <<sqrt(cp/rsq)<<std::endl;
std::cout<<GridLogMessage<< " VPGCR_step resid" <<sqrt(cp/rsq)<<std::endl;
if((k==nstep-1)||(cp<rsq)){
return cp;
}
@ -148,7 +148,7 @@ namespace Grid {
Linop.HermOp(z,tmp);
tmp=tmp-r;
std::cout<< " Preconditioner resid" <<sqrt(norm2(tmp)/norm2(r))<<std::endl;
std::cout<<GridLogMessage<< " Preconditioner resid" <<sqrt(norm2(tmp)/norm2(r))<<std::endl;
q[peri_kp]=Az;
p[peri_kp]=z;

4
lib/algorithms/iterative/SchurRedBlack.h
View File

@ -89,7 +89,7 @@ namespace Grid {
//////////////////////////////////////////////////////////////
// Call the red-black solver
//////////////////////////////////////////////////////////////
std::cout << "SchurRedBlack solver calling the MpcDagMp solver" <<std::endl;
std::cout<<GridLogMessage << "SchurRedBlack solver calling the MpcDagMp solver" <<std::endl;
_HermitianRBSolver(_HermOpEO,src_o,sol_o); assert(sol_o.checkerboard==Odd);
///////////////////////////////////////////////////
@ -108,7 +108,7 @@ namespace Grid {
RealD ns = norm2(in);
RealD nr = norm2(resid);
std::cout << "SchurRedBlackDiagMooee solver true unprec resid "<< std::sqrt(nr/ns) <<" nr "<< nr <<" ns "<<ns << std::endl;
std::cout<<GridLogMessage << "SchurRedBlackDiagMooee solver true unprec resid "<< std::sqrt(nr/ns) <<" nr "<< nr <<" ns "<<ns << std::endl;
}
};

12
lib/lattice/Lattice_ET.h
View File

@ -132,18 +132,18 @@ inline void CBFromExpression(int &cb,const T1& lat) // Lattice leaf
assert(cb==lat.checkerboard);
}
cb=lat.checkerboard;
// std::cout<<"Lattice leaf cb "<<cb<<std::endl;
// std::cout<<GridLogMessage<<"Lattice leaf cb "<<cb<<std::endl;
}
template<class T1,typename std::enable_if<!is_lattice<T1>::value, T1>::type * = nullptr >
inline void CBFromExpression(int &cb,const T1& notlat) // non-lattice leaf
{
// std::cout<<"Non lattice leaf cb"<<cb<<std::endl;
// std::cout<<GridLogMessage<<"Non lattice leaf cb"<<cb<<std::endl;
}
template <typename Op, typename T1>
inline void CBFromExpression(int &cb,const LatticeUnaryExpression<Op,T1 > &expr)
{
CBFromExpression(cb,std::get<0>(expr.second));// recurse
// std::cout<<"Unary node cb "<<cb<<std::endl;
// std::cout<<GridLogMessage<<"Unary node cb "<<cb<<std::endl;
}
template <typename Op, typename T1, typename T2>
@ -151,7 +151,7 @@ inline void CBFromExpression(int &cb,const LatticeBinaryExpression<Op,T1,T2> &ex
{
CBFromExpression(cb,std::get<0>(expr.second));// recurse
CBFromExpression(cb,std::get<1>(expr.second));
// std::cout<<"Binary node cb "<<cb<<std::endl;
// std::cout<<GridLogMessage<<"Binary node cb "<<cb<<std::endl;
}
template <typename Op, typename T1, typename T2, typename T3>
inline void CBFromExpression( int &cb,const LatticeTrinaryExpression<Op,T1,T2,T3 > &expr)
@ -159,7 +159,7 @@ inline void CBFromExpression( int &cb,const LatticeTrinaryExpression<Op,T1,T2,T3
CBFromExpression(cb,std::get<0>(expr.second));// recurse
CBFromExpression(cb,std::get<1>(expr.second));
CBFromExpression(cb,std::get<2>(expr.second));
// std::cout<<"Trinary node cb "<<cb<<std::endl;
// std::cout<<GridLogMessage<<"Trinary node cb "<<cb<<std::endl;
}
////////////////////////////////////////////
@ -370,7 +370,7 @@ using namespace Grid;
tmp.func(eval(0,v1),eval(0,v2));
auto var = v1+v2;
std::cout<<typeid(var).name()<<std::endl;
std::cout<<GridLogMessage<<typeid(var).name()<<std::endl;
v3=v1+v2;
v3=v1+v2+v1*v2;

2
lib/lattice/Lattice_base.h
View File

@ -221,7 +221,7 @@ PARALLEL_FOR_LOOP
template<class robj> strong_inline Lattice<vobj> & operator = (const Lattice<robj> & r){
this->checkerboard = r.checkerboard;
conformable(*this,r);
std::cout<<"Lattice operator ="<<std::endl;
std::cout<<GridLogMessage<<"Lattice operator ="<<std::endl;
PARALLEL_FOR_LOOP
for(int ss=0;ss<_grid->oSites();ss++){
this->_odata[ss]=r._odata[ss];

2
lib/lattice/Lattice_reduction.h
View File

@ -125,7 +125,7 @@ template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<
assert(grid!=NULL);
// FIXME
std::cout<<"WARNING ! SliceSum is unthreaded "<<grid->SumArraySize()<<" threads "<<std::endl;
std::cout<<GridLogMessage<<"WARNING ! SliceSum is unthreaded "<<grid->SumArraySize()<<" threads "<<std::endl;
const int Nd = grid->_ndimension;
const int Nsimd = grid->Nsimd();

3
lib/lattice/Lattice_rng.h
View File

@ -233,7 +233,8 @@ namespace Grid {
int words=sizeof(scalar_object)/sizeof(scalar_type);
std::vector<scalar_object> buf(Nsimd);
PARALLEL_FOR_LOOP
for(int ss=0;ss<osites;ss++){
for(int si=0;si<Nsimd;si++){

5
lib/lattice/Lattice_transfer.h
View File

@ -23,7 +23,7 @@ inline void subdivides(GridBase *coarse,GridBase *fine)
template<class vobj> inline void pickCheckerboard(int cb,Lattice<vobj> &half,const Lattice<vobj> &full){
half.checkerboard = cb;
int ssh=0;
PARALLEL_FOR_LOOP
//PARALLEL_FOR_LOOP
for(int ss=0;ss<full._grid->oSites();ss++){
std::vector<int> coor;
int cbos;
@ -40,7 +40,7 @@ PARALLEL_FOR_LOOP
template<class vobj> inline void setCheckerboard(Lattice<vobj> &full,const Lattice<vobj> &half){
int cb = half.checkerboard;
int ssh=0;
PARALLEL_FOR_LOOP
//PARALLEL_FOR_LOOP
for(int ss=0;ss<full._grid->oSites();ss++){
std::vector<int> coor;
int cbos;
@ -158,6 +158,7 @@ template<class vobj,class CComplex>
fine_inner = localInnerProduct(fineX,fineY);
blockSum(coarse_inner,fine_inner);
PARALLEL_FOR_LOOP
for(int ss=0;ss<coarse->oSites();ss++){
CoarseInner._odata[ss] = coarse_inner._odata[ss];
}

14
lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
View File

@ -10,12 +10,12 @@ namespace Grid {
void ContinuedFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata)
{
// How to check Ls matches??
// std::cout << Ls << " Ls"<<std::endl;
// std::cout << zdata->n << " - n"<<std::endl;
// std::cout << zdata->da << " -da "<<std::endl;
// std::cout << zdata->db << " -db"<<std::endl;
// std::cout << zdata->dn << " -dn"<<std::endl;
// std::cout << zdata->dd << " -dd"<<std::endl;
// std::cout<<GridLogMessage << Ls << " Ls"<<std::endl;
// std::cout<<GridLogMessage << zdata->n << " - n"<<std::endl;
// std::cout<<GridLogMessage << zdata->da << " -da "<<std::endl;
// std::cout<<GridLogMessage << zdata->db << " -db"<<std::endl;
// std::cout<<GridLogMessage << zdata->dn << " -dn"<<std::endl;
// std::cout<<GridLogMessage << zdata->dd << " -dd"<<std::endl;
assert(zdata->db==Ls);// Beta has Ls coeffs
@ -55,7 +55,7 @@ namespace Grid {
See[s] = Aee[s] - 1.0/See[s-1];
}
for(int s=0;s<Ls;s++){
std::cout <<"s = "<<s<<" Beta "<<Beta[s]<<" Aee "<<Aee[s] <<" See "<<See[s] <<std::endl;
std::cout<<GridLogMessage <<"s = "<<s<<" Beta "<<Beta[s]<<" Aee "<<Aee[s] <<" See "<<See[s] <<std::endl;
}
}

2
lib/qcd/action/fermion/DomainWallFermion.h
View File

@ -32,7 +32,7 @@ namespace Grid {
Approx::zolotarev_data *zdata = Approx::higham(eps,this->Ls);// eps is ignored for higham
assert(zdata->n==this->Ls);
std::cout << "DomainWallFermion with Ls="<<Ls<<std::endl;
std::cout<<GridLogMessage << "DomainWallFermion with Ls="<<Ls<<std::endl;
// Call base setter
this->CayleyFermion5D::SetCoefficientsTanh(zdata,1.0,0.0);

2
lib/qcd/action/fermion/MobiusFermion.h
View File

@ -30,7 +30,7 @@ namespace Grid {
{
RealD eps = 1.0;
std::cout << "MobiusFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Tanh approx"<<std::endl;
std::cout<<GridLogMessage << "MobiusFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Tanh approx"<<std::endl;
Approx::zolotarev_data *zdata = Approx::higham(eps,this->Ls);// eps is ignored for higham
assert(zdata->n==this->Ls);

2
lib/qcd/action/fermion/MobiusZolotarevFermion.h
View File

@ -34,7 +34,7 @@ namespace Grid {
Approx::zolotarev_data *zdata = Approx::zolotarev(eps,this->Ls,0);
assert(zdata->n==this->Ls);
std::cout << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
std::cout<<GridLogMessage << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
// Call base setter
this->CayleyFermion5D::SetCoefficientsZolotarev(hi,zdata,b,c);

12
lib/qcd/action/fermion/PartialFractionFermion5D.cc
View File

@ -260,12 +260,12 @@ namespace Grid {
void PartialFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata){
// check on degree matching
// std::cout << Ls << " Ls"<<std::endl;
// std::cout << zdata->n << " - n"<<std::endl;
// std::cout << zdata->da << " -da "<<std::endl;
// std::cout << zdata->db << " -db"<<std::endl;
// std::cout << zdata->dn << " -dn"<<std::endl;
// std::cout << zdata->dd << " -dd"<<std::endl;
// std::cout<<GridLogMessage << Ls << " Ls"<<std::endl;
// std::cout<<GridLogMessage << zdata->n << " - n"<<std::endl;
// std::cout<<GridLogMessage << zdata->da << " -da "<<std::endl;
// std::cout<<GridLogMessage << zdata->db << " -db"<<std::endl;
// std::cout<<GridLogMessage << zdata->dn << " -dn"<<std::endl;
// std::cout<<GridLogMessage << zdata->dd << " -dd"<<std::endl;
assert(Ls == (2*zdata->da -1) );
// Part frac

2
lib/qcd/action/fermion/WilsonFermion5D.cc
View File

@ -100,7 +100,7 @@ void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int
assert(dirdisp<=7);
assert(dirdisp>=0);
//PARALLEL_FOR_LOOP
PARALLEL_FOR_LOOP
for(int ss=0;ss<Umu._grid->oSites();ss++){
for(int s=0;s<Ls;s++){
int sU=ss;

20
lib/qcd/action/fermion/WilsonKernelsHand.cc
View File

@ -360,11 +360,11 @@ void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
MULT_2SPIN(Xp);
}
XP_RECON;
// std::cout << "XP_RECON"<<std::endl;
// std::cout << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
// std::cout << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
// std::cout << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
// std::cout << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
// std::cout<<GridLogMessage << "XP_RECON"<<std::endl;
// std::cout<<GridLogMessage << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
// std::cout<<GridLogMessage << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
// std::cout<<GridLogMessage << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
// std::cout<<GridLogMessage << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
// Yp
offset = st._offsets [Yp][ss];
@ -446,11 +446,11 @@ void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
MULT_2SPIN(Xm);
}
XM_RECON_ACCUM;
// std::cout << "XM_RECON_ACCUM"<<std::endl;
// std::cout << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
// std::cout << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
// std::cout << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
// std::cout << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
// std::cout<<GridLogMessage << "XM_RECON_ACCUM"<<std::endl;
// std::cout<<GridLogMessage << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
// std::cout<<GridLogMessage << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
// std::cout<<GridLogMessage << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
// std::cout<<GridLogMessage << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
// Ym

2
lib/qcd/action/gauge/WilsonGaugeAction.h
View File

@ -18,7 +18,7 @@ namespace Grid{
virtual RealD S(const GaugeField &U) {
RealD plaq = WilsonLoops<MatrixField,GaugeField>::avgPlaquette(U);
std::cout << "Plaq : "<<plaq << "\n";
std::cout<<GridLogMessage << "Plaq : "<<plaq << "\n";
double vol = U._grid->gSites();
return beta*(1.0 -plaq)*(Nd*(Nd-1.0))*vol*0.5;
};

18
lib/qcd/hmc/HMC.h
View File

@ -36,15 +36,15 @@ namespace Grid{
RealD prob = std::exp(-DeltaH);
random(sRNG,rn_test);
std::cout<< "--------------------------------------------\n";
std::cout<< "dH = "<<DeltaH << " Random = "<< rn_test
std::cout<<GridLogMessage<< "--------------------------------------------\n";
std::cout<<GridLogMessage<< "dH = "<<DeltaH << " Random = "<< rn_test
<< "\nAcc. Probability = " << ((prob<1.0)? prob: 1.0)<< " ";
if((prob >1.0) || (rn_test <= prob)){ // accepted
std::cout <<"-- ACCEPTED\n";
std::cout<<GridLogMessage <<"-- ACCEPTED\n";
return true;
} else { // rejected
std::cout <<"-- REJECTED\n";
std::cout<<GridLogMessage <<"-- REJECTED\n";
return false;
}
}
@ -53,12 +53,12 @@ namespace Grid{
MD.init(U); // set U and initialize P and phi's
RealD H0 = MD.S(U); // initial state action
std::cout<<"Total H before = "<< H0 << "\n";
std::cout<<GridLogMessage<<"Total H before = "<< H0 << "\n";
MD.integrate(U);
RealD H1 = MD.S(U); // updated state action
std::cout<<"Total H after = "<< H1 << "\n";
std::cout<<GridLogMessage<<"Total H after = "<< H1 << "\n";
return (H1-H0);
}
@ -81,17 +81,17 @@ namespace Grid{
// Thermalizations
for(int iter=1; iter <= Params.ThermalizationSteps; ++iter){
std::cout << "-- # Thermalization step = "<< iter << "\n";
std::cout<<GridLogMessage << "-- # Thermalization step = "<< iter << "\n";
DeltaH = evolve_step(Uin);
std::cout<< " dH = "<< DeltaH << "\n";
std::cout<<GridLogMessage<< " dH = "<< DeltaH << "\n";
}
// Actual updates (evolve a copy Ucopy then copy back eventually)
LatticeLorentzColourMatrix Ucopy(Uin._grid);
for(int iter=Params.StartingConfig;
iter < Params.Nsweeps+Params.StartingConfig; ++iter){
std::cout << "-- # Sweep = "<< iter << "\n";
std::cout<<GridLogMessage << "-- # Sweep = "<< iter << "\n";
Ucopy = Uin;
DeltaH = evolve_step(Ucopy);

4
lib/qcd/hmc/integrators/Integrator.h
View File

@ -95,7 +95,7 @@ namespace Grid{
//Initialization of momenta and actions
void init(LatticeLorentzColourMatrix& U){
std::cout<< "Integrator init\n";
std::cout<<GridLogMessage<< "Integrator init\n";
MDutils::generate_momenta(*P,pRNG);
for(int level=0; level< as.size(); ++level){
@ -119,7 +119,7 @@ namespace Grid{
RealD H = Hsum.real();
std::cout << "H_p = "<< H << "\n";
std::cout<<GridLogMessage << "H_p = "<< H << "\n";
// Actions
for(int level=0; level<as.size(); ++level)

40
lib/qcd/hmc/integrators/Integrator_algorithm.h
View File

@ -39,29 +39,29 @@ namespace Grid{
if(clock[level] == 0){ // initial half step
Integ->update_P(U,level,lambda*eps);
++clock[level];
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< clock[level] <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< clock[level] <<std::endl;
}
if(level == fl){ // lowest level
Integ->update_U(U,0.5*eps);
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"U "<< (clock[level]+1) <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"U "<< (clock[level]+1) <<std::endl;
}else{ // recursive function call
step(U,level+1,clock, Integ);
}
Integ->update_P(U,level,(1.0-2.0*lambda)*eps);
++clock[level];
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< (clock[level]) <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< (clock[level]) <<std::endl;
if(level == fl){ // lowest level
Integ->update_U(U,0.5*eps);
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"U "<< (clock[level]+1) <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"U "<< (clock[level]+1) <<std::endl;
}else{ // recursive function call
step(U,level+1,clock, Integ);
}
@ -71,14 +71,14 @@ namespace Grid{
Integ->update_P(U,level,lambda*eps);
++clock[level];
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< clock[level] <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< clock[level] <<std::endl;
}else{ // bulk step
Integ->update_P(U,level,lambda*2.0*eps);
clock[level]+=2;
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< clock[level] <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< clock[level] <<std::endl;
}
}
@ -112,13 +112,13 @@ namespace Grid{
if(clock[level] == 0){ // initial half step
Integ->update_P(U, level,eps/2.0);
++clock[level];
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< 0.5*clock[level] <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
}
if(level == fl){ // lowest level
Integ->update_U(U, eps);
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"U "<< 0.5*(clock[level]+1) <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"U "<< 0.5*(clock[level]+1) <<std::endl;
}else{ // recursive function call
step(U, level+1,clock, Integ);
}
@ -126,14 +126,14 @@ namespace Grid{
Integ->update_P(U, level,eps/2.0);
++clock[level];
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< 0.5*clock[level] <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
}else{ // bulk step
Integ->update_P(U, level,eps);
clock[level]+=2;
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< 0.5*clock[level] <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
}
}

18
lib/qcd/hmc/integrators/Integrator_base.h
View File

@ -101,7 +101,7 @@ namespace Grid{
//Initialization of momenta and actions
void init(LatticeLorentzColourMatrix& U,
GridParallelRNG& pRNG){
std::cout<< "Integrator init\n";
std::cout<<GridLogMessage<< "Integrator init\n";
if (!P)
P = new LatticeLorentzColourMatrix(U._grid);
MDutils::generate_momenta(*P,pRNG);
@ -172,13 +172,13 @@ namespace Grid{
if(clock[level] == 0){ // initial half step
Integ->update_P(U, level,eps/2);
++clock[level];
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< 0.5*clock[level] <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
}
if(level == fl){ // lowest level
Integ->update_U(U, eps);
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"U "<< 0.5*(clock[level]+1) <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"U "<< 0.5*(clock[level]+1) <<std::endl;
}else{ // recursive function call
step(U, level+1,clock, Integ);
}
@ -186,14 +186,14 @@ namespace Grid{
Integ->update_P(U, level,eps/2);
++clock[level];
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< 0.5*clock[level] <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
}else{ // bulk step
Integ->update_P(U, level,eps);
clock[level]+=2;
for(int l=0; l<level;++l) std::cout<<" ";
std::cout<<"P "<< 0.5*clock[level] <<std::endl;
for(int l=0; l<level;++l) std::cout<<GridLogMessage<<" ";
std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
}
}

28
lib/qcd/utils/SUn.h
View File

@ -372,7 +372,7 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
LatticeReal d(grid); d=zero;
LatticeReal alpha(grid);
// std::cout<<"xi "<<xi <<std::endl;
// std::cout<<GridLogMessage<<"xi "<<xi <<std::endl;
alpha = toReal(2.0*xi);
do {
@ -468,11 +468,11 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
LatticeMatrix Vcheck(grid);
Vcheck = zero;
Vcheck = where(Accepted,V*adj(V) - 1.0,Vcheck);
// std::cout << "SU3 check " <<norm2(Vcheck)<<std::endl;
// std::cout<<GridLogMessage << "SU3 check " <<norm2(Vcheck)<<std::endl;
assert(norm2(Vcheck)<1.0e-4);
// Verify the link stays in SU(3)
// std::cout <<"Checking the modified link"<<std::endl;
// std::cout<<GridLogMessage <<"Checking the modified link"<<std::endl;
Vcheck = link*adj(link) - 1.0;
assert(norm2(Vcheck)<1.0e-4);
/////////////////////////////////
@ -483,42 +483,42 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
for(int gen=0;gen<generators();gen++){
Matrix ta;
generator(gen,ta);
std::cout<< "Nc = "<<ncolour<<" t_"<<gen<<std::endl;
std::cout<<ta<<std::endl;
std::cout<<GridLogMessage<< "Nc = "<<ncolour<<" t_"<<gen<<std::endl;
std::cout<<GridLogMessage<<ta<<std::endl;
}
}
static void testGenerators(void){
Matrix ta;
Matrix tb;
std::cout<<"Checking trace ta tb is 0.5 delta_ab"<<std::endl;
std::cout<<GridLogMessage<<"Checking trace ta tb is 0.5 delta_ab"<<std::endl;
for(int a=0;a<generators();a++){
for(int b=0;b<generators();b++){
generator(a,ta);
generator(b,tb);
Complex tr =TensorRemove(trace(ta*tb));
std::cout<<tr<<" ";
std::cout<<GridLogMessage<<tr<<" ";
if(a==b) assert(abs(tr-Complex(0.5))<1.0e-6);
if(a!=b) assert(abs(tr)<1.0e-6);
}
std::cout<<std::endl;
std::cout<<GridLogMessage<<std::endl;
}
std::cout<<"Checking hermitian"<<std::endl;
std::cout<<GridLogMessage<<"Checking hermitian"<<std::endl;
for(int a=0;a<generators();a++){
generator(a,ta);
std::cout<<a<<" ";
std::cout<<GridLogMessage<<a<<" ";
assert(norm2(ta-adj(ta))<1.0e-6);
}
std::cout<<std::endl;
std::cout<<GridLogMessage<<std::endl;
std::cout<<"Checking traceless"<<std::endl;
std::cout<<GridLogMessage<<"Checking traceless"<<std::endl;
for(int a=0;a<generators();a++){
generator(a,ta);
Complex tr =TensorRemove(trace(ta));
std::cout<<a<<" ";
std::cout<<GridLogMessage<<a<<" ";
assert(abs(tr)<1.0e-6);
}
std::cout<<std::endl;
std::cout<<GridLogMessage<<std::endl;
}
// reunitarise??

6
lib/tensors/Tensor_extract_merge.h
View File

@ -57,11 +57,11 @@ inline void extract(typename std::enable_if<!isGridTensor<vsimd>::value, const v
extracted[i]=buf[i*s];
for(int ii=1;ii<s;ii++){
if ( buf[i*s]!=buf[i*s+ii] ){
std::cout << " SIMD extract failure splat = "<<s<<" ii "<<ii<<" " <<Nextr<<" "<< Nsimd<<" "<<std::endl;
std::cout<<GridLogMessage << " SIMD extract failure splat = "<<s<<" ii "<<ii<<" " <<Nextr<<" "<< Nsimd<<" "<<std::endl;
for(int vv=0;vv<Nsimd;vv++) {
std::cout<< buf[vv]<<" ";
std::cout<<GridLogMessage<< buf[vv]<<" ";
}
std::cout<<std::endl;
std::cout<<GridLogMessage<<std::endl;
assert(0);
}
assert(buf[i*s]==buf[i*s+ii]);