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

Thread loop constructs changing a little

Browse Source
This commit is contained in:
Peter Boyle 2019-06-15 12:54:11 +01:00
parent 462900b48d
commit cb336aa8f8
8 changed files with 22 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
Hadrons/A2AMatrix.hpp
View File

@ -170,7 +170,7 @@ public:
if ((MatLeft::Options == Eigen::RowMajor) and
(MatRight::Options == Eigen::ColMajor))
{
thread_loop( (unsigned int r = 0; r < a.rows(); ++r),
thread_for(r,a.rows(),
{
C tmp;
#ifdef USE_MKL
@ -186,7 +186,7 @@ public:
}
else
{
thread_loop( (unsigned int c = 0; c < a.cols(); ++c),
thread_for(c,a.cols(),
{
C tmp;
#ifdef USE_MKL
@ -646,13 +646,14 @@ void A2AMatrixBlockComputation<T, Field, MetadataType, TIo>
bytes += kernel.bytes(N_iii, N_jjj);
START_TIMER("cache copy");
thread_loop_collapse( 5, (int e =0;e<next_;e++),
for(int s =0;s< nstr_;s++)
for(int t =0;t< nt_;t++)
for(int iii=0;iii< N_iii;iii++)
for(int jjj=0;jjj< N_jjj;jjj++)
{
thread_for_collapse( 5,e,next_,{
for(int s =0;s< nstr_;s++)
for(int t =0;t< nt_;t++)
for(int iii=0;iii< N_iii;iii++)
for(int jjj=0;jjj< N_jjj;jjj++)
{
mBlock(e,s,t,ii+iii,jj+jjj) = mCacheBlock(e,s,t,iii,jjj);
}
});
STOP_TIMER("cache copy");
}

1
Hadrons/Global.cc
View File

@ -29,7 +29,6 @@ See the full license in the file "LICENSE" in the top level distribution directo
#include <Hadrons/Global.hpp>
using namespace Grid;
using namespace QCD;
using namespace Hadrons;
HadronsLogger Hadrons::HadronsLogError(1,"Error");

1
Hadrons/Global.hpp
View File

@ -50,7 +50,6 @@ See the full license in the file "LICENSE" in the top level distribution directo
#define BEGIN_HADRONS_NAMESPACE \
namespace Grid {\
using namespace QCD;\
namespace Hadrons {\
using Grid::operator<<;\
using Grid::operator>>;

2
Hadrons/Modules/MNPR/FourQuark.hpp
View File

@ -140,7 +140,7 @@ void TFourQuark<FImpl1, FImpl2>::tensorprod(LatticeSpinColourSpinColourMatrix &l
auto lret_v = lret.View();
auto a_v = a.View();
auto b_v = b.View();
thread_loop( (auto site=lret_v.begin();site<lret_v.end();site++) ,{
thread_foreach( site,lret_v,{
vTComplex left;
for(int si=0; si < Ns; ++si){
for(int sj=0; sj < Ns; ++sj){

1
Hadrons/TimerArray.cc
View File

@ -28,7 +28,6 @@ See the full license in the file "LICENSE" in the top level distribution directo
#include <Hadrons/TimerArray.hpp>
using namespace Grid;
using namespace QCD;
using namespace Hadrons;
void TimerArray::startTimer(const std::string &name)

12
Hadrons/Utilities/Contractor.cc
View File

@ -30,7 +30,6 @@ See the full license in the file "LICENSE" in the top level distribution directo
#include <Hadrons/TimerArray.hpp>
using namespace Grid;
using namespace QCD;
using namespace Hadrons;
#define TIME_MOD(t) (((t) + par.global.nt) % par.global.nt)
@ -353,11 +352,12 @@ int main(int argc, char* argv[])
tAr.startTimer("Transpose caching");
lastTerm[t].resize(ref.rows(), ref.cols());
thread_loop( (unsigned int j = 0; j < ref.cols(); ++j),
for (unsigned int i = 0; i < ref.rows(); ++i)
{
lastTerm[t](i, j) = ref(i, j);
});
thread_for( j,ref.cols(),{
for (unsigned int i = 0; i < ref.rows(); ++i)
{
lastTerm[t](i, j) = ref(i, j);
}
});
tAr.stopTimer("Transpose caching");
}
bytes = par.global.nt*lastTerm[0].rows()*lastTerm[0].cols()*sizeof(ComplexD);

12
Hadrons/Utilities/ContractorBenchmark.cc
View File

@ -205,7 +205,7 @@ void fullTrBenchmark(const unsigned int ni, const unsigned int nj, const unsigne
auto nr = a.rows(), nc = a.cols();
res = 0.;
thread_loop( (unsigned int i = 0; i < nr; ++i),
thread_for(i,nr,
{
ComplexD tmp = 0.;
@ -225,7 +225,7 @@ void fullTrBenchmark(const unsigned int ni, const unsigned int nj, const unsigne
auto nr = a.rows(), nc = a.cols();
res = 0.;
thread_loop( (unsigned int j = 0; j < nc; ++j),
thread_for(j,nc,
{
ComplexD tmp = 0.;
@ -248,7 +248,7 @@ void fullTrBenchmark(const unsigned int ni, const unsigned int nj, const unsigne
[](ComplexD &res, const MatLeft &a, const MatRight &b)
{
res = 0.;
thread_loop( (unsigned int r = 0; r < a.rows(); ++r),
thread_for(r,a.rows(),
{
ComplexD tmp;
@ -263,7 +263,7 @@ void fullTrBenchmark(const unsigned int ni, const unsigned int nj, const unsigne
[](ComplexD &res, const MatLeft &a, const MatRight &b)
{
res = 0.;
thread_loop( (unsigned int c = 0; c < a.cols(); ++c),
thread_for(c,a.cols(),
{
ComplexD tmp;
@ -284,7 +284,7 @@ void fullTrBenchmark(const unsigned int ni, const unsigned int nj, const unsigne
[](ComplexD &res, const MatLeft &a, const MatRight &b)
{
res = 0.;
thread_loop( (unsigned int r = 0; r < a.rows(); ++r),
thread_for(r,a.rows()
{
ComplexD tmp;
@ -299,7 +299,7 @@ void fullTrBenchmark(const unsigned int ni, const unsigned int nj, const unsigne
[](ComplexD &res, const MatLeft &a, const MatRight &b)
{
res = 0.;
thread_loop( (unsigned int c = 0; c < a.cols(); ++c),
thread_for(c,a.cols(),
{
ComplexD tmp;

1
Hadrons/Utilities/EigenPackCast.cc
View File

@ -29,7 +29,6 @@ See the full license in the file "LICENSE" in the top level distribution directo
#include <Hadrons/Environment.hpp>
using namespace Grid;
using namespace QCD;
using namespace Hadrons;
template <typename FOut, typename FIn>