portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-14 01:35:36 +00:00
Code Releases Activity

GPU support

Browse Source
This commit is contained in:
Peter Boyle 2019-01-01 15:07:29 +00:00
parent 8c91e82ee8
commit c43a2b599a
1 changed files with 39 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

72
benchmarks/Benchmark_meson_field.cc
View File

@ -67,9 +67,9 @@ void sliceInnerProductMesonField(std::vector< std::vector<ComplexD> > &mat,
// sum across these down to scalars // sum across these down to scalars
// splitting the SIMD // splitting the SIMD
std::vector<vector_type,alignedAllocator<vector_type> > lvSum(rd*Lblock*Rblock); std::vector<vector_type,alignedAllocator<vector_type> > lvSum(rd*Lblock*Rblock);
parallel_for (int r = 0; r < rd * Lblock * Rblock; r++){ thread_loop( (int r = 0; r < rd * Lblock * Rblock; r++),{
lvSum[r] = Zero(); lvSum[r] = Zero();
} });
std::vector<scalar_type > lsSum(ld*Lblock*Rblock,scalar_type(0.0)); std::vector<scalar_type > lsSum(ld*Lblock*Rblock,scalar_type(0.0));
@ -79,7 +79,7 @@ void sliceInnerProductMesonField(std::vector< std::vector<ComplexD> > &mat,
std::cout << GridLogMessage << " Entering first parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering first parallel loop "<<std::endl;
// Parallelise over t-direction doesn't expose as much parallelism as needed for KNL // Parallelise over t-direction doesn't expose as much parallelism as needed for KNL
parallel_for(int r=0;r<rd;r++){ thread_loop((int r=0;r<rd;r++),{
int so=r*grid->_ostride[orthogdim]; // base offset for start of plane int so=r*grid->_ostride[orthogdim]; // base offset for start of plane
@ -110,11 +110,11 @@ void sliceInnerProductMesonField(std::vector< std::vector<ComplexD> > &mat,
} }
} }
} }
} });
std::cout << GridLogMessage << " Entering second parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering second parallel loop "<<std::endl;
// Sum across simd lanes in the plane, breaking out orthog dir. // Sum across simd lanes in the plane, breaking out orthog dir.
parallel_for(int rt=0;rt<rd;rt++){ thread_loop((int rt=0;rt<rd;rt++),{
Coordinate icoor(Nd); Coordinate icoor(Nd);
@ -139,7 +139,7 @@ void sliceInnerProductMesonField(std::vector< std::vector<ComplexD> > &mat,
} }
}} }}
} });
std::cout << GridLogMessage << " Entering non parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering non parallel loop "<<std::endl;
for(int t=0;t<fd;t++) for(int t=0;t<fd;t++)
@ -199,14 +199,14 @@ void sliceInnerProductMesonFieldGamma(std::vector< std::vector<ComplexD> > &mat,
int MFlvol = ld*Lblock*Rblock*Ngamma; int MFlvol = ld*Lblock*Rblock*Ngamma;
std::vector<vector_type,alignedAllocator<vector_type> > lvSum(MFrvol); std::vector<vector_type,alignedAllocator<vector_type> > lvSum(MFrvol);
parallel_for (int r = 0; r < MFrvol; r++){ thread_loop( (int r = 0; r < MFrvol; r++),{
lvSum[r] = Zero(); lvSum[r] = Zero();
} });
std::vector<scalar_type > lsSum(MFlvol); std::vector<scalar_type > lsSum(MFlvol);
parallel_for (int r = 0; r < MFlvol; r++){ thread_loop( (int r = 0; r < MFlvol; r++),{
lsSum[r]=scalar_type(0.0); lsSum[r]=scalar_type(0.0);
} });
int e1= grid->_slice_nblock[orthogdim]; int e1= grid->_slice_nblock[orthogdim];
int e2= grid->_slice_block [orthogdim]; int e2= grid->_slice_block [orthogdim];
@ -215,7 +215,7 @@ void sliceInnerProductMesonFieldGamma(std::vector< std::vector<ComplexD> > &mat,
std::cout << GridLogMessage << " Entering first parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering first parallel loop "<<std::endl;
// Parallelise over t-direction doesn't expose as much parallelism as needed for KNL // Parallelise over t-direction doesn't expose as much parallelism as needed for KNL
parallel_for(int r=0;r<rd;r++){ thread_loop((int r=0;r<rd;r++),{
int so=r*grid->_ostride[orthogdim]; // base offset for start of plane int so=r*grid->_ostride[orthogdim]; // base offset for start of plane
@ -252,11 +252,11 @@ void sliceInnerProductMesonFieldGamma(std::vector< std::vector<ComplexD> > &mat,
} }
} }
} }
} });
std::cout << GridLogMessage << " Entering second parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering second parallel loop "<<std::endl;
// Sum across simd lanes in the plane, breaking out orthog dir. // Sum across simd lanes in the plane, breaking out orthog dir.
parallel_for(int rt=0;rt<rd;rt++){ thread_loop((int rt=0;rt<rd;rt++),{
iScalar<vector_type> temp; iScalar<vector_type> temp;
Coordinate icoor(Nd); Coordinate icoor(Nd);
@ -282,7 +282,7 @@ void sliceInnerProductMesonFieldGamma(std::vector< std::vector<ComplexD> > &mat,
} }
}}} }}}
} });
std::cout << GridLogMessage << " Entering non parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering non parallel loop "<<std::endl;
for(int t=0;t<fd;t++) for(int t=0;t<fd;t++)
@ -347,14 +347,14 @@ void sliceInnerProductMesonFieldGamma1(std::vector< std::vector<ComplexD> > &mat
int MFlvol = ld*Lblock*Rblock; int MFlvol = ld*Lblock*Rblock;
Vector<SpinMatrix_v > lvSum(MFrvol); Vector<SpinMatrix_v > lvSum(MFrvol);
parallel_for (int r = 0; r < MFrvol; r++){ thread_loop( (int r = 0; r < MFrvol; r++),{
lvSum[r] = Zero(); lvSum[r] = Zero();
} });
Vector<SpinMatrix_s > lsSum(MFlvol); Vector<SpinMatrix_s > lsSum(MFlvol);
parallel_for (int r = 0; r < MFlvol; r++){ thread_loop( (int r = 0; r < MFlvol; r++),{
lsSum[r]=scalar_type(0.0); lsSum[r]=scalar_type(0.0);
} });
int e1= grid->_slice_nblock[orthogdim]; int e1= grid->_slice_nblock[orthogdim];
int e2= grid->_slice_block [orthogdim]; int e2= grid->_slice_block [orthogdim];
@ -363,7 +363,7 @@ void sliceInnerProductMesonFieldGamma1(std::vector< std::vector<ComplexD> > &mat
std::cout << GridLogMessage << " Entering first parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering first parallel loop "<<std::endl;
// Parallelise over t-direction doesn't expose as much parallelism as needed for KNL // Parallelise over t-direction doesn't expose as much parallelism as needed for KNL
parallel_for(int r=0;r<rd;r++){ thread_loop((int r=0;r<rd;r++),{
int so=r*grid->_ostride[orthogdim]; // base offset for start of plane int so=r*grid->_ostride[orthogdim]; // base offset for start of plane
@ -394,11 +394,11 @@ void sliceInnerProductMesonFieldGamma1(std::vector< std::vector<ComplexD> > &mat
} }
} }
} }
} });
std::cout << GridLogMessage << " Entering second parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering second parallel loop "<<std::endl;
// Sum across simd lanes in the plane, breaking out orthog dir. // Sum across simd lanes in the plane, breaking out orthog dir.
parallel_for(int rt=0;rt<rd;rt++){ thread_loop((int rt=0;rt<rd;rt++),{
Coordinate icoor(Nd); Coordinate icoor(Nd);
ExtractBuffer<SpinMatrix_s> extracted(Nsimd); ExtractBuffer<SpinMatrix_s> extracted(Nsimd);
@ -422,10 +422,10 @@ void sliceInnerProductMesonFieldGamma1(std::vector< std::vector<ComplexD> > &mat
} }
}} }}
} });
std::cout << GridLogMessage << " Entering third parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering third parallel loop "<<std::endl;
parallel_for(int t=0;t<fd;t++) thread_loop((int t=0;t<fd;t++)
{ {
int pt = t / ld; // processor plane int pt = t / ld; // processor plane
int lt = t % ld; int lt = t % ld;
@ -443,7 +443,7 @@ void sliceInnerProductMesonFieldGamma1(std::vector< std::vector<ComplexD> > &mat
} }
} }
}} }}
} });
std::cout << GridLogMessage << " Done "<<std::endl; std::cout << GridLogMessage << " Done "<<std::endl;
// defer sum over nodes. // defer sum over nodes.
return; return;
@ -490,14 +490,16 @@ void sliceInnerProductMesonFieldGammaMom(std::vector< std::vector<ComplexD> > &m
int MFlvol = ld*Lblock*Rblock*Nmom; int MFlvol = ld*Lblock*Rblock*Nmom;
Vector<SpinMatrix_v > lvSum(MFrvol); Vector<SpinMatrix_v > lvSum(MFrvol);
parallel_for (int r = 0; r < MFrvol; r++){ thread_loop( (int r = 0; r < MFrvol; r++),
{
lvSum[r] = Zero(); lvSum[r] = Zero();
} });
Vector<SpinMatrix_s > lsSum(MFlvol); Vector<SpinMatrix_s > lsSum(MFlvol);
parallel_for (int r = 0; r < MFlvol; r++){ thread_loop( (int r = 0; r < MFlvol; r++),
{
lsSum[r]=scalar_type(0.0); lsSum[r]=scalar_type(0.0);
} });
int e1= grid->_slice_nblock[orthogdim]; int e1= grid->_slice_nblock[orthogdim];
int e2= grid->_slice_block [orthogdim]; int e2= grid->_slice_block [orthogdim];
@ -506,7 +508,8 @@ void sliceInnerProductMesonFieldGammaMom(std::vector< std::vector<ComplexD> > &m
std::cout << GridLogMessage << " Entering first parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering first parallel loop "<<std::endl;
// Parallelise over t-direction doesn't expose as much parallelism as needed for KNL // Parallelise over t-direction doesn't expose as much parallelism as needed for KNL
parallel_for(int r=0;r<rd;r++){ thread_loop((int r=0;r<rd;r++),
{
int so=r*grid->_ostride[orthogdim]; // base offset for start of plane int so=r*grid->_ostride[orthogdim]; // base offset for start of plane
@ -545,11 +548,12 @@ void sliceInnerProductMesonFieldGammaMom(std::vector< std::vector<ComplexD> > &m
} }
} }
} }
} });
std::cout << GridLogMessage << " Entering second parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering second parallel loop "<<std::endl;
// Sum across simd lanes in the plane, breaking out orthog dir. // Sum across simd lanes in the plane, breaking out orthog dir.
parallel_for(int rt=0;rt<rd;rt++){ thread_loop((int rt=0;rt<rd;rt++),
{
Coordinate icoor(Nd); Coordinate icoor(Nd);
ExtractBuffer<SpinMatrix_s> extracted(Nsimd); ExtractBuffer<SpinMatrix_s> extracted(Nsimd);
@ -575,10 +579,10 @@ void sliceInnerProductMesonFieldGammaMom(std::vector< std::vector<ComplexD> > &m
} }
}}} }}}
} });
std::cout << GridLogMessage << " Entering third parallel loop "<<std::endl; std::cout << GridLogMessage << " Entering third parallel loop "<<std::endl;
parallel_for(int t=0;t<fd;t++) thread_loop((int t=0;t<fd;t++),
{ {
int pt = t / ld; // processor plane int pt = t / ld; // processor plane
int lt = t % ld; int lt = t % ld;
@ -602,7 +606,7 @@ void sliceInnerProductMesonFieldGammaMom(std::vector< std::vector<ComplexD> > &m
}} }}
} }
}} }}
} });
std::cout << GridLogMessage << " Done "<<std::endl; std::cout << GridLogMessage << " Done "<<std::endl;
// defer sum over nodes. // defer sum over nodes.
return; return;