mirror of
https://github.com/paboyle/Grid.git
synced 2024-11-10 07:55:35 +00:00
254 lines
11 KiB
C++
254 lines
11 KiB
C++
/*************************************************************************************
|
|
|
|
Grid physics library, www.github.com/paboyle/Grid
|
|
|
|
Source file: ./benchmarks/Benchmark_su3.cc
|
|
|
|
Copyright (C) 2015
|
|
|
|
Author: Peter Boyle <paboyle@ph.ed.ac.uk>
|
|
Author: Peter Boyle <peterboyle@Peters-MacBook-Pro-2.local>
|
|
|
|
This program is free software; you can redistribute it and/or modify
|
|
it under the terms of the GNU General Public License as published by
|
|
the Free Software Foundation; either version 2 of the License, or
|
|
(at your option) any later version.
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
GNU General Public License for more details.
|
|
|
|
You should have received a copy of the GNU General Public License along
|
|
with this program; if not, write to the Free Software Foundation, Inc.,
|
|
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
|
|
|
|
See the full license in the file "LICENSE" in the top level distribution directory
|
|
*************************************************************************************/
|
|
/* END LEGAL */
|
|
#include <Grid/Grid.h>
|
|
|
|
using namespace std;
|
|
using namespace Grid;
|
|
using namespace Grid::QCD;
|
|
|
|
int main (int argc, char ** argv)
|
|
{
|
|
Grid_init(&argc,&argv);
|
|
#define LMAX (32)
|
|
#define LMIN (16)
|
|
#define LINC (4)
|
|
|
|
int64_t Nloop=2000;
|
|
|
|
std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
|
|
std::vector<int> mpi_layout = GridDefaultMpi();
|
|
|
|
int64_t threads = GridThread::GetThreads();
|
|
std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
|
|
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << "= Benchmarking SU3xSU3 x= x*y"<<std::endl;
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << " L "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
|
|
std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
|
|
|
|
for(int lat=LMIN;lat<=LMAX;lat+=LINC){
|
|
|
|
std::vector<int> latt_size ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
|
|
int64_t vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
|
|
GridCartesian Grid(latt_size,simd_layout,mpi_layout);
|
|
GridParallelRNG pRNG(&Grid); pRNG.SeedFixedIntegers(std::vector<int>({45,12,81,9}));
|
|
|
|
LatticeColourMatrix z(&Grid); random(pRNG,z);
|
|
LatticeColourMatrix x(&Grid); random(pRNG,x);
|
|
LatticeColourMatrix y(&Grid); random(pRNG,y);
|
|
|
|
double start=usecond();
|
|
for(int64_t i=0;i<Nloop;i++){
|
|
x=x*y;
|
|
}
|
|
double stop=usecond();
|
|
double time = (stop-start)/Nloop*1000.0;
|
|
|
|
double bytes=3.0*vol*Nc*Nc*sizeof(Complex);
|
|
double footprint=2.0*vol*Nc*Nc*sizeof(Complex);
|
|
double flops=Nc*Nc*(6.0+8.0+8.0)*vol;
|
|
std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<footprint<<" \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
|
|
|
|
}
|
|
|
|
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << "= Benchmarking SU3xSU3 z= x*y"<<std::endl;
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << " L "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
|
|
std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
|
|
|
|
for(int lat=LMIN;lat<=LMAX;lat+=LINC){
|
|
|
|
std::vector<int> latt_size ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
|
|
int64_t vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
|
|
|
|
GridCartesian Grid(latt_size,simd_layout,mpi_layout);
|
|
GridParallelRNG pRNG(&Grid); pRNG.SeedFixedIntegers(std::vector<int>({45,12,81,9}));
|
|
|
|
LatticeColourMatrix z(&Grid); random(pRNG,z);
|
|
LatticeColourMatrix x(&Grid); random(pRNG,x);
|
|
LatticeColourMatrix y(&Grid); random(pRNG,y);
|
|
|
|
double start=usecond();
|
|
for(int64_t i=0;i<Nloop;i++){
|
|
z=x*y;
|
|
}
|
|
double stop=usecond();
|
|
double time = (stop-start)/Nloop*1000.0;
|
|
|
|
double bytes=3*vol*Nc*Nc*sizeof(Complex);
|
|
double flops=Nc*Nc*(6+8+8)*vol;
|
|
std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<" \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
|
|
|
|
}
|
|
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << "= Benchmarking SU3xSU3 mult(z,x,y)"<<std::endl;
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << " L "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
|
|
std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
|
|
|
|
for(int lat=LMIN;lat<=LMAX;lat+=LINC){
|
|
|
|
std::vector<int> latt_size ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
|
|
int64_t vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
|
|
|
|
GridCartesian Grid(latt_size,simd_layout,mpi_layout);
|
|
GridParallelRNG pRNG(&Grid); pRNG.SeedFixedIntegers(std::vector<int>({45,12,81,9}));
|
|
|
|
LatticeColourMatrix z(&Grid); random(pRNG,z);
|
|
LatticeColourMatrix x(&Grid); random(pRNG,x);
|
|
LatticeColourMatrix y(&Grid); random(pRNG,y);
|
|
|
|
double start=usecond();
|
|
for(int64_t i=0;i<Nloop;i++){
|
|
mult(z,x,y);
|
|
}
|
|
double stop=usecond();
|
|
double time = (stop-start)/Nloop*1000.0;
|
|
|
|
double bytes=3*vol*Nc*Nc*sizeof(Complex);
|
|
double flops=Nc*Nc*(6+8+8)*vol;
|
|
std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<" \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
|
|
|
|
}
|
|
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << "= Benchmarking SU3xSU3 mac(z,x,y)"<<std::endl;
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << " L "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
|
|
std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
|
|
|
|
for(int lat=LMIN;lat<=LMAX;lat+=LINC){
|
|
|
|
std::vector<int> latt_size ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
|
|
int64_t vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
|
|
|
|
GridCartesian Grid(latt_size,simd_layout,mpi_layout);
|
|
GridParallelRNG pRNG(&Grid); pRNG.SeedFixedIntegers(std::vector<int>({45,12,81,9}));
|
|
|
|
LatticeColourMatrix z(&Grid); random(pRNG,z);
|
|
LatticeColourMatrix x(&Grid); random(pRNG,x);
|
|
LatticeColourMatrix y(&Grid); random(pRNG,y);
|
|
|
|
double start=usecond();
|
|
for(int64_t i=0;i<Nloop;i++){
|
|
mac(z,x,y);
|
|
}
|
|
double stop=usecond();
|
|
double time = (stop-start)/Nloop*1000.0;
|
|
|
|
double bytes=3*vol*Nc*Nc*sizeof(Complex);
|
|
double flops=Nc*Nc*(6+8+8)*vol;
|
|
std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<" \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
|
|
|
|
}
|
|
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << "= Benchmarking SU3xSU3 CovShiftForward(z,x,y)"<<std::endl;
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << " L "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
|
|
std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
|
|
|
|
for(int lat=LMIN;lat<=LMAX;lat+=LINC){
|
|
|
|
std::vector<int> latt_size ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
|
|
int64_t vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
|
|
|
|
GridCartesian Grid(latt_size,simd_layout,mpi_layout);
|
|
GridParallelRNG pRNG(&Grid); pRNG.SeedFixedIntegers(std::vector<int>({45,12,81,9}));
|
|
|
|
LatticeColourMatrix z(&Grid); random(pRNG,z);
|
|
LatticeColourMatrix x(&Grid); random(pRNG,x);
|
|
LatticeColourMatrix y(&Grid); random(pRNG,y);
|
|
|
|
for(int mu=0;mu<4;mu++){
|
|
double start=usecond();
|
|
for(int64_t i=0;i<Nloop;i++){
|
|
z = PeriodicBC::CovShiftForward(x,mu,y);
|
|
}
|
|
double stop=usecond();
|
|
double time = (stop-start)/Nloop*1000.0;
|
|
|
|
|
|
double bytes=3*vol*Nc*Nc*sizeof(Complex);
|
|
double flops=Nc*Nc*(6+8+8)*vol;
|
|
std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<" \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
|
|
}
|
|
}
|
|
#if 1
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << "= Benchmarking SU3xSU3 z= x * Cshift(y)"<<std::endl;
|
|
std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
|
|
std::cout<<GridLogMessage << " L "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
|
|
std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
|
|
|
|
for(int lat=LMIN;lat<=LMAX;lat+=LINC){
|
|
std::vector<int> latt_size ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
|
|
int64_t vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
|
|
|
|
GridCartesian Grid(latt_size,simd_layout,mpi_layout);
|
|
GridParallelRNG pRNG(&Grid); pRNG.SeedFixedIntegers(std::vector<int>({45,12,81,9}));
|
|
|
|
LatticeColourMatrix z(&Grid); random(pRNG,z);
|
|
LatticeColourMatrix x(&Grid); random(pRNG,x);
|
|
LatticeColourMatrix y(&Grid); random(pRNG,y);
|
|
LatticeColourMatrix tmp(&Grid);
|
|
|
|
for(int mu=0;mu<4;mu++){
|
|
double tshift=0;
|
|
double tmult =0;
|
|
|
|
double start=usecond();
|
|
for(int64_t i=0;i<Nloop;i++){
|
|
tshift-=usecond();
|
|
tmp = Cshift(y,mu,-1);
|
|
tshift+=usecond();
|
|
tmult-=usecond();
|
|
z = x*tmp;
|
|
tmult+=usecond();
|
|
}
|
|
double stop=usecond();
|
|
double time = (stop-start)/Nloop;
|
|
tshift = tshift/Nloop;
|
|
tmult = tmult /Nloop;
|
|
|
|
double bytes=3*vol*Nc*Nc*sizeof(Complex);
|
|
double flops=Nc*Nc*(6+8+8)*vol;
|
|
std::cout<<GridLogMessage<<std::setprecision(3) << "total us "<<time<<" shift "<<tshift <<" mult "<<tmult<<std::endl;
|
|
time = time * 1000; // convert to NS for GB/s
|
|
std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<" \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
|
|
}
|
|
}
|
|
#endif
|
|
Grid_finalize();
|
|
}
|