portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-16 14:57:05 +01:00
Code Releases Activity

Adding support for general Nc in the benchmark outputs

Browse Source
This commit is contained in:
Guido Cossu
2018-01-25 13:46:31 +01:00
parent b00d2d2c39
commit cff3bae155
5 changed files with 66 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
benchmarks/Benchmark_dwf_sweep.cc
View File

@ -51,6 +51,7 @@ int main (int argc, char ** argv)
{
Grid_init(&argc,&argv);
std::cout << GridLogMessage<< "*****************************************************************" <<std::endl;
std::cout << GridLogMessage<< "* Kernel options --dslash-generic, --dslash-unroll, --dslash-asm" <<std::endl;
std::cout << GridLogMessage<< "*****************************************************************" <<std::endl;
@ -107,6 +108,7 @@ void benchDw(std::vector<int> & latt4, int Ls, int threads,int report )
GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
GridCartesian * FGrid = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
long unsigned int single_site_flops = 8*QCD::Nc*(7+16*QCD::Nc);
std::vector<int> seeds4({1,2,3,4});
std::vector<int> seeds5({5,6,7,8});
@ -196,7 +198,7 @@ void benchDw(std::vector<int> & latt4, int Ls, int threads,int report )
if ( ! report ) {
double volume=Ls; for(int mu=0;mu<Nd;mu++) volume=volume*latt4[mu];
double flops=1344*volume*ncall;
double flops=single_site_flops*volume*ncall;
std::cout <<"\t"<<NP<< "\t"<<flops/(t1-t0)<< "\t";
}
@ -228,7 +230,7 @@ void benchDw(std::vector<int> & latt4, int Ls, int threads,int report )
if(!report){
double volume=Ls; for(int mu=0;mu<Nd;mu++) volume=volume*latt4[mu];
double flops=(1344.0*volume*ncall)/2;
double flops=(single_site_flops*volume*ncall)/2.0;
std::cout<< flops/(t1-t0);
}
}
@ -237,6 +239,7 @@ void benchDw(std::vector<int> & latt4, int Ls, int threads,int report )
#define CHECK_SDW
void benchsDw(std::vector<int> & latt4, int Ls, int threads, int report )
{
long unsigned int single_site_flops = 8*QCD::Nc*(7+16*QCD::Nc);
GridCartesian * UGrid = SpaceTimeGrid::makeFourDimGrid(latt4, GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
@ -321,7 +324,7 @@ void benchsDw(std::vector<int> & latt4, int Ls, int threads, int report )
Counter.Report();
} else {
double volume=Ls; for(int mu=0;mu<Nd;mu++) volume=volume*latt4[mu];
double flops=1344*volume*ncall;
double flops=single_site_flops*volume*ncall;
std::cout<<"\t"<< flops/(t1-t0);
}
@ -358,7 +361,7 @@ void benchsDw(std::vector<int> & latt4, int Ls, int threads, int report )
CounterSdw.Report();
} else {
double volume=Ls; for(int mu=0;mu<Nd;mu++) volume=volume*latt4[mu];
double flops=(1344.0*volume*ncall)/2;
double flops=(single_site_flops*volume*ncall)/2.0;
std::cout<<"\t"<< flops/(t1-t0);
}
}