Tracing replaces self timing hooks

benchmarks/Benchmark_ITT.cc

@@ -420,7 +420,6 @@ public:
 	FGrid->Broadcast(0,&ncall,sizeof(ncall));
 
 	//	std::cout << GridLogMessage << " Estimate " << ncall << " calls per second"<<std::endl;
-	Dw.ZeroCounters();
 
 	time_statistics timestat;
 	std::vector<double> t_time(ncall);
@@ -589,7 +588,6 @@ public:
 	FGrid->Broadcast(0,&ncall,sizeof(ncall));
 
 	//	std::cout << GridLogMessage << " Estimate " << ncall << " calls per second"<<std::endl;
-	Ds.ZeroCounters();
 
 	time_statistics timestat;
 	std::vector<double> t_time(ncall);

benchmarks/Benchmark_dwf.cc

@@ -186,7 +186,6 @@ int main (int argc, char ** argv)
 
   if (1) {
     FGrid->Barrier();
-    Dw.ZeroCounters();
     Dw.Dhop(src,result,0);
     std::cout<<GridLogMessage<<"Called warmup"<<std::endl;
     double t0=usecond();
@@ -231,7 +230,6 @@ int main (int argc, char ** argv)
       exit(-1);
     }
     assert (norm2(err)< 1.0e-4 );
-    Dw.Report();
   }
 
   if (1)
@@ -306,7 +304,6 @@ int main (int argc, char ** argv)
   if ( WilsonKernelsStatic::Opt == WilsonKernelsStatic::OptInlineAsm ) std::cout << GridLogMessage<< "* Using Asm Nc=3   WilsonKernels" <<std::endl;
   std::cout << GridLogMessage<< "*********************************************************" <<std::endl;
   {
-    Dw.ZeroCounters();
     FGrid->Barrier();
     Dw.DhopEO(src_o,r_e,DaggerNo);
     double t0=usecond();
@@ -328,7 +325,6 @@ int main (int argc, char ** argv)
     std::cout<<GridLogMessage << "Deo mflop/s =   "<< flops/(t1-t0)<<std::endl;
     std::cout<<GridLogMessage << "Deo mflop/s per rank   "<< flops/(t1-t0)/NP<<std::endl;
     std::cout<<GridLogMessage << "Deo mflop/s per node   "<< flops/(t1-t0)/NN<<std::endl;
-    Dw.Report();
   }
   Dw.DhopEO(src_o,r_e,DaggerNo);
   Dw.DhopOE(src_e,r_o,DaggerNo);

benchmarks/Benchmark_gparity.cc

@@ -93,7 +93,6 @@ int main (int argc, char ** argv)
   int ncall =1000;
   if (1) {
     FGrid->Barrier();
-    Dw.ZeroCounters();
     Dw.Dhop(src,result,0);
     std::cout<<GridLogMessage<<"Called warmup"<<std::endl;
     double t0=usecond();
@@ -112,7 +111,6 @@ int main (int argc, char ** argv)
     std::cout<<GridLogMessage << "mflop/s =   "<< flops/(t1-t0)<<std::endl;
     std::cout<<GridLogMessage << "mflop/s per rank =  "<< flops/(t1-t0)/NP<<std::endl;
     std::cout<<GridLogMessage << "mflop/s per node =  "<< flops/(t1-t0)/NN<<std::endl;
-    Dw.Report();
   }
 
 
@@ -134,7 +132,6 @@ int main (int argc, char ** argv)
   GparityDomainWallFermionD DwD(Umu_d,*FGrid_d,*FrbGrid_d,*UGrid_d,*UrbGrid_d,mass,M5);
   if (1) {
     FGrid_d->Barrier();
-    DwD.ZeroCounters();
     DwD.Dhop(src_d,result_d,0);
     std::cout<<GridLogMessage<<"Called warmup"<<std::endl;
     double t0=usecond();
@@ -153,7 +150,6 @@ int main (int argc, char ** argv)
     std::cout<<GridLogMessage << "mflop/s =   "<< flops/(t1-t0)<<std::endl;
     std::cout<<GridLogMessage << "mflop/s per rank =  "<< flops/(t1-t0)/NP<<std::endl;
     std::cout<<GridLogMessage << "mflop/s per node =  "<< flops/(t1-t0)/NN<<std::endl;
-    DwD.Report();
   }
 #endif
   Grid_finalize();

benchmarks/Benchmark_mooee.cc

@@ -103,35 +103,30 @@ int main (int argc, char ** argv)
 #define BENCH_DW(A,...)			\
     Dw. A (__VA_ARGS__);				\
     FGrid->Barrier();				\
-    Dw.CayleyZeroCounters();      \
     t0=usecond();				\
     for(int i=0;i<ncall;i++){			\
       Dw. A (__VA_ARGS__);				\
     }						\
     t1=usecond();				\
     FGrid->Barrier();				\
-    Dw.CayleyReport();					\
     std::cout<<GridLogMessage << "Called " #A " "<< (t1-t0)/ncall<<" us"<<std::endl;\
     std::cout<<GridLogMessage << "******************"<<std::endl;
 
 #define BENCH_ZDW(A,in,out)			\
     zDw. A (in,out);				\
     FGrid->Barrier();				\
-    zDw.CayleyZeroCounters();      \
     t0=usecond();				\
     for(int i=0;i<ncall;i++){			\
       zDw. A (in,out);				\
     }						\
     t1=usecond();				\
     FGrid->Barrier();				\
-    zDw.CayleyReport();							\
     std::cout<<GridLogMessage << "Called ZDw " #A " "<< (t1-t0)/ncall<<" us"<<std::endl;\
     std::cout<<GridLogMessage << "******************"<<std::endl;
 
 #define BENCH_DW_SSC(A,in,out)			\
     Dw. A (in,out);				\
     FGrid->Barrier();				\
-    Dw.CayleyZeroCounters();      \
     t0=usecond();				\
     for(int i=0;i<ncall;i++){			\
       __SSC_START ;				\
@@ -140,7 +135,6 @@ int main (int argc, char ** argv)
     }						\
     t1=usecond();				\
     FGrid->Barrier();				\
-    Dw.CayleyReport();					\
     std::cout<<GridLogMessage << "Called " #A " "<< (t1-t0)/ncall<<" us"<<std::endl;\
     std::cout<<GridLogMessage << "******************"<<std::endl;
 

benchmarks/Benchmark_wilson.cc

@@ -155,7 +155,6 @@ int main (int argc, char ** argv)
   //int ncall=1;
 
   // Counters
-  Dw.ZeroCounters();
   Grid.Barrier();
 
   double t0=usecond();
@@ -201,7 +200,6 @@ int main (int argc, char ** argv)
   err = ref-result;
   std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<<std::endl;
 
-  Dw.Report();
   
   // guard
   double err0 = norm2(err);