Ex8 and minor improvements

sheet3/345/benchmark.cpp

@@ -82,8 +82,12 @@ vector<double> benchmark_D(const vector<double>& coeff, const vector<double>& x)
 double norm2(const vector<double>& x)
 {
     double s = 0.0;
-    for (unsigned int i = 0; i < x.size(); ++i)
-        s += x[i]*x[i];
+    double xi;
+    for (unsigned int i = 0; i < x.size(); ++i){
+        xi = x[i];
+        s += xi*xi;
+    }
+        
     return sqrt(s);
 }
 
@@ -116,7 +120,7 @@ vector<double> matrixMultColumnWise(const vector<double> &A, const vector<double
             double sum = 0.0;
             for (unsigned int k = 0; k < L; k++)
             {
-                sum += A[k*L+i]*B[k*N+j];
+                sum += A[k*M+i]*B[k*N+j];
             }
             C[i*N+j] = sum;
         }

sheet3/345/main.cpp

@@ -131,6 +131,7 @@ int main(int argc, char **argv)
     double bytesC = (MC * LC  + LC * NC  + MC * NC)* sizeof(double);
 
     cout << "\n===== Benchmark C =====\n";
+    cout << guardC << endl;
     cout << "bytes: " << bytesC << endl;
     cout << "Timing in sec. : " << tC << endl;
     cout << "GFLOPS         : " << flopsC / tC / 1024 / 1024 / 1024 << endl;
@@ -172,6 +173,7 @@ int main(int argc, char **argv)
     double bytesD = (p + 2 * ND)*sizeof(double);
 
     cout << "\n===== Benchmark D =====\n";
+    cout << guardD << endl;
     cout << "bytes: " << bytesD << endl;
     cout << "Timing in sec. : " << tD << endl;
     cout << "GFLOPS         : " << flopsD / tD / 1024 / 1024 / 1024 << endl;
@@ -208,6 +210,8 @@ int main(int argc, char **argv)
     cout << "GFLOPS         : " << 2.0 * NA / tA / 1024 / 1024 / 1024 << endl;
     cout << "GiByte/s       : "
          <<  NA * sizeof(xA[0]) / tA / 1024 / 1024 / 1024 << endl;
+
+    //a bit faster due to only accessing one vector 
 }
 
 
@@ -240,6 +244,8 @@ int main(int argc, char **argv)
     cout << "GFLOPS         : " << 5.0 * NA / tA / 1024 / 1024 / 1024 << endl;
     cout << "GiByte/s       : "
          << 2.0 * NA * sizeof(xA[0]) / tA / 1024 / 1024 / 1024 << endl;
+
+    //in comparison to benchmark A: a bit slower runtime but more than double the amount of FLOPS therefor also more GFLOPS
 }
 
 
@@ -276,11 +282,16 @@ int main(int argc, char **argv)
     double bytesC = (MC * LC  + LC * NC  + MC * NC)* sizeof(double);
 
     cout << "\n===== Benchmark 5C =====\n";
+    cout << guardC << endl;
     cout << "bytes: " << bytesC << endl;
     cout << "Timing in sec. : " << tC << endl;
     cout << "GFLOPS         : " << flopsC / tC / 1024 / 1024 / 1024 << endl;
     cout << "GiByte/s       : " << bytesC / tC / 1024 / 1024 / 1024 << endl;
+
+    //slower than rowwise access, due to incoherent acces in the vector memory of A
+    
+    //Transpose matrix, the it is also row wise-access or reorder loops 
 }
 
 return 0;
-}  // memory for x and y will be deallocated by their destructors
+}  

sheet3/345/output.txt

@@ -0,0 +1,51 @@
+g++ -c -g -O0 -funroll-all-loops -std=c++17 -Wall -pedantic -Wextra -Weffc++ -Woverloaded-virtual -Wfloat-equal -Wshadow -Wredundant-decls -Winline -fmax-errors=1  -flto -o main.o main.cpp
+g++ -c -g -O0 -funroll-all-loops -std=c++17 -Wall -pedantic -Wextra -Weffc++ -Woverloaded-virtual -Wfloat-equal -Wshadow -Wredundant-decls -Winline -fmax-errors=1  -flto -o mylib.o mylib.cpp
+g++ -c -g -O0 -funroll-all-loops -std=c++17 -Wall -pedantic -Wextra -Weffc++ -Woverloaded-virtual -Wfloat-equal -Wshadow -Wredundant-decls -Winline -fmax-errors=1  -flto -o benchmark.o benchmark.cpp
+g++  main.o mylib.o benchmark.o  -g -O0  -llapack -lblas -flto -o main.GCC_
+./main.GCC_ 
+
+===== Benchmark A =====
+<xA,yA> = 1.4e+06
+Timing in sec. : 0.00893637
+GFLOPS         : 0.291808
+GiByte/s       : 2.33446
+
+===== Benchmark B =====
+340000
+bytes: 2.31472e+07
+Timing in sec. : 0.0133897
+GFLOPS         : 0.402029
+GiByte/s       : 1.61001
+
+===== Benchmark C =====
+7.37196e+07
+bytes: 2.4e+07
+Timing in sec. : 8.67235
+GFLOPS         : 0.21478
+GiByte/s       : 0.00257736
+
+===== Benchmark D =====
+10500
+bytes: 3.20001e+07
+Timing in sec. : 0.101087
+GFLOPS         : 0.515935
+GiByte/s       : 0.294821
+
+===== Benchmark 5A =====
+NORM = 150114
+Timing in sec. : 0.00703533
+GFLOPS         : 0.370658
+GiByte/s       : 1.48263
+
+===== Benchmark 5B =====
+<xA,yA> = 1.4e+06
+Timing in sec. : 0.0108377
+GFLOPS         : 0.601533
+GiByte/s       : 1.92491
+
+===== Benchmark 5C =====
+7.37196e+07
+bytes: 2.4e+07
+Timing in sec. : 15.2407
+GFLOPS         : 0.122215
+GiByte/s       : 0.00146658