Task 5, 5* some fixes and cleanup

sheet5/5/getmatrix.cpp

@@ -8,6 +8,7 @@
 #include <iostream>
 #include <list>
 #include <vector>
+#include <omp.h>
 using namespace std;
 
 
@@ -168,33 +169,64 @@ void CRS_Matrix::Debug() const
 
 void CRS_Matrix::CalculateLaplace(vector<double> &f)
 {
+    double t_start = omp_get_wtime();
+
+    double t0 ,t1,t2;
     assert(_mesh.NdofsElement() == 3);               // only for triangular, linear elements
     //cout << _nnz << " vs. " << _id[_nrows] << "  " << _nrows<< endl;
     assert(_nnz == _id[_nrows]);
 
-    for (int k = 0; k < _nrows; ++k)
+    #pragma omp parallel
     {
-        _sk[k] = 0.0;
-    }
-    for (int k = 0; k < _nrows; ++k)
-    {
-        f[k] = 0.0;
-    }
+        #pragma omp for
+        for (int k = 0; k < _nrows; ++k)
+            {
+                _sk[k] = 0.0;
+            }
 
-    double ske[3][3], fe[3];
-    //  Loop over all elements
-    auto const nelem = _mesh.Nelems();
-    auto const &ia   = _mesh.GetConnectivity();
-    auto const &xc   = _mesh.GetCoords();
+        #pragma omp barrier
+        #pragma omp single
+        t0 = omp_get_wtime();
+        
+
+        #pragma omp for   
+        for (int k = 0; k < _nrows; ++k)
+        {
+            f[k] = 0.0;
+        }
+
+        #pragma omp barrier
+        #pragma omp single
+        t1 = omp_get_wtime();
+       
+
+        double ske[3][3], fe[3];
+        //  Loop over all elements
+        auto const nelem = _mesh.Nelems();
+        auto const &ia   = _mesh.GetConnectivity();
+        auto const &xc   = _mesh.GetCoords();
+
+        #pragma omp barrier
+        #pragma omp single
+        t2 = omp_get_wtime();
+        
+        #pragma omp for
+        for (int i = 0; i < nelem; ++i)
+        {
+            CalcElem(ia.data() + 3 * i, xc.data(), ske, fe);
+            AddElem_3(ia.data() + 3 * i, ske, fe, f);
+        }
+        //Debug();
 
-    for (int i = 0; i < nelem; ++i)
-    {
-        CalcElem(ia.data() + 3 * i, xc.data(), ske, fe);
-        AddElem_3(ia.data() + 3 * i, ske, fe, f);
     }
-    //Debug();
+    double t3 = omp_get_wtime();
+    cout << "Zero matrix: " << (t0 - t_start) << " sec\n";
+    cout << "Zero RHS: " << (t1 - t0)     << " sec\n";
+    cout << "Element loop: " << (t3 - t2)    << " sec\n";
+    cout << "Total assembly:" << (t3 - t_start) << " sec\n";
+   
 
-    return;
+    
 }
 
 void CRS_Matrix::ApplyDirichletBC(std::vector<double> const &u, std::vector<double> &f)
@@ -286,7 +318,8 @@ void CRS_Matrix::Defect(vector<double> &w,
 {
     assert( _nrows == static_cast<int>(w.size()) );
     assert( w.size() == u.size() && u.size() == f.size() );
-
+    
+    #pragma omp parallel for
     for (int row = 0; row < _nrows; ++row)
     {
         double wi = f[row];
@@ -340,8 +373,10 @@ void CRS_Matrix::AddElem_3(int const ial[3], double const ske[3][3], double cons
                 assert(ip >= 0);
             }
 #endif
+            #pragma omp atomic
             _sk[ip] += ske[i][j];
         }
+        #pragma omp atomic
         f[ii] += fe[i];
     }
 }

sheet5/5/jacsolve.cpp

@@ -23,6 +23,7 @@ void JacobiSolve(CRS_Matrix const &SK, vector<double> const &f, vector<double> &
 
     cout << endl << " Start Jacobi solver for " << nrows << " d.o.f.s"  << endl;
     //  Choose initial guess
+    #pragma omp parallel for
     for (int k = 0; k < nrows; ++k)
     {
         u[k] = 0.0;                          //  u := 0
@@ -44,6 +45,7 @@ void JacobiSolve(CRS_Matrix const &SK, vector<double> const &f, vector<double> &
     // Iteration sweeps
     int iter  = 0;
     double sigma = sigma0;
+    // i dont parallelize this cause its iteration dependent
     while ( sigma > tol2 * sigma0 && maxiter > iter)
     {
         ++iter;

sheet5/5/main.cpp

@@ -11,6 +11,7 @@
 #include <chrono>             // timing
 #include <cmath>
 #include <iostream>
+#include <omp.h>
 using namespace std;
 using namespace std::chrono;  // timing
 
@@ -19,7 +20,6 @@ int main(int, char ** )
 {
     const int numprocs = 1;
     const int myrank   = 0;
-
     if (myrank == 0)
     {
         cout << "\n There are " << numprocs << " processes running.\n \n";
@@ -121,6 +121,7 @@ int main(int, char ** )
 
             //mesh.Write_ascii_matlab("uv.txt", uv);
             //mesh.Visualize(uv);
+            
         }
         return 0;
     }

sheet5/5/vdop.cpp

@@ -11,7 +11,7 @@ void vddiv(vector<double> &x, vector<double> const &y,
 {
     assert( x.size() == y.size() && y.size() == z.size() );
     size_t n = x.size();
-
+    #pragma omp parallel for
     for (size_t k = 0; k < n; ++k)
     {
         x[k] = y[k] / z[k];
@@ -26,7 +26,7 @@ void vdaxpy(std::vector<double> &x, std::vector<double> const &y,
 {
     assert( x.size() == y.size() && y.size() == z.size() );
     size_t n = x.size();
-
+    #pragma omp parallel for
     for (size_t k = 0; k < n; ++k)
     {
         x[k] = y[k] + alpha * z[k];
@@ -41,6 +41,7 @@ double dscapr(std::vector<double> const &x, std::vector<double> const &y)
     size_t n = x.size();
 
     double    s = 0.0;
+     #pragma omp parallel for reduction(+:s)
     for (size_t k = 0; k < n; ++k)
     {
         s += x[k] * y[k];