I am using `Eigen 3.3.7` and using it to solve a linear system and am using the sparse matrix and solver facilities within Eigen.

So, I have openMP enabled and querying the number of threads returns the number of threads I have setup using openMP.

Now, the setup is as follows:

Code: Select all

Eigen::SparseMatrix<double> A;
// Fill the matrix
A.setFromTriplets(tripletList.begin(), tripletList.end());

vector_t b_vec;
// fill the vector
Eigen::ConjugateGradient<Eigen::SparseMatrix<double> > cg;
cg.setMaxIterations(256);
cg.setTolerance(0.001);
cg.compute(A);
x = cg.solve(b_vec);


I am not seeing anything in terms of computation time with OpenMP and I was wondering if I need to do anything else or if it is even possible to gain some speedup through parallel computation with this solver?

according to https://eigen.tuxfamily.org/dox/classEi ... dient.html

you have to set the _UpLo template parameter, because the default is that the triangular part that will be used for the computations. It can be Lower, Upper, or Lower|Upper in which the full matrix entries will be considered. Default is Lower, best performance is Lower|Upper.