Enable OpenMP parallelization of row-major-sparse * dense products.
I observed significant speed-up of the CG solver.
1 file changed