On Tue, Feb 14, 2017 at 4:59 AM, Lailai Zhu <lailaizh...@gmail.com> wrote:
> hi, Proveen, > > After implementing and testing by following your solutions, I realized two > issues. > When you use the mumps solver, how does it scale? i mean with a couple of > processors, say 4. > In my case, I tried to solve a test Laplace equations, the time of solving > the system increases > when only using two processors compared to one. If i use the cg solver, > then solving process > roughly scales to 6 processors, not so nice, but still improves. > > I dont have any numbers regarding parallel performance. I will tell you if I get any numbers on this. In my HDG code for compressible Navier-Stokes, I found gmres to be faster than mumps. So I use mumps only when I am not too particular about performance. > The second question is about the ->initialize function, I only see the > constructor and solve > member functions for trilino's direct solver? > I forgot about this. The initialize function is currently in the master branch on github. See https://www.dealii.org/developer/doxygen/deal.II/classTrilinosWrappers_1_1SolverDirect.html You will have to compile the deal.II master branch to use this function. Best praveen -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
