> Also note that while a block-ILU preconditioner will reduce the time spent > per solver iteration, the total number of iterations may increase.
I know the problem. I have already used pARMS from Saad which is the only parallel ILU0 solver that I know of which scales nicely. Unfortunately, it requires MPI which requires a lot of effort on my side as my program is not written with MPI. Anyway, this is the computation of the ILU0 which does not scale, not the iteration process. So it must be the overhead you were talking about. > FYI: The next release will bring a massively parallel ILU0 preconditioner > [1]. I can point you to the feature branch in the developer repository on > GitHub if you don't want to wait. I'll be glad to try it. Can you give me the branch ? Is it documented ? Is it OpenMP ready ? ------------------------------------------------------------------------------ _______________________________________________ ViennaCL-support mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/viennacl-support
