On Fri, Mar 16, 2012 at 13:36, Pham Van <pham_ha at yahoo.com> wrote:
> Thank Jed for very quick reply. > > The two magic lines did the trick. The question is why they have been > excluded? For me its just fine right now. > I assume because people reported that it didn't work. I have enabled them again in petsc-dev. > > I have one question thought: I have unsymmetric matrix. It is solved just > fine with 1 processor, but if I increased to 2 processors mumps produce the > following error (the matrix is solved just fine when I use superlu_dist > package both uni and multi-proccesors, the problem is superlu_dist is too > slow): > > [0]PETSC ERROR: --------------------- Error Message > ------------------------------------ > [0]PETSC ERROR: Error in external library! > [0]PETSC ERROR: Error reported by MUMPS in numerical factorization phase: > INFO(1)=-9, INFO(2)=2055660 > As far as we can tell, MUMPS is competing for the most arcane error reporting system. You have to read the user's manual to translate those numbers into anything understandable. ?9 Main internal real/complex workarray S too small. If INFO(2) is positive, then the number of entries that are missing in S at the moment when the error is raised is available in INFO(2). If INFO(2) is negative, then its absolute value should be multiplied by 1 million. If an error ?9 occurs, the user should increase the value of ICNTL(14) before calling the factorization (JOB=2) again, except if ICNTL(23) is provided, in which case ICNTL(23) should be increased. You can use -mat_mumps_icntl_14 to control that parameter. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20120316/e5fec6b8/attachment.htm>
