Was the problem due to PARMETIS ordering? Setting colperm = NATURAL may incur lots of fill-ins. Also, I notice that the matrix is very small to ask to use 64 processors.
Sherry On Thu, Aug 25, 2011 at 10:46 AM, Hong Zhang <hzhang at mcs.anl.gov> wrote: > On Thu, Aug 25, 2011 at 12:45 PM, Gabriele Jost <gjost at tacc.utexas.edu> > wrote: > > Thanks Hong. I am not sure I can switch, don't see the module available > at my site. > > Can I use the flags you mentioned on the regular petsc 3.1? > Yes. > Hong > > > ________________________________________ > > From: petsc-dev-bounces at mcs.anl.gov [petsc-dev-bounces at mcs.anl.gov] on > behalf of Hong Zhang [hzhang at mcs.anl.gov] > > Sent: Thursday, August 25, 2011 10:33 AM > > To: For users of the development version of PETSc > > Cc: Victor Eijkhout > > Subject: Re: [petsc-dev] KSP Viewer problem in petsc3.1 > > > > Can you switch to petsc-dev? > > Quite few bugs have been fixed in petsc-dev and superlu_dist since > petsc-3.1. > > You may try '-mat_superlu_dist_equil no -mat_superlu_dist_colperm > NATURAL' > > first. > > > > Hong > > > > On Thu, Aug 25, 2011 at 11:19 AM, Satish Balay <balay at mcs.anl.gov> > wrote: > >> Its a patch to petsc - its already be the latest patched tarball - > petsc-3.1-p8.tar.gz > >> [attaching it] > >> > >> Satish > >> > >> On Thu, 25 Aug 2011, Gabriele Jost wrote: > >> > >>> Hello! > >>> > >>> I am running into a problem when using petsc 3.1 and SuperLU. > >>> > >>> KSP Object: > >>> type: preonly > >>> maximum iterations=5000, initial guess is zero > >>> tolerances: relative=1e-06, absolute=1e-50, divergence=10000 > >>> left preconditioning > >>> using PRECONDITIONED norm type for convergence test > >>> PC Object: > >>> type: lu > >>> LU: out-of-place factorization > >>> tolerance for zero pivot 1e-12 > >>> matrix ordering: natural > >>> factor fill ratio given 0, needed 0 > >>> Factored matrix follows: > >>> Matrix Object: > >>> type=mpiaij, rows=13608, cols=13608 > >>> package used to perform factorization: superlu_dist > >>> total: nonzeros=0, allocated nonzeros=27216 > >>> SuperLU_DIST run parameters: > >>> Process grid nprow 8 x npcol 8 > >>> Equilibrate matrix TRUE > >>> Matrix input mode 1 > >>> Replace tiny pivots TRUE > >>> Use iterative refinement FALSE > >>> Processors in row 8 col partition 8 > >>> Row permutation LargeDiag > >>> [59]PETSC ERROR: --------------------- Error Message > ------------------------------------ > >>> [60]PETSC ERROR: Petsc Release Version 3.1.0, Patch 5, Mon Sep 27 > 11:51:54 CDT 2010 > >>> [62]PETSC ERROR: [53]PETSC ERROR: Unknown column permutation! > >>> See docs/changes/index.html for recent updates. > >>> > >>> > >>> I googled around and found that a similar problem had been reported to > the Petsc folks before. The answer was as follows: > >>> > >>> On Wed, 15 Dec 2010, Satish Balay wrote: > >>> > >>> > Can you try the attached patch - and see if this goes away? > >>> > > >>> > patch -Np1 < superlu_dist.patch > >>> > > >>> > >>> > >>> Is the patch above included in the SLU that is part of the petsc 3.1 > distribution? If not, could we get it? > >>> > >>> Thanks so much in advance and best regards, > >>> Gabriele > >>> > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110825/a92821da/attachment.html>