[petsc-users] command line option for superlu_dist
Matthew Knepley
knepley at gmail.com
Tue Jan 5 12:13:45 CST 2010
1) The solution norm you report is NaN, which indicates memory corruption in
your code
2) The number of allocated nonzeros is really suspicious. There are only 2
nonzeros per row?
Matt
On Tue, Jan 5, 2010 at 12:01 PM, <hxie at umn.edu> wrote:
>
> The information for superlu_dist
> Linear solve converged due to CONVERGED_ITS iterations 1
> KSP Object:
> type: preonly
> maximum iterations=1000, initial guess is zero
> tolerances: relative=1e-06, absolute=1e-10, divergence=100
> left preconditioning
> PC Object:
> type: lu
> LU: out-of-place factorization
> matrix ordering: natural
> LU: tolerance for zero pivot 1e-12
> LU: factor fill ratio needed 0
> Factored matrix follows
> Matrix Object:
> type=mpiaij, rows=291896, cols=291896
> package used to perform factorization: superlu_dist
> total: nonzeros=0, allocated nonzeros=583792
> SuperLU_DIST run parameters:
> Process grid nprow 2 x npcol 2 Equilibrate matrix
> TRUE Matrix input mode 1 Replace tiny pivots TRUE
> Use iterative refinement FALSE Processors in row 2 col
> partition 2 Row permutation LargeDiag Column
> permutation MMD_AT_PLUS_A
> Parallel symbolic factorization FALSE Repeated
> factorization SamePattern
> linear system matrix followed by preconditioner matrix:
> Matrix Object:
> type=mpiaij, rows=291896, cols=291896
> total: nonzeros=33670280, allocated nonzeros=36641300
> using I-node (on process 0) routines: found 21607 nodes, limit used is
> 5
> Matrix Object:
> type=mpiaij, rows=291896, cols=291896
> total: nonzeros=33670280, allocated nonzeros=36641300
> using I-node (on process 0) routines: found 21607 nodes, limit used is
> 5
>
> WARNING: Tolerance not met by convergence norms
>
> -----------------------------------------------------------------------------
> Solution norm = NaN
> Total CPU time = 39.0 min
>
> For mumps, it takes only 2min to get some result "Solution norm =
> 0.255E+03".
>
>
>
--
What most experimenters take for granted before they begin their experiments
is infinitely more interesting than any results to which their experiments
lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20100105/e046d246/attachment.htm>
More information about the petsc-users
mailing list