[petsc-users] memory increase in a transient calculation but no valgrind errors

[Matthew Knepley]

On Sat, Dec 24, 2011 at 9:36 AM, Dominik Szczerba <dominik at itis.ethz.ch>wrote:

> I am doing a transient calculation, allocating and deallocating
> several matrices and vectors both between nonlinear iterations as well
> as between time steps.
> I measure (with dstat) a constant linear memory increase as the
> calculation progresses. I have run my solver on a small example
> through valgrind, however, no leaks / unfreed memory have been
> reported. Are there any other possibilities to find out where the
> memory goes to?
>
> Many thanks for any hints.


Its always possible that dstat is showing memory fragmentation. If the
problems are the same size, you might benefit from pooling.

   Matt


>
> Dominik

-- 
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/20111224/64163266/attachment.htm>