On Thu, 21 Apr 2011, Kirk, Benjamin (JSC-EG311) wrote:

> All that is needed is to have redistribute pack and unpack both the old and
> current DOF information as well?

I believe so.

> I won't be as busy as you next week - hell I might be able to work on this
> rather than heckling you...

I'm hoping you'll be nearly as busy as me a little while longer.  ;-)
My talk is going to brag about our robustness heuristics, but none of
those has been nearly as reliable for hypersonics convergence as "let
Ben write the config file".  And even still we're not quite to a
baseline capsule solution with turbulence and/or ablation.  Definitely
won't to have UQ results finished by next Thursday (are you coming to
the visit?) at this rate but I'm hoping to at least have them queued.

> Not promising anything, but if you give me a simple procedure and
> patches from HEAD to cause the error I can take a shot at
> redistributing the DofIndices as well and see if that fixes it.

src/mesh/parallel_mesh.C, ParallelMesh::partition() has our
workaround.  Comment out the "if(!serial) return" lines there, and do
a "LIBMESH_RUN='mpirun -np 6' METHOD=dbg make run_examples" to trigger
the problem.  (I think you can even get the failure from -np 2).
---
Roy

------------------------------------------------------------------------------
Fulfilling the Lean Software Promise
Lean software platforms are now widely adopted and the benefits have been 
demonstrated beyond question. Learn why your peers are replacing JEE 
containers with lightweight application servers - and what you can gain 
from the move. http://p.sf.net/sfu/vmware-sfemails
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to