[CIG-SHORT] Storing intermediate results

[Brad Aagaard]

Xiao,

PyLith does not currently support checkpointing.

You should consider whether you really need a uniform resolution mesh. 
You may be able to create a mesh with the same accuracy but many fewer 
DOF with a variable resolution mesh.

You should also write out the PETSc log summary --petsc.log_summary to 
see if there are obvious bottlenecks. For example, ASCII I/O is much 
slower then HDF5 I/O.

Regards,
Brad



On 12/11/2014 02:16 PM, Ma, Xiao wrote:
> Hi,
> I am modeling a pretty large problem (2D problem 100kmX30km Domain with 12.5m uniformly mesh). I am using a cluster system to run the simulations, however there is a wall-time limit for the cluster. I am wondering can Pylith store the solutions and use them in the next run (a new wall-time)?
> Best,
> Xiao
>
>
>
> _______________________________________________
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>