[petsc-users] Prometheus vs GAMG for elasticity/plasticity problems

[Jed Brown]

Thomas Gross <tgross at ilsb.tuwien.ac.at> writes:

> Dear petsc users/developers,
>
> I am using petsc in a parallel FEAP framework (http://www.ce.berkeley.edu/projects/feap/) for mechanics oriented problems (mainly linear elasticity or plasticity).
> In earlier petsc versions (3.2.7) I always used the petsc options:
> -ksp_type cg -pc_type prometheus (with standard settings) and achieved a great level of performance.
>
> After updating to petsc 3.4.3, where the “Prometheus” preconditioner was replaced by “GAMG”, I can not reach the same level of performance. My settings are:
> -ksp_type cg -pc_type gamg -pc_gamg_type agg -pc_gamg_agg_nsmooths 1 (settings recommended for FEAP).
>
> Is there a set of GAMG settings, which are equivalent to the standard Prometheus settings?

Please send -log_summary -ksp_monitor -ksp_view output for both cases.
Make sure to use the same smoother you used with Prometheus.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140127/585d8c5d/attachment.pgp>