[petsc-users] boomerAmg scalability

[Mohamad M. Nasr-Azadani]

Hi Ravi,

I have been using BoomerAMG as a preconditioner joint with an iterative
solver, e.g. GMRES of BiCGs for regular 3D CFD problems.
On the top of my head, I can not remember if I had the strong scaling tests
done (I will look into it and let you know if you found any), but for the
weak-scaling case, I definitely saw some scaling issues.
As the size of the system increases, the number of iterations does also
increase (somewhat significantly for my test problem, i.e. incompressible
N-S with complex geometry) which ultimately deteriorates the weak-scaling
behaviors.

This is also reported in the reports given by the hypre team, cf.
https://computation.llnl.gov/casc/linear_solvers/pubs/pmis_report.pdf

(cf. see Table 6.1 for the Stokes flow simulation results and scaling).

Also, inherited with the multigrid nature, there are always
fine-tuning factors which are, unfortunately, problem dependent.

Good luck,
Mohamad

On Thu, Dec 15, 2011 at 8:23 AM, Ravi Kannan <rxk at cfdrc.com> wrote:

> Dear All,****
>
> ** **
>
> This is Ravi Kannan from CFD Research Corporation. Recently, we are
> experimenting with the BoomerAMG preconditioner for some “stiff” CFD
> problems. In that regard, all the other standard solver-preconditioner
> combinations failed for the current CFD problem. The boomer is the only one
> which is able to provide with “converged” solutions.****
>
> ** **
>
> We noticed that the scalability of this boomer preconditioner is really
> poor. For instance, even with a cell size of 2 million, we cannot scale to
> even 16 partitions (in contrast, the other solver-preconditioner
> combinations like the BI-CGS/BJacobi gave good enough scalability).****
>
> ** **
>
> Are we missing something? Do we need to use a more latest version of
> boomer?****
>
> ** **
>
> Thanks,****
>
> Ravi. ****
>
> ** **
>
> ** **
>
> ** **
>
> ** **
>
> ** **
>
> *From:* petsc-dev-bounces at mcs.anl.gov [mailto:
> petsc-dev-bounces at mcs.anl.gov] *On Behalf Of *Bobby Philip
> *Sent:* Thursday, December 15, 2011 9:22 AM
> *To:* For users of the development version of PETSc
> *Subject:* Re: [petsc-dev] controlling vector values while doing matrix
> free operations****
>
> ** **
>
> Hmmmm..sorry guys - the entire thread got put away in a folder by my mail
> reader and I just discovered all your emails :-)****
>
> ** **
>
> SNESVI looks interesting but as someone pointed out I am using matrix
> free. And as I understand from Barry a matrix free version of SNESVI is not
> implemented. The idea of flipping the sign seems to be a poor mans approach
> that might work for me though I would still need some mechanisms in SNES to
> put the hooks in.****
>
> ** **
>
> Bobby****
>
> ** **
>
> On Dec 14, 2011, at 10:45 PM, Jed Brown wrote:****
>
>
>
> ****
>
> On Wed, Dec 14, 2011 at 19:40, Dmitry Karpeev <karpeev at mcs.anl.gov> wrote:
> ****
>
> The trouble is that the constraints can get "eliminated" only when they
> become active. ****
>
> ** **
>
> That's not what I meant. I meant to transform the algebraic system so that
> those extra variables were eliminated. The point is that we tend to put a
> lot of effort into designing effective preconditioners for a standard
> formulation (e.g. conservative variables), and that is partly lost of we
> have this other system in which the evaluations of constitutive relations
> are added explicitly.****
>
> ** **
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20111219/3a5362c1/attachment-0001.htm>