dolfin team mailing list archive

["Garth N. Wells" <g.n.wells@xxxxxxxxxx>]

We're working on parallel assembly and solution at the moment. It will 
use PETSc, and should in place in the not too distant future.

I just fixed the test program src/test/passembly/main.cpp in the 
development version. It should compile now. It is very much a test 
program, so it's pretty rough.

If you have any further questions, dolfin-dev@xxxxxxxxxx is the best 
place to ask.

Garth

Michael Ulbrich wrote:
> Hi,
>
> I have a question concerning fenics/dolfin:
>
> You write on the dolfin homepage that it can use parallel
> linear algebra via PETSc.
> I compiled dolfin with petsc und mpi enabled, but,
> unfortunately, I am not able to run, e.g.,
>
> src/demo/solvers/elasticity-updated/dolfin-elasticity-updated
>
> with parallel performance. If I run it with (I use mpich
> from the petsc externalpackages directory)
>
> mpiexec -n 4 ./dolfin-elasticity-updated
>
> it is started four times, but does not seem to do parallel
> linear algebra. It just seems to compute four times the same.
> Am I doing something wrong or is parallel PETSc not yet supported?
>
> I also tried passembly, but got a compiler error message
> that a.update(map) is not available.
>
> Thanks and best regards,
>
> Michael
>
> ---------------------------------------------------------------------
> Prof. Dr. Michael Ulbrich
> Chair of Mathematical Optimization
> Center for Mathematical Sciences
> M1 - Optimization
> TU Muenchen
> Boltzmannstr. 3
> 85747 Garching b. Muenchen
> Germany
> Phone: +49 89 289-17931
> Fax:   +49 89 289-17932
> Email: mulbrich@xxxxxxxxx
> ---------------------------------------------------------------------