On Thu, Feb 26, 2015 at 8:36 AM, Martin Sandve Alnæs <[email protected]> wrote:
> This doesn't happen on the buildbots, is it something about my installation?
> Does anyone else get this? You don't have to run all tests to check, just
> cd <dolfin>
> cd test/unit/python
> py.test la/test_krylov_solver.py
>
>
>>               assert niter < 12
> E               assert 15 < 12
>
> la/test_krylov_solver.py:119: AssertionError
> -------------------------------------------------------------------------------------------
> Captured stdout call
> --------------------------------------------------------------------------------------------
> Testing method 'petsc_amg' with 4 x 4 mesh
> Solving linear system of size 50 x 50 (PETSc Krylov solver).
>   3 iterations.
> Testing method 'petsc_amg' with 8 x 8 mesh
> Solving linear system of size 162 x 162 (PETSc Krylov solver).
>   4 iterations.
> Testing method 'petsc_amg' with 16 x 16 mesh
> Solving linear system of size 578 x 578 (PETSc Krylov solver).
>   6 iterations.
> Testing method 'petsc_amg' with 32 x 32 mesh
> Solving linear system of size 2178 x 2178 (PETSc Krylov solver).
>   8 iterations.
> Testing method 'petsc_amg' with 64 x 64 mesh
> Solving linear system of size 8450 x 8450 (PETSc Krylov solver).
>   10 iterations.
> Testing method 'ml_amg' with 4 x 4 mesh
> Solving linear system of size 50 x 50 (PETSc Krylov solver).
>   4 iterations.
> Testing method 'ml_amg' with 8 x 8 mesh
> Solving linear system of size 162 x 162 (PETSc Krylov solver).
>   6 iterations.
> Testing method 'ml_amg' with 16 x 16 mesh
> Solving linear system of size 578 x 578 (PETSc Krylov solver).
>   8 iterations.
> Testing method 'ml_amg' with 32 x 32 mesh
> Solving linear system of size 2178 x 2178 (PETSc Krylov solver).
>   11 iterations.
> Testing method 'ml_amg' with 64 x 64 mesh
> Solving linear system of size 8450 x 8450 (PETSc Krylov solver).
>   15 iterations.

I get the same error. The reason it doesn't fail on the buildbots is
probably because none of them have PETSc built with ml.

Johannes
_______________________________________________
fenics mailing list
[email protected]
http://fenicsproject.org/mailman/listinfo/fenics

Reply via email to