Sorry, I wanted to say, that the imposition of boundary
conditions in the setup is apparently dominant.
Is this the general case in your applications?
Thank You
Marek
2016-07-30 13:09 GMT+02:00 Vinetou Incucuna <fregew...@gmail.com>:
> ah, sorry for interruption.
> From the log it is evident, than the Navier-Stokes
> equation setup times are reasonable and that the
> constraint matrix preparation, including Boundary
> conditions.
> Is this evaluation sound, should I strive for better times?
>
> Thank You
>
>
> Marek
>
> 2016-07-30 13:06 GMT+02:00 Vinetou Incucuna <fregew...@gmail.com>:
>
>> Hello,
>> problem was,
>>
>> - that I have some meaningless calls for
>>
>> /* current_solution_phase.reinit(locally_owned_dofs_phase,
>> locally_relevant_dofs_phase, mpi_communicator);
>> current_solution_phase = 0;
>> */
>> /*solution_update.reinit(locally_owned_dofs_phase,
>> locally_relevant_dofs_phase, mpi_communicator);
>> solution_update = 0;
>>
>> These vectors remained in my solver from the previous development.
>>
>> - I ran the setup in the debug mode, after the switch to release mode
>> I reached the following results
>>
>> started on 4x24 cores
>> 549250 dofs for Cahn Hilliard system
>>
>> dof renumbering, wall time: 0.194882s.
>> reinit of system matrix, wall time: 0.025027s.
>> Setup CH, wall time: 0.464638s.
>>
>> 21841796 dofs for Navier-Stokes system
>> constraint matrix preparation, wall time: 92.1981s.
>> computing_timer.enter_subsection("constraint matrix preparation");
>> constraint_matrix_nse.clear();
>> constraint_matrix_nse.reinit(locally_relevant_dofs_nse);
>> DoFTools::make_hanging_node_constraints(dof_handler_nse,
>> constraint_matrix_nse);
>> std::vector<bool> component_mask(dim + 1, false);
>> for (int i = 0; i < dim; ++i)
>> component_mask[i] = true; // velocities
>>
>> VectorTools::interpolate_boundary_values(dof_handler_nse, 0,
>> ConstantFunction<dim>(0., dim + 1), constraint_matrix_nse,
>> component_mask);
>>
>> VectorTools::interpolate_boundary_values(dof_handler_nse, 1,
>> InflowVelocityBoundaryValues<dim>(),
>> constraint_matrix_nse,
>> component_mask);
>>
>> /* std::set<types::boundary_id> bound_set;
>> bound_set.insert(1);
>>
>> VectorTools::compute_no_normal_flux_constraints(dof_handler_nse,
>> 0,
>> bound_set, constraint_matrix_nse);
>> */
>> constraint_matrix_nse.close();
>> computing_timer.leave_subsection();
>>
>>
>>
>> making sparsity pattern, wall time: 3.96472s.
>> reinit of system matrix NS, wall time: 3.58887s.
>> Setup NS, wall time: 134.87s.
>>
>
>
--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see
https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
