search

KratosMultiphysics / Kratos

Kratos Multiphysics (A.K.A Kratos) is a framework for building parallel multi-disciplinary simulation software. Modularity, extensibility and HPC are the main objectives. Kratos has BSD license and is written in C++ with extensive Python interface.
https://kratosmultiphysics.github.io/Kratos/
Other
1.01k stars 244 forks source link

Modifying analysis to run using MPI on a cluster #1771

Closed jaabell closed 6 years ago

jaabell commented 6 years ago

Dear Kratos devs,

I could use some help here. I have a kratos model using PFEM FSI, developed in GiD, which I can run on my laptop using OpenMP. I wish to run this same example, and a finer version of it, on a cluster using MPI and MPI/OpenMP parallelism. I have already compiled Kratos and all dependencies using intel compilers on the cluster and the test runs fine.

How do I go about modifying the ProjectParameters.json and MainKratos.py files to use MPI? I've attached the files I have right now.... PFEM_MalladoGrueso.zip

Also related: How can I find out what options are available on the ProjectParameters.json file? Is there any documentation specifying everything available?

Thanks! Regards from Chile

Jose Abell

pooyan-dadvand commented 6 years ago

As far as I know the PFEM application is not mpi compatible and it’s actual remeshing step makes it difficult to be ported

Meanwhile I would refer you to @KratosMultiphysics/pfem which maintain this app

maceligueta commented 6 years ago

Some months back we discovered a library that could do the job in MPI. Tess2 by Diatomic, in GitHub. However, no more efforts were dedicated in this direction so far.

jaabell commented 6 years ago

Hi Pooyan and Miguel Angel. Thanks for your replies! I recently saw a presentation/paper on PFEM2... how is that coming along?

maceligueta commented 6 years ago

Let me reference @pablobecker for this.

pablobecker commented 6 years ago

I am no longer working on it, but I know @miguelmaso and @dcagri are using it, altough for their specific research and certainly not in MPI. The problem is that i never ported the movement of particles to MPI. Juan Marcelo Gimenez did the porting, but the current implementation uses OpenFoam instead of Kratos FEM.

pablobecker commented 6 years ago

If you are interested in geomechanics applications, why don't you take a look at the MPM application ? ping @iiaconeta

philbucher commented 6 years ago

Closing since inactive Please reopen otherwise