Overview

Modern simulation methods require parallel high-performance computers and suitable software environments. In accordance with our mission, IMCS is dedicated to international top-level research in the field of numerical mathematics and computer-aided simulation in engineering, which is why both local, exclusively available hardware equipment and research software that we have developed ourselves and that can be massively parallelized are of crucial importance. Last but not least, our research partners from science and industry also benefit from the excellent technical resources that are currently installed at the University of the Federal Armed Forces in Munich and that are available to our partners within the framework of research cooperations.
 

Hardware Resources

Currently, a high-performance computer (Linux cluster) for parallel calculations within the framework of finite element methods (FEM) and other discretization methods is being installed at IMCS. Commissioning is planned for the end of 2018. The most important technical data of the planned supercomputer are:

  • Linux cluster of compute nodes with multi-core CPUs
  • e.g. 480 cores in 20 Intel Xeon Skylake-SP nodes (2x12 core)
  • e.g. 196 GB DDR4 RAM per node
  • e.g. Melllanox Infiniband Network
  • e.g. rack cabinet with integrated water cooling

 

40-IMG_Server.jpg 21-IMG_Netzwerk.jpg

 

Research Software BACI

Together with other research partners, our institute is developing the parallel multiphysics research code BACI ("Bavarian Advanced Computational Initiative"). BACI was established more than 10 years ago at the Institute for Computational Mechanics at the Technical University of Munich (Prof. Wolfgang A. Wall) and has developed into one of the world's leading research codes in the field of computational solid mechanics and fluid dynamics and in particular for coupled multi-field problems. Large parts of BACI are based on finite element methods (FEM), but alternative discretization methods such as discontinuous Galerkin methods (DG), particle methods and mesh-free methods have also been successfully integrated. The research software is implemented throughout in object-oriented programming (C++) using modern software design and is parallelized with MPI for distributed memory hardware architectures.

 

compsim04.png  compsim06.png