Erasmus Langer
Siegfried Selberherr
Bindu Balakrishna
Oskar Baumgartner
Hajdin Ceric
Johann Cervenka
Otmar Ertl
Wolfgang Gös
Klaus-Tibor Grasser
Philipp Hehenberger
René Heinzl
Hans Kosina
Goran Milovanovic
Neophytos Neophytou
Roberto Orio
Vassil Palankovski
Mahdi Pourfath
Karl Rupp
Franz Schanovsky
Philipp Schwaha
Ivan Starkov
Franz Stimpfl
Viktor Sverdlov
Oliver Triebl
Stanislav Tyaginov
Martin-Thomas Vasicek
Stanislav Vitanov
Paul-Jürgen Wagner
Thomas Windbacher

Franz Stimpfl
Dipl.-Ing.
stimpfl(!at)iue.tuwien.ac.at
Biography:
Franz Stimpfl was born in Vienna, Austria, in 1980. He studied computer science at the Technische Universität Wien, where he received the degree of Diplomingenieur in 2007. He joined the Institute for Microelectronics in October 2007, where he is currently working on his doctoral degree. His research activities include mesh generation and modern software paradigms.

Parallel Multi-Segment Mesh Generation

The wide variety of different simulations imposes a multitude of constraints and requirements on mesh generation, not only from Technology Computer Aided Design (TCAD), but also in the field of scientific computing as a whole. Driven by the increasing availability of computational power, the growth of complexity in physical models and, due to size reduction, more accurate geometrical features further intensify the importance of mesh generation.
To cope with increasing requirements of simulations, inputs restricted to manifold surfaces are no longer sufficient and a shift towards multiple segment meshes, thus resulting in non-manifold surfaces, is in progress.
We propose an approach for meshing multiple segments, by decomposing the non-manifold surface into separate manifold segments. However, in following this approach, the interfaces between segments pose a problem. The interfaces need to be unique, ensure consistency, and, as a result of consistency and uniqueness, only need to be meshed once.
The segmentation of the surface and volume meshing process accommodates the possibility of increased use of parallelization. By utilizing our approach only a few synchronization points are needed during the meshing process, such as the creation of the irreducible edges, the construction of volume segments using charts as boundaries, and the merging of the meshed volumes.
Since the meshing process of non-manifold surfaces is broken down to manifolds, existing meshing algorithms, which have already been used, tested, and their results verified, can be applied. Our approach supports exchange of the actual meshing kernel and adaptation to the needs of the application, due to the utilization of appropriate programming paradigms to facilitate a modular design.
Besides the programming paradigms, the right choice of the underlying datastructure is of utmost importance. Topological operations, such as the boundary and the co-boundary, are applied to identify irreducible edges. Therefore a solid topological framework, the Generic Scientific Simulation Environment (GSSE), is used as the foundation of the meshing process.


An example of a multi-segment CMOS device, which can be meshed in parallel.


Home | Activities | Staff | Publications | Sponsors | Music | Contact Us