Zur Erlangung des akademischen Grades eines Diplom-Ingenieur (FH) über das Thema Simulation of Advanced Large-Scale HPC Architectures

The rapid development of massively parallel systems in the High Performance Computing (HPC) area requires good and efficient scalability of the applications. The next generation’s design of supercomputer is today not certain in terms of what will be the computational resources, memory and I/O capabilities. How-ever it is guaranteed that they become even more parallel due to developments such as multicore. Obtaining the optimal performance on these machines is not only a matter of hardware it is also an issue of programming design. Therefore co-development of hard- and software is needed. The question is: how to test algorithm’s on machines which do not exist today. To address the programming issues in terms of scalability and fault tolerance for the next generation, this project’s aim is to design and develop a simulator based on Parallel Discrete Event Simulation (PDES) for testing arbitrary Message Pass-ing Interface (MPI) based applications. Massive parallel environments with at least 107 virtual processes can be simulated. In comparison todays most denst supercomputer combine 105 cores togeter. The simulation it self is a transparent layer where the MPI applications run on top. These layer is designed to provide mechanism’s for collecting metric data as well as test the applications behavior against failure. To do so a fault injection based on distibution models and in a directly manner can be done