4.6.3 Parallelization Method

Since a cluster of workstations is usually the standard simulation environment to perform process simulation and optimization, it makes sense to parallelize the Monte-Carlo ion implantation simulation process which is one of the most time consuming process simulation steps, to avoid a potential bottle neck in a process simulation flow.

Therefore a Parallelization method based on MPI [76] (Message Passing Interface), as a process communication library, which allows a distributed simulation on a cluster of single and multi processor workstations is implemented. The merit of MPI is that it provides a comfortable and fast interface to define process communication and that it is available for a wide variety of hardware platforms. As the parallelization strategy is focused on a cluster of workstations connected by slow networks, the Parallelization method was designed in a way that the communication overhead is kept so low that it has almost no influence on the performance gain. Additionally all sophisticated physical models and methods developed for the single processor version are used without any restriction.