In this paper, I examine the possibilities of parallelization and performance enhancement in software model based graph transformations. In model driven development using graph transformations is one of the most popular ways to process models. A graph transformation consists of multiple graph rewriting steps, in each step a pattern (called the match) is searched and replaced by the given rule. As the size of the model grows, the processing time increases exponentially. Therefore, optimizing graph transformations and speeding up pattern searches are of key importance in model processing. One possible direction is to apply the transformation – particularly the pattern matching phase – in parallel. Considering the personal computer architecture of nowadays, the idea of multicore execution comes natural.
The goal of this thesis is to elaborate the results of my research in parallel graph transformation in multicore systems. The thesis starts with a brief introduction about the mathematical and software development background of graph transformation. As next, the details of my approach are elaborated. The thesis is not limited to the theoretical background of the proposed solution, but it also introduces the practical realization of the system. The implementation of the approach is based on the VMTS framework developed at AAIT. Along the introduction of the development, a complete case-study with various measures is elaborated showing the efficiency of parallelization methods in practice.