Fully utilizing the potential of parallel architectures is known to be a challenging task. In the past the software developer had to deal with this challenge using a variety of specialized programming models. The Insieme compiler, currently under development by the Distributed and Parallels Systems group at the University of Innsbruck, tries to support developers in this challenging task. The compiler automatically optimizes parallel applications for the execution on heterogenous and homogenous systems. In this master thesis the development of a testing framework used in the Insieme compiler project is shown. Input codes for integration testing and permanent monitoring of the compiler performance were collected. An integration test framework to automatize the execution of test codes was built. Each test code was classified by several metrics (e.g. scalability, memory consumption), those metrics are useful to optimize the compilation process for a high number of miscellaneous test codes.
The second part of the thesis shows a performance analysis of a product application. The analysis focuses on the scalability on parallel shared memory systems. Proposals to speed up the application are made.