Author: Bernhard Saumweber Title: Supporting the PGAS Model on Modern Shared Memory Systems Abstract: As shared memory nodes and processor architecture get more complex in the era of exascale computing, traditional programming models such as Pthreads and OpenMP are becoming less well suited to exploit their full potential while programming models such as Partitioned Global Address Space and Single Program Multiple Data become more attractive. DASH is a C++ template library that implements these paradigms and provides data structures, algorithms and additional abstractions such as patterns and hierarchical teams for high performance computing. To support OpenMP-style programming, DASH was extended in this thesis with additional data structures in a library called DASH-OMP, and by a Parallel Abstraction Layer (PAL) that allows to write more abstract applications that use either OpenMP or DASH for parallelization. DASH-OMP and PAL are evaluated using microbenchmarks as well as various larger benchmark applications.