Parallelization of Loop Nests with General Bounds in the Polyhedron Model

The polytope model is one possible (mathematical) basis for par-allelizing sequential computer programs automatically. It proved to be well suited for the parallelization of loop nests containing only for loops whose bounds satisfy several restrictions. Recent research efforts propose an extension of the polytope model, the polyhedron model, and provide an implementation for perfectly nested while loops. The first part of this thesis examines the implications of different loop types for target code generation and integrates the results in a class hierarchy. The second part extends the polyhedron model to imperfect loop nests containing general for loops, while loops and if statements. We provide also one possible implementation