Platform-independent modeling of explicitly parallel programs

Inhalt:
We propose a model-driven approach to parallel programming of SPMD-style, explicitly parallel computations. We define an executable, platform-independent modeling language with explicitly parallel control and data flow for an abstract parallel computer with shared address space, and implement it as an extension of UML2 activity diagrams and a generator for Fork source code that can be compiled and executed on a high-level abstract parallel machine simulator. We also sketch how to refine the modeling language to address more realistic parallel platforms.