Systems design space exploration by serial dataflow program execution

The increasing ubiquity of heterogeneous parallel computing platforms nowadays creates the challenge to fully exploit the available computational power when porting existing programs or developing new applications with portability in mind. Existing design space exploration methods focus on specialized applications amenable to compile-time analysis. Real-world applications, however, tend to exhibit complex behavior that depends on input data and even timing. This paper proposes a methodology for creating a finite (approximate) representation of the design space of general streaming applications, based on detailed tracking of a serial run of the program. Homotopy theoretic methods are used to demonstrate how the design space of a program can be reconstructed from its serial execution trajectory. Moreover, the concept of a dependency graph of a dataflow program defined in the literature is extended with the definition of two new kinds of dependencies - the Guard Enable and Disable - and the 3-tuple notion needed to represent them.