A branch-and-bound method for optimal transformation of data flow graphs for observing hardware constraints

The paper introduces a new branch-and-bound-method which produces a data path from a given procedural behavior description observing precise hardware constraints and minimizing the execution time of the hardware algorithm to be implemented. The method starts with a maximal parallel flow graph and establishes supplementary edges in the graph to such an extent that the hardware constraints are observed for all deducible implementations of the resulting flow graphs.<
>