A new approach for operation scheduling with data flow graph reforming

In behavioral synthesis, the description of the system is transformed into a data flow graph (DFG) first, and then scheduling and allocation are carried out. In this study, under the resource constraint, the fixed DFG with minimal operation count does not necessarily achieve a shortest schedule; its schedule may be longer than that of some algebraically equivalent DFGs. To explore a shortest schedule among those algebraically equivalent DFGs, a new approach is developed which modifies the topology of the graph and does scheduling at the same time. An example shows that the schedule obtained by this method is better than that of the original DFG.<>