Automatic generation of don´t cares for the controlling finite state machine from the corresponding behavioral description

A symbolic analysis method is described and applied to register-transfer microprograms describing control units of hardware controllers. It automatically generates program invariants and next uses them to optimize the logic realization of the control finite state machine (FSM). The method calculates the microprogram invariants, symbolic states, and relations. These data are used to find products of predicate values (predicates are the signals from the data path to the control unit) that have constrained values for some of the control unit states. If a state transition exists from internal state S_i to state S_j under a product of predicate values that cannot be satisfied in state S_i , then this transition is replaced in the state table with a don´t care. This is beneficial because a state table composed of a large number of don´t cares facilitates the processes that optimize the realization of the FSM logic circuit: the state minimization, the state assignment of input and internal states, and the Boolean minimization. A computer implementation of these methods that uses extensively symbol manipulation and artificial intelligence techniques is also described