Encoding multiple outputs for improved column compaction

The authors introduce EP1, a new encoding problem where the objective is to find an encoding so that the binary logic function has a minimal number of outputs after column compaction. Whereas state assignment is restricted to functions which have only one multiple-valued output variable (the next state variable), EP1 is important for functions which have many multiple-valued output variables, for example, control stores and other components of finite state machines which often occupy large fractions of the total chip area. This is particularly useful in high-level synthesis, where datapath components can often be reallocated or resynthesized to accept arbitrary or near-arbitrary control codes. A heuristic solution to EP1 is given, and experimental results are provided.<>