Local and global microcode compaction using reduction operators

Abstract

The problem of compacting microcode has received considerable attention, but there remains much room for improvement. The major obstacle is the NP-completeness of the associated optimization problem and the coupling between code generation and compaction. Reduction operators are one form of heuristic technique that have been used effectively in scene analysis. By abstracting the microcode compaction problem as a constraint satisfaction problem, we can utilize some developed heuristic techniques. This approach is demonstrated along with experimental results obtained from a computer implementation. A comparison is made with several existing methods.