An evolutionary technique for local microcode compaction

Abstract

In this paper we present a variant of the simulated evolution technique for local microcode compaction. Simulated evolution is a general optimization method based on an analogy with the natural selection process in biological evolution. The proposed technique combines simulated evolution with list scheduling, in which simulated evolution is used to determine suitable priorities which lead to a good solution by applying list scheduling as a decoding heuristic. The proposed technique is an effective method that yields good results without problem-specific parameter tuning on test problems of very different sizes and structures. This is achieved by establishing a reasonable balance between exploration of the search space and exploitation of good solutions found in an acceptable CPU time. We demonstrate the effectiveness of our technique by comparing it with the existing microcode compaction techniques for randomly generated data dependency graphs. The proposed scheme offers considerable improvement in the number of microinstructions compared with the existing techniques with comparable CPU time.