With the advent of mobile computing and communications, the power consumption of microprocessors becomes an increasingly important issue. In particular, the investigation of power consumption in individual sub-blocks of processors has shown that power consumed by memory components accounts the greatest percentage of the total power consumption in microprocessors. In the controller design of microprogrammed processors, minimizing the control memory size is crucial in many power-critical applications since the size of the memory is directly related to the power consumption. It is also true for the instruction memory in application-specific VLIW (Very Long Instruction Word) processors.
In this thesis, we present a heuristic algorithm called CLASSIC for the minimization of the control memory width in microprogrammed processors or the instruction memory width of application-specific VLIW processors. CLASSIC results in nearly optimal solutions with the time complexity of $O(n^2)$, where n denotes the number of microoperations. In this thesis, we also propose the so-called incompleteness relations which are exploited for the minimization of the control memory width. Compared to the compatibility class method which applies to fields of mutually exclusive microoperations, our method applies to fields of mutually incomplete microoperations. Since incomplete relation includes exclusive relation, candidate fields in our method are the superset of those in the compatibility class method. This is the reason why our method always achieves better compression ratio than the compatibility class method.
The key features of our algorithm are that 1) it can exploit more combinations of microoperations than compatibility class scheme, and that 2) it can exploit a larger set of incompleteness combinations than the pure AND/OR set scheme. Moreover, CLASSIC can obtain nearly optimal solutions in a reasonable computation time even for big examples, which were not successfully handled ...