The role of instruction sequencing in structured microprogramming

Abstract

The current literature is filled with descriptions of various microprogrammed processors and discussions of the improvements in performance that can be realized through microprogramming. Thus, Tucker and Flynn [1] describe a dynamically microprogrammed processor and give several examples of problem-oriented programming in which the performance of the microcode was much better than that of assembly language code (System/360, normalized technology). Recently, Abd-alla and Karlgaard [2] have developed an algorithm for the synthesis of applications-oriented microcode for a dynamically microprogrammed computer. Their paper gives examples of problem-oriented architectures (realized through specialized instruction sets) which have much better performance than the corresponding general purpose architectures. This trend toward realizing specialized computer systems by means of writable control store probably means that more people will be writing microcode in the future. In particular, it seems worthwhile to consider the relation of the instruction sequencing functions of a given machine to both the ease of writing correct microcode and the size of control memory required for that machine. So far there seems to have been little or no discussion of this topic in the literature [3].