authors proceed with the discussion of optimization techniques: (I) (2) (3) removal of redundant actions, removal of negated actions, code motion, such as taking an action (node of a graph) that need be executed once, out of a loop, (4) (5) (6) recognition of parallel actions, recognition of fan-out results of actions, filling of timing gaps with possible parallel actions. The authors discuss each of these items at some length, and t h e i r points are valid. The items I - 3 , which hold for general programs and microprograms, are well explained in terms of the graph representation. However, items 4-6 which are c h a r a c t e r i s t i c of microprograms, are concerned with actions inside nodes; the graph representation is not useful for t h e i r discussion. This would seem to indicate something missing in the means of representation. i t is not clear how timing dependencies are represented. The authors' In p a r t i c u l a r , specification of a microaction contains no indication of the times in the cycle of operation at which operands must be present,
/lp/association-for-computing-machinery/review-of-functional-characteristics-of-a-multilingual-processor-by-h-BqJHx0oBq7