Limitations of synchronization primitives with conditional branching and global variables

Abstract

A formal model of the process concept is presented. This model can represent sets of processes that use the synchronization primitive PV or one of the many generalizations of PV.The study of synchronization problems is; then reduced to the study of relations between sets of processes.For one relation --"simulate" --it is possible to show that there are differences between several synchronization primitives.These differences show that the relative "power" of these synchronization primitives is not the same.

A comparison of two Petri net types

“…To demonstrate that simulate2 is not "too strong" we present in Fig. 2 a hierarchy proved in [3] between PNcom, PNIog, PNout and PN with respect to Liptons' simulation rules [5] which are stronger than simulate2 and which nevertheless seem to be very reasonable and useful particularly with regard to realtime programming.…”

Synchronization Problems Solvable by Generalized PV Systems

“…Lipton [18,19] and Lipton, Snyder, and Zalcstein [20] have studied the relative powers, in a precise sense, of various systems of synchronization primitives. The results of [18-2% however, are mainly of the form: "A system of type Tl can solve a particular synchronization problem S and no system of type T2 can solve S. Thus TI is more powerful than T2.…”