Key points are not available for this paper at this time.
Measurement and evaluation of computer systems requires a model of the computer under control of the operating system that shows the interrelation between task flow and allocation of resources. For multiprogramming and multiprocessing systems, the model needs to show parallelism of operations. Furthermore, it should allow easy shift into greater or less detail as required by the measurement investigation, while providing perspective on the whole system. An adaptation of Petri Nets has been found well suited to this purpose. The representation shows parallelism, and allows compression of either sequential or parallel operations when less detail is required. The paper describes a multiprocessor, multiprogramming system, the CDC 6400, in terms of Petri Nets and shows how this type of representation lends itself to planning system measurements.
Jerre D. Noe (Fri,) studied this question.