Roll in/out and Usage of Large Capacity Core Memory in a Time-sharing System

Abstract

In this paper the analysis of roll in/out operation and the usage of large capacity core memory (LCM) in a time-sharing system (TSS) are described. A TSS model based on a few assumptions is proposed, and the relation between roll in/out and system performance is analyzed. The system is classified into two types, i.e. swapping limited and CPU limited, according to whether the maximum system performance is restricted by capability of swapping devices or central processing units, and using LCM in TSS is very effective for the swapping limited case because of decrease of system overhead due to roll in/out operation and increase of system performance. The usage of LCM is divided into two main classes, i.e. as roll-out area and roll-in area, and a system model with LCM is analyzed and discussed in each class. One of the typical cases of using LCM as roll-in area is the implementation of conversational language in interpretive mode, then, the condition on which the interpreter program using LCM is operated in swapping limited state in spite of decrease of swapping overhead and increase of CPU time is obtained, and a numerical example is showed. Finally, an optimum algorithm for the use of LCM, which betters system performance in man-computer communication, is suggested.