{"title":"A time invariant working set model for independent reference","authors":"Vidyadhar Phalke","doi":"10.1145/1122018.1122046","DOIUrl":null,"url":null,"abstract":"Under the Independent Reference Model (IRM) of program behavior, we study properties of the C space-time products as defined by Prieve and Fabry [21] for variable memory management algorithms. In particular, we look at the behavior of Denning's Working Set [11] and VMIN [21] algorithms. From the observed properties of these algorithms, we derive an algorithm which either keeps a page always in memory, or does not keep it at all. We show the online optimality of this algorithm under the assumption that the access probabilities are known a priori. Further, we propose a practical version of this algorithm for database and disk buffer management. By using our algorithm, we obtained space-time product improvements up to 41% over the Working Set algorithm.","PeriodicalId":349974,"journal":{"name":"ACM-SE 33","volume":"27 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1995-03-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"ACM-SE 33","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/1122018.1122046","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
Under the Independent Reference Model (IRM) of program behavior, we study properties of the C space-time products as defined by Prieve and Fabry [21] for variable memory management algorithms. In particular, we look at the behavior of Denning's Working Set [11] and VMIN [21] algorithms. From the observed properties of these algorithms, we derive an algorithm which either keeps a page always in memory, or does not keep it at all. We show the online optimality of this algorithm under the assumption that the access probabilities are known a priori. Further, we propose a practical version of this algorithm for database and disk buffer management. By using our algorithm, we obtained space-time product improvements up to 41% over the Working Set algorithm.