{"title":"The effect of simultaneity on sorting operations","authors":"S. Feerst, F. Sherwood","doi":"10.1145/612201.612252","DOIUrl":null,"url":null,"abstract":"The sorting method to be considered is a standard tape merge, modified to take advantage of whatever simultaneous operation capabillties are available. The two types of simultaneity to be considered here are: a) Read-Write and b) Read-Write-Compute. For analysis purposes the sorting method divides naturally into two parts: Phase One-the first tape passt wherein sorted strings of some length are produced from the original randomly arranged data~ by mesh sorting1 and Phase Two-all subsequent tape passes1 wherein the sorted strings produced by Phase One are merged until a single sorted string is obtained. Phase One A division of the internal memory into input and output sections indicates that with no simultaneity, or with read-write simultaneity~ sorted strings of length A/2 are produced1 where A equals the available work area. in the latter caser sharing the reading and writing produces some time advantage. if complete read-write-compute simultaneity exists the sorting method could produce sorted strings of length A/4 and minimize the time required for Phase One. These strings would be half the length of those produced using simultaneous read-write. The longer sorted strings can be handled with fewer passes in Phase Twot more than offsetting the extra time required to produce them in Phase One. Therefore1 even when simultaneous read-write-compute is available~, it is preferable to utilize only simultaneous read-write for Phase One.","PeriodicalId":109454,"journal":{"name":"ACM '59","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1959-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"ACM '59","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/612201.612252","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
The sorting method to be considered is a standard tape merge, modified to take advantage of whatever simultaneous operation capabillties are available. The two types of simultaneity to be considered here are: a) Read-Write and b) Read-Write-Compute. For analysis purposes the sorting method divides naturally into two parts: Phase One-the first tape passt wherein sorted strings of some length are produced from the original randomly arranged data~ by mesh sorting1 and Phase Two-all subsequent tape passes1 wherein the sorted strings produced by Phase One are merged until a single sorted string is obtained. Phase One A division of the internal memory into input and output sections indicates that with no simultaneity, or with read-write simultaneity~ sorted strings of length A/2 are produced1 where A equals the available work area. in the latter caser sharing the reading and writing produces some time advantage. if complete read-write-compute simultaneity exists the sorting method could produce sorted strings of length A/4 and minimize the time required for Phase One. These strings would be half the length of those produced using simultaneous read-write. The longer sorted strings can be handled with fewer passes in Phase Twot more than offsetting the extra time required to produce them in Phase One. Therefore1 even when simultaneous read-write-compute is available~, it is preferable to utilize only simultaneous read-write for Phase One.