Show simple item record

dc.contributor.authorEgan, C.
dc.contributor.authorSteven, G.B.
dc.contributor.authorVintan, L.
dc.date.accessioned2008-07-03T14:24:23Z
dc.date.available2008-07-03T14:24:23Z
dc.date.issued2002
dc.identifier.citationEgan , C , Steven , G B & Vintan , L 2002 , ' Cached Two-Level Adaptive Branch Predictors with Multiple Stages ' , Lecture Notes in Computer Science , vol. 2002 , pp. 179-191 . https://doi.org/10.1007/3-540-45997-9
dc.identifier.issn0302-9743
dc.identifier.otherPURE: 86809
dc.identifier.otherPURE UUID: 8cd323a9-037f-4ef1-871a-474c10c31286
dc.identifier.otherdspace: 2299/2171
dc.identifier.otherScopus: 0345340029
dc.identifier.urihttp://hdl.handle.net/2299/2171
dc.descriptionThe original publication is available at www.springerlink.com . Copyright Springer DOI : 10.1007/3-540-45997-9
dc.description.abstractDuring the last decade, the accuracy of branch predictors was significantly improved by the development of Two-Level Adaptive Branch Predictors. However, although these predictors deliver very high prediction rates, they have several disadvantages. Firstly, the size of the secondlevel Pattern History Table (PHT) increases exponentially as a function of history register length and therefore becomes very costly if a large amount of branch history is exploited. Secondly, many of the prediction counters in the PHT are never used. Thirdly, predictions are frequently generated from non-initialised counters. Finally, several branches may update the same counter, resulting in interference between branch predictions. In this paper, we quantify the performance of a novel family of multi-stage Two-Level Adaptive Predictors. In each two-level predictor, the PHT is replaced by a Prediction Cache. Unlike a PHT, a Prediction Cache saves only relevant branch prediction information. Furthermore, predictions are never based on uninitialised entries and interference between branches is eliminated. In the case of a Prediction Cache miss in the first stage, our two-stage predictors uses a default two-bit prediction counter stored in a second stage. We demonstrate that a two-stage Cached Predictor is more accurate than a conventional two-level predictor and quantify the crucial contribution made by the second prediction stage in achieving this high accuracy. We then extend our Cached Predictor by adding a third stage and demonstrate that a Three-Stage Cached Predictor further improves the accuracy of cached predictors.en
dc.language.isoeng
dc.relation.ispartofLecture Notes in Computer Science
dc.rights/dk/atira/pure/core/openaccesspermission/open
dc.titleCached Two-Level Adaptive Branch Predictors with Multiple Stagesen
dc.contributor.institutionSchool of Computer Science
dc.description.statusPeer reviewed
dc.relation.schoolSchool of Computer Science
rioxxterms.versionofrecordhttps://doi.org/10.1007/3-540-45997-9
rioxxterms.typeJournal Article/Review
herts.preservation.rarelyaccessedtrue
herts.rights.accesstyperestrictedAccess


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record