返回
An Overview of Cache Optimization Techniques and Cache-Aware Numerical Algorithms.pdf

An Overview of Cache Optimization Techniques and Cache-Aware Numerical Algorithms.pdf

ID：34141126

大小：316.67 KB

页数：23页

时间：2019-03-04

An Overview of Cache Optimization Techniques and Cache-Aware Numerical Algorithms.pdf_第1页
An Overview of Cache Optimization Techniques and Cache-Aware Numerical Algorithms.pdf_第2页
An Overview of Cache Optimization Techniques and Cache-Aware Numerical Algorithms.pdf_第3页
An Overview of Cache Optimization Techniques and Cache-Aware Numerical Algorithms.pdf_第4页
An Overview of Cache Optimization Techniques and Cache-Aware Numerical Algorithms.pdf_第5页
资源描述：

《An Overview of Cache Optimization Techniques and Cache-Aware Numerical Algorithms.pdf》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、AnverviewofCacheptimizationTechniques?andCache{AwareumericalAlgorithms12arkusowarschikandChristianWei1ehrstuhlfurSystemsimulation(nformatik10)nstitutfurnformatikFriedrich{Alexander{UniversitatErlangen{urnberg,Germanyarkus.owarschik@cs.fau.de2ehrstuhlfurRechnertechnikundRechnerorganis

2、ation(RR{TU)FakultatfurnformatikTechnischeUniversitatunchen,GermanyChristian.Weiss@cs.tum.edu1ntroductionnordertomitigatetheimpactofthegrowinggapbetweenCUspeedandmainmemoryperformance,today'scomputerarchitecturesimplementhierar-chicalmemorystructures.Theideabehindthisapproachistohideboththe

3、lowmainmemorybandwidthandthelatencyofmainmemoryaccesseswhichisslowincontrasttothe oating{pointperformanceoftheCUs.Usually,thereisasmallandexpensivehighspeedmemorysittingontopofthehierarchywhichisusuallyintegratedwithintheprocessorchiptoprovidedatawithlowlatencyandhighbandwidth;i.e.,theCUregister

4、s.ovingfurtherawayfromtheCU,thelayersofmemorysuccessivelybecomelargerandslower.Themem-orycomponentswhicharelocatedbetweentheprocessorcoreandmainmemoryarecalledcachememoriesorcaches.Theyareintendedtocontaincopiesofmainmemoryblockstospeedupaccessestofrequentlyneededdata[31,33].Thenextlowerleveloft

5、hememoryhierarchyisthemainmemorywhichislargebutalsocomparativelyslow.Whileexternalmemorysuchasharddiskdrivesorremotememorycomponentsinadistributedcomputingenvironmentrepresentthelowerendofanycommonhierarchicalmemorydesign,thispaperfocusesonoptimizationtechniquesforenhancingcacheperformance.Theleve

6、lsofthememoryhierarchyusuallysubsetoneanothersothatdataresidingwithinasmallermemoryarealsostoredwithinthelargermemories.AtypicalmemoryhierarchyisshowninFig.1.EÆcientprogramexecutioncanonlybeexpectedifthecodesrespecttheunderlyinghierarchicalmemorydesign.Unfortunately,today'scompilerscannotintroduce

7、highlysophisticatedcache{basedtransformationsand,consequently,muchofthisoptimizatione ortislefttotheprogrammer[26,41].?ThisresearchisbeingsupportedinpartbytheDeutscheForschungsgemeinschaft(GermanScienceFoundation

当前文档最多预览五页，下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页，下载文档查看全文
温馨提示：
1. 部分包含数学公式或PPT动画的文件，查看预览时可能会显示错乱或异常，文件下载后无此问题，请放心下载。
2. 本文档由用户上传，版权归属用户，天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容，确认文档内容符合您的需求后进行下载，若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误，付费完成后未能成功下载的用户请联系客服处理。