Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!apple!fernwood!portal!cup.portal.com!mmm
From: mmm@cup.portal.com (Mark Robert Thorson)
Newsgroups: comp.arch
Subject: Re: memory management implementations
Message-ID: <41003@cup.portal.com>
Date: 7 Apr 91 01:59:57 GMT
References: <1726@spim.mips.COM> <1991Apr02.194431.11154@cello.hpl.hp.com>
Organization: The Portal System (TM)
Lines: 7

Someone once told me (hence I could be all wrong when repeating it) that
the clock or "scheduled sweep" algorithms are preferred to LRU
because although they don't have quite the best average performance,
they also lack LRU's worst-case performance.  For example, let's say
somebody is repeatedly processing the same array;  under LRU, every new
array element you move to is the element you visited longest-ago in time,
hence most likely to have been paged to disk.