Xref: utzoo comp.arch:11132 comp.sys.mips:118 Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!ncar!asuvax!mcdphx!udc!chant!aglew From: aglew@urbana.mcd.mot.com (Andy-Krazy-Glew) Newsgroups: comp.arch,comp.sys.mips Subject: Re: Memory utilization & inter-process contention Message-ID: Date: 24 Aug 89 19:37:47 GMT References: <3332@blake.acs.washington.edu> Sender: aglew@urbana.mcd.mot.com Organization: Work: Motorola MCD, Urbana Design Center; School: University of Illinois at Urbana-Champaign Lines: 20 In-reply-to: lgy@blake.acs.washington.edu's message of 22 Aug 89 05:53:58 GMT ..> L. Yaffe talks about thrashing (2 25M processes on a 32M system), asks: > Which (if any) high-performance Unix systems are capable of avoiding >this type of inter-process contention? What works best - actual process >swapping layered on top of demand paging, better paging algorithms, >clever scheduling based on adaptive process priorities, ... ? Gould NPL had a "working set" scheduler for memory; a working set was determined by looking at the number of pages touched by a process in a given interval of process virtual time (not system time); a process would not run if the working set could not be loaded. Pervase Akhtar wrote a paper on it for a past UNIX conference. Working set schedulers are quite common in large (non-UNIX) systems. See any recent issue of ACM SIGMETRICS' Conference Proceedings. -- Andy "Krazy" Glew, Motorola MCD, aglew@urbana.mcd.mot.com 1101 E. University, Urbana, IL 61801, USA. {uunet!,}uiucuxc!udc!aglew My opinions are my own; I indicate my company only so that the reader may account for any possible bias I may have towards our products.