Return to BSD News archive
Newsgroups: comp.os.386bsd.misc Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!hookup!swrinde!ihnp4.ucsd.edu!library.ucla.edu!csulb.edu!csus.edu!netcom.com!gokings From: gokings@netcom.com (Russell Marrash) Subject: Re: Shared Library Status ? Message-ID: <gokingsCMp4Lt.HsA@netcom.com> Organization: Netcom Online Communications Services (408-241-9760 login: guest) References: <2lqck0$11k@pdq.coe.montana.edu> <michaelv.763634402@ponderous.cc.iastate.edu> <CMnsJw.8p8@cogsci.ed.ac.uk> Date: Tue, 15 Mar 1994 07:53:53 GMT Lines: 29 In article <CMnsJw.8p8@cogsci.ed.ac.uk> richard@cogsci.ed.ac.uk (Richard Tobin) writes: >In article <michaelv.763634402@ponderous.cc.iastate.edu> michaelv@iastate.edu (Michael L. VanLoon) writes: >>90% sez who? :-) 90% under DOS, maybe. I don't think a multi-tasking >>OS that serves several-hundred interrupts per second and allocates >>processor time between several running processes with resident memory >>sets in the meg range is going to get an average 90% hit rate with an >>8k cache. I'd be pleased to be proven wrong on this point, though. Stuff deleted! > >Of course, that's just my guess. If you want figures, I suggest >posting to comp.arch. I believe that some systems (eg Mips) make it >reasonably easy to measure this sort of thing. > >-- Richard Hit rates are also dependent upon the type of tasks that are being run. If you are fetching a lot of sequential data then your hit rate will increase. A set-associative cache will give you a better hit rate then direct-mapped, but is more expensive to implement because you have to add the logic for the replacement sceam. The Intel 486 uses set-associative if I remember correctly. Also I know from experence that a 16k set associative cache will give you an average hit rate of 93%. All though I don't have the numbers for 8k, I don't think 90% is unreasonable. Russell Marrash gokings@netcom.com