*BSD News Article 33236


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!howland.reston.ans.net!europa.eng.gtefsd.com!MathWorks.Com!news2.near.net!news.delphi.com!usenet
From: John Dyson <dysonj@delphi.com>
Newsgroups: comp.unix.bsd
Subject: Re: 4.4-lite?
Date: Thu, 21 Jul 94 01:19:10 -0500
Organization: Delphi (info@delphi.com email, 800-695-4005 voice)
Lines: 19
Message-ID: <xQ+RCHu.dysonj@delphi.com>
References: <2vgvc7$3tg@spruce.cic.net> <30finf$98e@pdq.coe.montana.edu> <Ct75oE.75p@newsserver.aggregate.com> <30h9jl$fg4@pdq.coe.montana.edu> <Ct8o9E.8My@newsserver.aggregate.com>
NNTP-Posting-Host: bos1d.delphi.com
X-To: Rob Healey <rhealey@sirius.aggregate.com>

Rob Healey <rhealey@sirius.aggregate.com> writes:
 
>	OK. EXACTLY what are these VM enhancements. They've been touted
>	up and down but no specifics have ever been given. Let's hear
>	'em! How exactly is FreeBSD's VM so much more lightning fast
>	than NetBSD and how does that apply to non-x86 architectures,
>	ESPECIALLY big endian?
 
Much of the info is in the release notes for FreeBSD.  Additionally,
I can probably dig up some of the postings that I made to comp.os.386bsd.dev...
and repost them.  The code will work fine on big endian architectures, but
some of the enhancements are non-starters on machines with TLB-only like
sparcs.
 
The FreeBSD VM code is NOT lightning fast, but there are some algorithmic
improvements that DO usefully increase %cpu for running user processes
on memory starved systems.  Also, process start-up time is significantly
improved.  You can actually measure the speed improvements, and they
are significant.