Return to BSD News archive
Path: sserve!newshost.anu.edu.au!munnari.oz.au!foxhound.dsto.gov.au!fang.dsto.gov.au!yoyo.aarnet.edu.au!news.adelaide.edu.au!news.cs.su.oz.au!harbinger.cc.monash.edu.au!msuinfo!agate!howland.reston.ans.net!gatech!newsxfer.itd.umich.edu!gumby!yale!yale.edu!noc.near.net!news.delphi.com!usenet From: John Dyson <dysonj@delphi.com> Newsgroups: comp.os.386bsd.questions Subject: Re: linux's I/O calls faster than NetBSD's ? Date: Wed, 9 Mar 94 19:13:40 -0500 Organization: Delphi (info@delphi.com email, 800-695-4005 voice) Lines: 18 Message-ID: <BE3J-z0.dysonj@delphi.com> References: <2lhv9r$pbt@homea.ensta.fr> <2limt7$16k@homea.ensta.fr> NNTP-Posting-Host: bos1b.delphi.com X-To: Manuel Bouyer <bouyer@bsdtest.ensta.fr> I think that I have found out the reason why *BSD seems slow... I have heard that Linux has a unified VM/Buffer cache and can use more memory for caching because of the dynamic nature of the cache size. FreeBSD doesn't have that yet :-) (soon!!!) I am not ready to discuss the details, but FreeBSD will have a very sleek scheme (probably in -current in about 2-3 weeks.)) We have been planning the capability for about 1-2 months, but wanted to remove some performance bottlenecks and debug some new VM capabilities. We are about ready for the switch (just need to bring up a second machine -- debugging low level fs stuff is not safe on a production system.) So your observations are showing the limited size of the *BSD buffer caches. I do have an experimental dynamically sized vfs_bio, but we are not going to use the primative technique that I used in the code I wrote a year ago. Take care -- and give *BSD a chance, it really is goood... :-) John dyson@implode.root.com