*BSD News Article 88672


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!news.cs.su.oz.au!metro!metro!munnari.OZ.AU!news.ecn.uoknor.edu!news.wildstar.net!news.ececs.uc.edu!cloudbreak.rs.itd.umich.edu!newsxfer3.itd.umich.edu!howland.erols.net!news.bbnplanet.com!cam-news-hub1.bbnplanet.com!uunet!in1.uu.net!192.207.105.50!prodigy.com!darkstar.prodigy.com!davidsen
From: davidsen@tmr.com (bill davidsen)
Newsgroups: comp.os.linux.networking,comp.unix.bsd.bsdi.misc,comp.unix.bsd.misc
Subject: Re: Linux vs BSD
Date: 5 Feb 1997 20:35:08 GMT
Organization: TMR Associates, Schenectady NY
Lines: 32
Message-ID: <5daqts$1f74@usenet1y.prodigy.net>
References: <32DFFEAB.7704@usa.net> <5cabgm$hp0@fg70.rz.uni-karlsruhe.de> <5chl91$bsi@usenet.srv.cis.pitt.edu> <87d8uqu5vt.fsf@localhost.xs4all.nl>
NNTP-Posting-Host: darkstar.prodigy.com
Originator: davidsen@darkstar.prodigy.com
Xref: euryale.cc.adfa.oz.au comp.os.linux.networking:67697 comp.unix.bsd.bsdi.misc:5938 comp.unix.bsd.misc:2390

In article <87d8uqu5vt.fsf@localhost.xs4all.nl>,
Peter Mutsaers  <plm@xs4all.nl> wrote:

| Added extra overhead of a user space implementation (this will always
| be less efficient than a kernel space implementation) your claim is
| clearly made up. Also it contradicts my own experiments with a recent
| Linux version at a completely clean ethernet (here at home).

I question your definition of "less efficient." It's true that you
will use a little less CPU writing in the kernel, but that's true of
everything, and no one says "read the hardware clock directly from a
user program for efficiency."

In user or kernel space you should be able to drive the hardware to
it's limits, and that's what counts. For portability I would use a
stub in the kernel, but you could map the buffers and registers into
user space, I guess.

This is _not_ a plea to move the network code, just a technical
comment on the posting.

I have seen ftp report very high transfer speeds myself, I knew
enough to disbelieve them. I wouldn't accuse someone of making up
numbers because they don't know how to benchmark. To send a file of
N bytes through a kernel with buffer space B the time is essentially
the transfer time of (N-B) bytes. If ftp reports the rate as N/sec
it is wrong, but the person reporting the number isn't necessarily
lying.
--
	-bill davidsen (davidsen@tmr.com)
"As a software development model, Anarchy does not scale well."
		-Dave Welch