*BSD News Article 31005


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!agate!ihnp4.ucsd.edu!swrinde!pipex!sunic!EU.net!ieunet!news.ieunet.ie!jkh
From: jkh@whisker.hubbard.ie (Jordan K. Hubbard)
Newsgroups: comp.os.386bsd.questions
Subject: Re: Linux or FreeBSD?
Date: 29 May 1994 09:39:59 GMT
Organization: Jordan Hubbard
Lines: 116
Message-ID: <JKH.94May29093959@whisker.hubbard.ie>
References: <CqH2z7.29E@dit.upm.es> <2s618a$34t@pdq.coe.montana.edu>
	<2s86fj$cn4@acmex.gatech.edu> <hastyCqJBED.9yz@netcom.com>
	<2s8mbn$e1o@acmex.gatech.edu> <JKH.94May29030102@whisker.hubbard.ie>
	<2s937t$7sp@acmex.gatech.edu>
NNTP-Posting-Host: whisker.hubbard.ie
In-reply-to: gt8134b@prism.gatech.edu's message of 28 May 1994 23:47:09 -0400

In article <2s937t$7sp@acmex.gatech.edu> gt8134b@prism.gatech.edu (Robert Sanders) writes:

   But this isn't one of those issues!  Sure, that's the subject, but the
   whole point of my two (now three) posts was to simply explain why
   some people who had used both Linux and FreeBSD came away with the
   subjective feeling that Linux was much faster.  I don't give a damn
   about feature comparisons from other people: my roommate and I went
   out and *tried* it.  In the same vein, I don't try to offer authoritative
   feature comparisons; I merely summarize my expereinces.

Well, I guess I should point out that I came in late to this
discussion and haven't seen your earlier posts, so I may have
misinterpreted your basic viewpoint from lack of sufficient context.
In fact, I do have to say that your basic approach is one that I
greatly approve of - I wish more people in this generally useless
FreeBSD vs Linux debate would actually _take the trouble_ to load and
run both for awhile before spouting off, and your willingness to do so
is only to your credit.  That said, I think that it's also pretty
important in these `debates' (only in quotes because so few of them up
to now have qualified for such an exalted label ;-) to differentiate
between the *subjective* points and the *objective* points, since it's
sometimes hard to tell the genuine technical gripes from the personal
preference issues.  Your later comments regarding packaging indicate
that you've probably realized that in retrospect, so I'll say no more
about it.

   You're fighting straw men, Jordan.  I never said that.  I'm not trying
   to win credibility for anyone, but if I were, it would be for FreeBSD.
   Two separate messages on this group expressed dismay at the relative
   speed of FreeBSD; one even said that out of Linux, Windows, etc., FreeBSD
   was the slowest OS he'd had on the machine.  I was trying to explain
   that it isn't that slow, that it's just tuned less for interactive response
   than Linux was. 

Ok, fair enough.  Again, blame me for coming in late.  I might also
note that FreeBSD's detractors should also bear in mind that we, too
are still evolving and still have more than a few optimizations to go
- we're not at all insensitive to the criticisms of those Linux users
who make unfavorable (and credible) comparisons between FreeBSD's and
Linux's interactive response.  When I met Linus in Holland last year,
one of the major points of discussion was, in fact, Linux's VM and
scheduling algorithms and how FreeBSD would do well to take a closer
look into how interactive processes were scheduled in environments
with small process counts.  It should always be noted, of course, that
there's no free lunch with these things.  Linux's scheduling and VM
code does offer superior performance with light job mixes, but the
`ramp' of performance vs number of system intensive processes actually
declines much faster on a Linux system as well, so you have to pay the
price somewhere.  At least this was my experience with Linux 0.99, and
also backed up by a number of others (some of whom were die-hard Linux
fans).

This is, again, just being honest about where the plusses and minuses
lie and my personal preference would, of course, be a pan Linux and
FreeBSD team actively engaged in looking at performance issues (in
both light and heavily loaded environements) across BOTH operating
systems, trying to merge in some of the best features of each in order
to strengthen the quality of both.  I would actively support such a
movement.

We should always bear in mind (and I don't necessarily mean you
personally, Robert, since I sense I'd be preaching to the choir) that
UNIX itself is what's living a somewhat besieged life here, not
FreeBSD or Linux specifically.  If we want to extend the lives of both
in the face of competition from a number of `young turk' operating
systems now coming down the pike, we may someday have to seriously
consider working together just a little more (and I almost cry at the
amount of effort being gratuitously spent on parallel efforts or
competing implementations where there's _really no need_ for a lot of
it).

This is not to say that I necessarily see this as an easy thing, as
the frequently touted FreeBSD/NetBSD mergers have pointed out in very
painful detail, simply that it's a damn shame that it has to be the
case.  I personally think that Linux and FreeBSD are also not so far
apart, ideologically speaking, and if FreeBSD and NetBSD cannot enjoy
a closer working relationship (though in truth it's far more cordial
as of late) then perhaps FreeBSD and Linux can.  I'd be more than keen
to extend the requisite olive branches!

   I don't know if the Kranenberg/FreeBSD shlib implementation uses fixed
   addresses; I doubt it does.  Has anyone benchmarked statically linked
   vs. dynamically linked executables on speed of *library* code?  I'll
   bet it's not insignificant.  Again, I know Linux's implementation isn't
   perfect, but there are good reasons to have it.  Linux developers aren't
   idiots.

I never said they were, simply that I wasn't a big fan of some of the
trade-offs they chose.  It's more than just a one-off hit to the
developers, as well.  Did you know, for example, that OI lets you
create your own subclasses interactively, flesh them out, then bring
them up in the builder?  Probably not since Linux's OI version has
never allowed you to do this (and boy, was I annoyed when I found this
out since I was trying to use the OI port to Linux to do some work!)
since it was too much trouble to generate a shared library
automatically without any intervention by the `customer'.  I am happy
to see that Linux is heading down the ELF route, and I expect that
this will render many points like this moot, I simply wanted to point
out that it's not always a `one time cost' borne only by the developer
and then completely transparent to all users thereafter.

   To sum it up: I wasn't contributing to the Linux-vs-FreeBSD thread
   itself, but merely trying to comment on the apparent speed difference.

Ok..  Like I said, we're both still evolving OS's with great potential
for gains in performance and functionality, and if we can learn from
eachother's implementations, all the better.  That is why an honest
appraisal of Linux's strengths is actually of *benefit* to me, and why
I generally have so little patience with these (in general, not yours)
Linux/FreeBSD flame wars that generate much heat but shed so little
useful light.

					Jordan

--
Jordan K. Hubbard	FreeBSD core team	Friend to mollusks