*BSD News Article 52100


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!news.sprintlink.net!howland.reston.ans.net!swrinde!sdd.hp.com!news1.best.com!shellx.best.com!shell1.best.com!not-for-mail
From: rcarter@best.com (Russell Carter)
Newsgroups: comp.unix.bsd.freebsd.misc,comp.unix.bsd.misc,comp.os.386bsd.misc,comp.os.386bsd.questions,comp.unix.bsd
Subject: Re: Ultimate BSD PC Hardware Setup
Date: 2 Oct 1995 19:12:49 -0700
Organization: Best Internet Communications, Inc. (info@best.com)
Lines: 109
Message-ID: <44q671$eu9@shell1.best.com>
NNTP-Posting-Host: shell1.best.com
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:6766 comp.unix.bsd.misc:248 comp.os.386bsd.misc:5725 comp.os.386bsd.questions:17620 comp.unix.bsd:16715


Well, I've been a trifle busy shipping 35G of disk (with clusters
attached :-) in the last 10 days or so, so didn't have time
to wade around the old net.  But the fun things we find!

To make sure there is no confusion, I *am* Geli Engineering, and
the workstation clusters Geli Engineering sells are powered by FreeBSD.
Every claim I make is supported by data on http://www.geli.com, and
if you notice questionable conclusions there please contact me!

I have *no* comments about NetBSD as a project, other than it is an 
excellent thing to do.  I do have comments about Microsoft, but that 
is not what I came to talk about.

What I would like to talk about is the value of advice on the internet
concerning peripherals for FreeBSD when the advisor has got real
problems with both conflict of interest and lack of experience on 
the material discussed.

Here's the meat.  I have had correspondence with Michael Van Loon
since November of '94, when I brought up NetBSD 1.0 and FreeBSD 2.0
in a head to head comparison on identical heavy duty hardware.  At
that time, he recommended NetBSD as an equivalent if not superior
alternative to FreeBSD.  You might infer accurately the results
of my analysis by the OS I chose to base my business on.

Now we find that

1. Michael is employed by Microsoft (a rather large multiple OS vendor)
   (reluctantly admitted...)
2. Giving hardware and setup advice in FreeBSD netgroups (FreeBSD is 
   an OS project)
3. Actively steering people to NetBSD on the basis that it is a fully
   equivalent alternative. (NetBSD is an OS project)
4. Nearly completely unblemished by experience with the equipment
   he recommends on FreeBSD. (reluctantly admitted...)

Now, it doesn't take a whole lot of imagination to cook up all kinds
of interesting scenarios for the motivation of this piece of advice
or that one or *that* one 8-0.

This isn't personal, but a reflection of my genuine distaste
for the use of guerrilla marketing.  I think a good way to
rehabilation would be for Michael to just flat out say:  "You 
should use NetBSD!  It's really *great*, and here is 
why: [datadatadata....]" whenever he gives advice on FreeBSD
hardware.  After all that is what he is doing.  I think.

Michael's specific comments spur me to output a bit of
good data on the NCR:

|1. 
|
|   >   >Just because my only system at home is EISA-based doesn't mean I have
|   >   >no experience with other platforms.  In fact, I have set up two
|   >   >different PCI systems.  I have used the NCR card, and it worked just
|   >   >great.  I have also heard enough reports on the BusLogic card to know
|   >   >it works adequately.  And, finally, I have heard enough reports on
|   >   >Adaptec's policies, some of them recent, that I maintain my stand that
|   >   >I have no intention of ever supporting them with my money.
|
|   Wow!  Two different systems! That's a lot of experience, I'm curious though,
|   were those FreeBSD systems?  Be honest...
|
|What's your point?  You're splitting hairs, and I have no interest in
|rehashing old ground.  Have you beat this useless tangent to death,
|yet?
|
|No.  They were not FreeBSD systems.  Who cares?!  NetBSD uses the
|exact same NCR driver as FreeBSD.  The experience applies adequately.

Well, at least until it's run through cpp, but you knew that right?
If so, why is it quite a bit slower in NetBSD?
Well, we both know it is more complicated than just the driver...
I have pretty good data here on NetBSD, but ideally the NetBSD
folks would publish something analagous to what is on http://www.geli.com.

|Plus, in addition to two PCI NetBSD systems, I and others around me,
|use literally hundreds of NCR PCI systems running Windows NT and
|Windows 95, many with multiple drives (and multiple devices).  They
|all work very nicely -- I have no reservations about recommending NCR
|PCI SCSI controllers.

Even with a Quantum Grand Prix 4.3 G?  I wouldn't do that, with any
system up to about at least the (current) FreeBSD-current.  In
other words, what nearly everyone has got access to.  Steffen thinks
he's got the problem licked but I have not had a chance to verify
it.  It certainly wasn't licked three weeks ago.  I had to ship
a web server with a 2940 in it.  I *do* *not* unequivically
recommend it with untested drives.  The same goes for the 2940
for that matter.  Drive manufacturers can be a hedgy lot, and this
is not an OS specific problem.  I'll post the results when
I get it to work on the Grand Prix.

On the other hand these drives shipped out the door with NCRs working
flawlessly this week:

Fujitsu M1606 1GB
Quantum Atlas 2.1GB
Seagate Hawk 4.3GB
Quantum Fireball 540M.

In addition, I don't hesitate to recommend it with any drive that
shows up on http://www.geli.com.  

[ and a lot of misinterpreted stuff deleted...]

Cheers,
Russell