*BSD News Article 35851


Return to BSD News archive

Newsgroups: comp.os.386bsd.questions
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!netnews.upenn.edu!news.cc.swarthmore.edu!psuvax1!news.pop.psu.edu!news.cac.psu.edu!howland.reston.ans.net!europa.eng.gtefsd.com!MathWorks.Com!zombie.ncsc.mil!paladin.american.edu!constellation!rex!ben
From: ben@rex.uokhsc.edu (Benjamin Z. Goldsteen)
Subject: Re: SCSI controller recomendations requested
Message-ID: <Cw6rsI.C4r@rex.uokhsc.edu>
Date: Thu, 15 Sep 1994 19:36:17 GMT
Reply-To: benjamin-goldsteen@uokhsc.edu
References: <Cw3nuw.E8@galaxia.network23.com> <356jdu$irb@agate.berkeley.edu> <MICHAELV.94Sep14135401@MindBender.HeadCandy.com>
Organization: Health Sciences Center, University of Oklahoma
Lines: 62

michaelv@MindBender.HeadCandy.com (Michael L. VanLoon) writes:

>In article <356jdu$irb@agate.berkeley.edu> jkh@violet.berkeley.edu (Jordan K. Hubbard) writes:

>   The Bt445S seems to come in a variety of confusing revisions, some
>   broken some not, thus the "luck" factor.  I have one that works beautifully,
>   and other folks have had the worst luck.  Buslogic basically really screwed
>   the pooch on their VLB card and the road is litered with their previous
>   failure attempts.  Avoid picking one of these failed attempts up and you're
>   fine, otherwise..  It's always hard to know what a dealer has in stock and
>   how old it is.

>Not to disagree with your findings, but I've also been told this has
>as much to do with the motherboard as with the SCSI card.  I've had
>people tell me they're using bt445s cards that aren't supposed to work
>right, but do, on their motherboards with no problems.  Other people
>have had problems even with the "corrected" SCSI boards, when used in
>flaky motherboards.

>VLB is a very loosely defined "standard".  It leaves a lot of
>implementation decisions up to the person designing the motherboard.
>Hence, you will not get consistent results across all VLB
>motherboards.

>   The other two give commendable stability and performance.

>I'd have to disagree here.  I've watched with my own eyes a
>demonstration of just how slow the 1542 boards are.  This was with
>Windoze NT, but it still applies.  We have several Dell Pentium boxes
>at a place where I work (one of my many jobs), that run Windoze NT.
>My machine is a 66MHz Pentium, and has a NCR 53c810 SCSI controller in
>it.  We recently acquired a Dell XPS 90 (90MHz Pentium) machine, that
>came supplied with an Adaptec 1542.  Both machines have 32MB of RAM,
>512k cache, and newish 1GB SCSI drives.

>We decided to see how fast the two machines booted, mainly to see just
>how fast the 90MHz Pentium would be.  We got some surprising results.
>We set them beside each other, got to the OS boot menu, and hit return
>for Windoze NT at the same time on both machines.  The XPS 90 screamed
>thru the self tests, and went right into the disk check screen while
>the 66 was still sitting checking memory.  The 66 got to the disk
>checking screen while the 90 was about half way thru.  Then all hell
>broke loose.  The 66 with the PCI SCSI checked thru the disks and was
>booting the OS before the 90 even finished checking disks.  The 66 was
>at the login banner as the 90 was just starting up the OS.  The
>difference in disk speed was simply astounding, and couldn't be made
>any clearer visually.

>While this doesn't apply directly to the a 486 running a VLB card,
>believe me, buying an ISA-bus SCSI controller for a modern-day
>computer is definitely going to make the SCSI controller your main
>bottleneck for throughput.

How can the ISA-bus be a bottleneck for disk drives that can't even
push what the ISA-bus can sustain (ISA-bus maxes at 4-5MB/sec).  While
we are on this subject, can somebody tell me why an IBM RS/6000-950 can
read a SCSI-2 disk at 3500KB/s on a SCSI-2 bus but only gets 2500KB/s
on a SCSI-1 bus (from the same SCSI-2 drive)?  I thought the sustained
rate on SCSI-1 was 5MB/sec.

-- 
Benjamin Z. Goldsteen