[comp.sys.ncr] NCR's 4000 series to replace towers???

barton@holston.UUCP (Barton A. Fisk) (04/05/91)

I have heard that people who used to order towers are now
ordering the 4000 series instead, is this true?

This is understandable since the 4000 boasts high performance
enhancements.

Does this mean that the tower is destined for the 
unsupported list???
-- 
uucp: holston!barton
pseudo: barton@holston.UUCP

nolan@helios.unl.edu (Michael Nolan) (04/05/91)

barton@holston.UUCP (Barton A. Fisk) writes:


>I have heard that people who used to order towers are now
>ordering the 4000 series instead, is this true?

First of all, it's the 3000 series, not the 4000 series.

>This is understandable since the 4000 boasts high performance
>enhancements.

Only at the higher end (3400 and above).  At the low end the 3000's are just
NCR's PC's with a slightly faster transfer rate (maybe) and unix.

>Does this mean that the tower is destined for the 
>unsupported list???

Well, eventually EVERYTHING goes in the bit bucket.  NCR's official policy
(as told to those in attendance at the NCR Unix User Group meeting last fall)
was that Tower development and production continues, and that the 'market
will decide'.  Yeah, right!  Perhaps some additional word on this will come
out at NUCON in San Antonio in 4 weeks, but I expect that the Tower 750 will
be the last new Tower.  (Assuming that the 750 is ever released as anything
other than an upgrade kit.)  NCR has said that SVR4 will probably NOT
be ported to the Tower, only to the 3000.  (This may change if AT&T gets
control of NCR.)

One thing NCR has been very good at in the past is continuing support for
older systems long past their sales lifetime.  (Tower XP and mini-Tower 
buyers may flame me for that statement, though.)  I'm not getting rid 
of my 1 1/2 year old Tower/450 right away, although I might get a 3400
just to have something 'interesting' to play around with.

Michael Nolan
nolan@helios.unl.edu

borcher@cs-col.Columbia.NCR.COM (Tom Borcher) (04/08/91)

In article <5900@holston.UUCP> barton@holston.UUCP (Barton A. Fisk) writes:
>
>I have heard that people who used to order towers are now
>ordering the 4000 series instead, is this true?

It is System 3000 (not 4000) and some people are opting for the new
stuff...human nature, you know

>Does this mean that the tower is destined for the 
>unsupported list???

Once any product is released it is destined for the unsupported list,
only the E.T.A. is unknown

vaigl-j@mad.cis.ohio-state.edu (James Vaigl) (04/09/91)

Someone said earlier that "the lower end sytem 3000's were nothing more
than PC's"  Well, having worked at NCR, (no longer) I can say that in
most of the Unix benchmarks we ran, the 916 (a 16Mhz 386) beat a 700
by *wide* margins on many tests.  There is not a tower made that can
come close to the 3433 ( a 486 at 33Mhz ).

BTW, I just stumbled onto this newgroup.  Is it for NCR people, customers,
the general public...?

--Jim

dick@ahds.UUCP (Dick Heijne CCS/TS) (04/11/91)

In article <105173@tut.cis.ohio-state.edu>, vaigl-j@mad.cis.ohio-state.edu (James Vaigl) writes:
> (...)  Well, having worked at NCR, (no longer) I can say that in
> (...)
> BTW, I just stumbled onto this newgroup.  Is it for NCR people, customers,
> the general public...?
> 
For everyone, excluding those who deserted NCR :-))
NO, ARGHH! JOKE!

Dick.

barton@holston.UUCP (Barton A. Fisk) (04/11/91)

In article <nolan.670834286@helios> nolan@helios.unl.edu writes:
>barton@holston.UUCP (Barton A. Fisk) writes:
>
>>I have heard that people who used to order towers are now
>>ordering the 4000 series instead, is this true?
>
>First of all, it's the 3000 series, not the 4000 series.
>

Ooops, exuse the finger slip, I meant 3000.

Question:

Do the high end (ie. 3400 and >) support symetric multiprocessing?

If so, what software is used to achieve this. Does NCR resell SCO
MPX?

How many processors are supported. I'm only interested in 486/33's.

Our service depot people here don't seem to know that much about it.
-- 
uucp: holston!barton
pseudo: barton@holston.UUCP

vaigl-j@carp.cis.ohio-state.edu (James Vaigl) (04/12/91)

The 3433 is a 486-33 uniprocessor.  I think you have to get to level
35xx before you start talking about multiprocessors.  

I don't know about Unix, but I'd say that it will be a *long* time
before you see a version of OS/2 that supports MP platforms, and when/if
it does, i'd say the odds are overwhelmingly against really symmetric 
use of processors.

--Jim

bill@wrangler.WLK.COM (Bill Kennedy) (04/12/91)

In article <106632@tut.cis.ohio-state.edu> vaigl-j@carp.cis.ohio-state.edu (James Vaigl) writes:
>The 3433 is a 486-33 uniprocessor.  I think you have to get to level
>35xx before you start talking about multiprocessors.  

I'm a little distant from the subject with this but I'm curious about a couple
of things.  There's no question that ssbn (33MHz 386, 12MB) outperforms
wrangler (32/400, 4MB) but wrangler is superior for intensive serial I/O.
I realize that this doesn't have anything to do with reality, general purpose
Pee Cee cloneware (ssbn) is pretty puny with serial I/O.  My point is that
the creaky old 400 serves a vital role here running the three modems and
system printer.  It's not snazzy but it's even tempered and reliable.  Each
time I feel like it's too pokey I just go log in on the 3B2 :-)

>I don't know about Unix, but I'd say that it will be a *long* time
>before you see a version of OS/2 that supports MP platforms, and when/if
>it does, i'd say the odds are overwhelmingly against really symmetric 
>use of processors.

I agree with Jim, it's going to be a while before symmetric multiprocessing
reaches down to the class of machine that you can pick up by yourself.  It
reminded me of a rumor I heard that the final buy for 32/450 and 32/650 is
set for late June of this year.  It sounds reasonable and credible, has
anyone heard the same thing?  I admitted I was a bit off the subject, but
day-to-day computing realities in Pipe Creek, Texas (yes, it's as remote
and small as it sounds :-) tend towards asymmetric multicomputers.  There
is still a lot of crunch left in the antiques and as they discontinue they
come into the budget reach of mere mortals.
-- 
Bill Kennedy  uucp      {att,cs.utexas.edu,pyramid!daver}!ssbn.wlk.com!bill
              internet    bill@ssbn.WLK.COM   or ssbn!bill@attmail.COM

wescott@Columbia.NCR.COM (Mike Wescott) (04/12/91)

In article <106632@tut.cis.ohio-state.edu> vaigl-j@carp.cis.ohio-state.edu (James Vaigl) writes:
> The 3433 is a 486-33 uniprocessor.  I think you have to get to level
> 35xx before you start talking about multiprocessors.  

Nope.  The high end of the 34xx will be MP.

--
	-Mike Wescott
	 mike.wescott@ncrcae.Columbia.NCR.COM

sleepy@wybbs.mi.org (Mike Faber) (04/15/91)

In article <5904@holston.UUCP> barton@holston.UUCP (Barton A. Fisk) writes:
>In article <nolan.670834286@helios> nolan@helios.unl.edu writes:
>
>Do the high end (ie. 3400 and >) support symetric multiprocessing?

This was one thing that NCR was VERy vague about.  They're going to use
(they say) loosely coupled processing, and the 3700 and 3800 are suppposed to
supercede all supercomputers in terms of MIPS.  The lowest level 3000 that is
multi processor - the 3500 - uses tightly coupled architechture, and will
use up to 6 486/50's.
>
>If so, what software is used to achieve this. Does NCR resell SCO
>MPX?
I don't think NCR knows whether they will.  Our rep will only be VERY vague
about what the high end stuff will be like.
>
>How many processors are supported. I'm only interested in 486/33's.
Which model?  The 3800 is supposed to use hunderds. (and do over 100,000 MIPS!)
>
>Our service depot people here don't seem to know that much about it.
Join the crowd.

--
sleepy@wybbs.uucp
Michael Faber

nbladt@aut.autelca.ascom.ch (Norbert Bladt) (04/15/91)

vaigl-j@mad.cis.ohio-state.edu (James Vaigl) writes:

>Someone said earlier that "the lower end sytem 3000's were nothing more
>than PC's"  Well, having worked at NCR, (no longer) I can say that in
and he is right. So, where is the difference between NCR and any other
company selling PC's for low-end (?) UNIX users ?

>most of the Unix benchmarks we ran, the 916 (a 16Mhz 386) beat a 700
>by *wide* margins on many tests.  There is not a tower made that can
>come close to the 3433 ( a 486 at 33Mhz ).
Now you can see how badly designed some expensive hardware can be !
I worked with a similar system as the tower several years ago and it
was much faster than our Tower 32/650.
I agree to the difference in speed, performance (and price !), we have a
Toshiba Laptop (5200/100, 20 Mhz) which is faster than the Tower. Not to
mention our no-name taiwanese 486/25 Mhz SCSI system !

>BTW, I just stumbled onto this newgroup.  Is it for NCR people, customers,
>the general public...?
I always thought, this group was for NCR people and customers which have
problems with their NCR systems (there are a lot: X.25, Xstation, etc.)

Norbert.
-- 
Norbert Bladt, Ascom Autelca AG, Worbstr. 201, CH-3073 Guemligen, Switzerland
Phone: +41 31 52 95 52			FAX: +41 31 52 95 44
Mail: nbladt@autelca.ascom.ch  UUCP: ..!uunet!mcsun!chx400!hslrswi!aut!nbladt

barton@holston.UUCP (Barton A. Fisk) (04/16/91)

In article <637@wrangler.WLK.COM>, bill@wrangler.WLK.COM (Bill Kennedy) writes:
> In article <106632@tut.cis.ohio-state.edu> vaigl-j@carp.cis.ohio-state.edu (James Vaigl) writes:
> >I don't know about Unix, but I'd say that it will be a *long* time
> >before you see a version of OS/2 that supports MP platforms, and when/if
> >it does, i'd say the odds are overwhelmingly against really symmetric 
> >use of processors.
> 


Well, my opinion is that OS/2 is another "New Coke" flop.


> I agree with Jim, it's going to be a while before symmetric multiprocessing
> reaches down to the class of machine that you can pick up by yourself.  It

My Systempro at work supports dual symmetric processors. I haven't tried
to pick it up lately. 
-- 
uucp: holston!barton
pseudo: barton@holston.UUCP

dmdc@ncrsea.Seattle.NCR.COM (Dennis M. Dooley) (04/17/91)

In article <391@wybbs.mi.org> sleepy@wybbs.UUCP (Mike Faber) writes:
>In article <5904@holston.UUCP> barton@holston.UUCP (Barton A. Fisk) writes:
>>In article <nolan.670834286@helios> nolan@helios.unl.edu writes:
>>
>>Do the high end (ie. 3400 and >) support symetric multiprocessing?
>
    Yes 3450 and 3550.

>This was one thing that NCR was VERy vague about.  They're going to use
>(they say) loosely coupled processing, and the 3700 and 3800 are suppposed to
>supercede all supercomputers in terms of MIPS.  The lowest level 3000 that is
>multi processor - the 3500 - uses tightly coupled architechture, and will
>use up to 6 486/50's.

   There is a 3600 and 3700 in NCR's future.  I am not aware of a 3800.

>>
>>If so, what software is used to achieve this. Does NCR resell SCO
>>MPX?

   UNIX SVR4 provided by NCR.  I can't speak for what SCO might do.

>I don't think NCR knows whether they will.  Our rep will only be VERY vague
>about what the high end stuff will be like.
>>
>>How many processors are supported. I'm only interested in 486/33's.

   To date, the 33Mhz is utilized on the uniprocessor platforms.  The MPs
   are slated to use the 50Mhz part.

>Which model?  The 3800 is supposed to use hunderds. (and do over 100,000 MIPS!)
>>
>>Our service depot people here don't seem to know that much about it.
>Join the crowd.
>
>--
>sleepy@wybbs.uucp
>Michael Faber

   There appears to be a general confusion on NCR's System 3000 family
   that I would hope to clear up.
   
   The NCR System 3000 platform is made up of 7 computing (and price-point)
   levels.  All levels incorporate the Intel ix86 processor which provides
   compatibility and scalability within the product family.  The general
   break-down is as follows:
   
   Level 1   Portables.  Notebook and Notepad systems
   
   Level 2   Entry level systems.  Desktop systems.
   
   Level 3   Large desktop systems.  Large desktop client/entry level server.

   Level 4   Tightly coupled symmetric multiprocessor.  Currently this is
             to be based on the 50Mhz i486 and is scalable from 1 to 4
             processors.
             
   Level 5   Tightly coupled symmetric multiprocessor.  Currently this is
             to be based on the 50Mhz i486 and is scalable from 2 to 8
             processors.
             
   Level 6   Loosely coupled multiprocessor technology.  Regional/Enterprise
	     OLTP/ Decision Support.
	     Scalable to hundreds of processors.  Capable of > 4,000 MIPS.
	     Up to 250 TPS.  Disk capacities over 100GBs.  Uses disk array
	     technology.

   Level 7   Loosely coupled multiprocessor technology.  Enterprise OLTP/
	     Decision Support.
	     Scalable to thousands of processors.  Capable of > 100,000 MIPS.
	     Up to 60,000 TPS.  Disk capacities in the TBs.  Uses disk array
	     technology.

   Expect more announcements on the Level 6 & 7 on May 13, 1991.

   The following operating systems are supported:

     AT&T UNIX SVR4 on levels 2 -7

     SCO UNIX SVR3.2 on levels 2-4 (Uniprocessors)

     OS/2 on levels 2-4 (Uniprocessors)

     DOS on 2-4 (Uniprocessors)

  -----------
  NCR is not responsible for the comments that I have made here.
________________________________________________________________________

Dennis M. Dooley      VOICEplus  421-1790     NCR Corporation
ncrsea!dmdc           (206) 643-4150          15400 S.E. 30th Pl.
dennis.dooley@ncrsea.Seattle.NCR.COM          Bellevue, WA. 98007