is the 3B2 dead

[Stephen J. Friedl]

In article <920 at gistdev.gist.com>, flint at gistdev.gist.com (Flint Pellett) writes:
> I don't believe anyone has yet defined "dead" as "useless".  The 3B2/600s
> on up are pretty nice machines, and are quite useful.

I have been watching this discussion on the death of the 3B2 for
some time wondering who would finally say it, so I will.  The
unfortunate answer is that the 3B2 line is probably not going
much anywhere.

First of all, those who say "AT&T has just released XX and YY
so they would never kill it now" are relying on the good sense
of a company that has not always demonstrated that this is
their strong suit.  They might do the right thing anyway, but
it's just as likely to be by accident than by design.

Note: I have a very high respect for the technical folks at AT&T
but they are not the ones making these kinds of decisions.

Another note: this is just my "market analysis", which is kind
of a stretch for me because I am primarily a technical guy.  Those
who have a good mind for this kind of thing are encouraged to take
it all with a grain of salt.


So, let's see where one could come to the above conclusion.  The
first question to ask is whether they are still doing development
on the microprocessor that forms the base for the entire 3B
family.  I believe they are not.  How many design wins did the
WE32?00 get outside of AT&T?  I know of two: Zilog and
Microproject.  Zilog had a 3B-compatible CPU boardset for use in
their aging Z8000 systems, and it was supposed to be pretty hot.
I know that the IRS bought sh*tloads of these Z8000 machines, but
I don't know if anybody upgraded to the new boardset.

Microproject was making their own box based on AT&T's VME card,
but they ran into financial/technical difficulties and I don't
even know if they are still around.  I remember getting a letter
from them saying that they found horrible bugs in the CPU and
that they were suing AT&T and fully expected to win and for the
time being their AP was under tight control.  Ahem. This is the
last I ever heard from Microproject.

The WE32000 family was built from the ground up to run UNIX and
is basically wonderful, but I have been told that CPU development
stopped about a year and a half ago, and it was too bad because
the mask was probably capable of 35-40MHz as opposed to the 24MHz
that they are running now.

If you don't have a faster CPU then you have to think about putting
them together in a box so they work in parallel.  I see all of you
waving your hands saying "the 3B4000!" but this is not the answer.
The product is clearly not going anywhere, and there is a good chance
that it is not even orderable anymore.

If you look further you will see that AT&T has attacked the high
end by OEMing Pyramid machines using the MIPS chip.  This is clearly
a departure from the 3B2 line (even though the name might be the
same, I'm not sure) and does not bode well for the high end.

Look on the other side and you see the low end being dominated by
the Intel machines.  These machines are becoming fast so quickly
that AT&T could never keep up with them with the 3B family.

It is clear to me that the 3B2 is being squeezed out on both
ends, and that it is not the direction of the future.

On the other hand, it won't die the same painful death as the
7300.  That machine never had real market acceptance (and was
not even made by AT&T) so they never hurt that many people
that mattered to them (college students don't matter).

The 3B2 family is being used in the Air Force contract, so it
is a fair bet that vendors will continue to support it (say,
WordPerfect and Informix) so you will not be left high and dry
like those unfortunate folks like me who have 3B15s.  Note that
the goverment price for these machines is ridiculously low --
it is purported to be only hundreds of dollars over AT&T's
manufacturing costs.  I have seen (but cannot repeat) some of
these prices and I thought they were typos.

It doesn't strike me that this Air Force deal is making AT&T
a killing, but my information on this is third hand and I am
even more out of my area of knowledge than than the rest of
this posting :-).

All told, the 3B2 will be around but it is not the place to
hitch your star for the future.  If you are maxed out on your
current configuration, you should think about going a different
direction (maybe even the Pyramid, but my choice would be a
Sequent).

This note has a great deal of speculation and second-guessing
and may even be totally off base.  Those wishing to agree or
disagree are encouraged to send mail, and off-the-record
comments are *very* welcome.  Let me know if I can repeat stuff
if anonymity is maintained.

     Steve

P.S. - I wonder if I should think about replacing my ``3B2 UNIX''
       license plate?

-- 
Stephen J. Friedl, KA8CMY / Software Consultant / Tustin, CA / 3B2-kind-of-guy
+1 714 544 6561  / friedl at mtndew.Tustin.CA.US  / {uunet,attmail}!mtndew!friedl

Sat-is-fac-tion (sat' is fak" shen) n: Seeing Pick Systems cave into UNIX