Whatever became of the DEC 10/20 systems...
Richard Snider
replace the Decsystem-10 or 20's with something faster... I belive
it was called the Jupiter Project and I heard that it died..
Does anyone know why ??
Various sources here comment on the fact that the KL CPU took too
much power for what it did, to word-size, etc...
....Rich
rsnider@xrtll
Sarr J. Blumson
DEC apparently believed in a single instruction set product line. You
will receive many flames discussing whether this was incredibly stupid
and they killed the greatest architecture in history or they were very
smart and the 10/20 was competing only with the VAX anyway.
It appeared to be a pure marketing decision, at any rate.
(Sarr Blumson)
Jim Haynes
Sometimes when this quesion comes up somebody will point out that an
architecture with 18-bit addresses doesn't have much future anymore.
Maybe somebody will explain how Honeywell keeps its 36-bit family
alive after you run out of address space.
--
hay...@cats.ucsc.edu
hay...@cats.bitnet
"Ya can argue all ya wanna, but it's dif'rent than it was!"
"No it aint! But ya gotta know the territory!"
Meredith Willson: "The Music Man"
leic...@zodiac.rutgers.edu
the DECsystem 10/20 or VAX groups - but in a position to hear a lot of stories
- I'll tell you that the business is complicated.
It is quite true that major powers at DEC (led by Gordon Bell) were pushing
hard to get DEC to focus on the VAX, with VMS, as its one architecture for
the future. The reasons for this ran from business issues (it's much, much
cheaper to support and sell one line than many overlapping lines) to technical
(the 18-bit address space of the 10/20 line was definitely a liability, and
everyone knew it) to, no doubt, political (various personalities were jockying
for position).
It's also true that the 10/20 team was quite influential within DEC, 10/20
customers were very visible, and as a result despite the pressures from the
"VAX is the future" crowd, the Jupiter work WAS proceeding.
The great day of decision came when the planned engineering completion date
for the Jupiter arrived. Yes, the hardware worked. Unfortunately, it was
much slower than it was supposed to be - I'm told it was more or less as fast
as the previous-generation KL processor. The Jupiter engineers argued that
with more time (and money, of course) they could get the thing running up to
spec. And they probably could have. However, the market window for the
machine was rapidly closing. Customers had been waiting loyally for a KL
replacement for years. The machine had been extensively pre-announced and
discussed, including speed and date of shipment, within the 10/20 community.
No matter what DEC did, it would piss off and lose some of these customers.
DEC decided to cut its loses and pull the plug.
One thing the KL community never understood was that they really were, by
this time, small potatoes. DEC was on the threshold of an explosion in its
markets. Even a successful Jupiter would have quickly become insignificant
at the new scale of DEC's operations. DEC has never denied that it's a
business, out there to make money. The 10/20 community may have been DEC's
big money maker at one time, but I'm not even sure where they rated compared
to the very successful PDP-11 line of that period: The 11/70 was a HUGE
success.
It's sad that so many people, at both DEC and among customers, took this
all so personally. If they hadn't, VMS might have learned much more from
TOP-10 and TOPS-20; as it is, it learned almost nothing, a result of
deliberate policy. If they hadn't, DEC's relations with academia might not
have soured as they did. There might not have been nearly as much blood on
the floor.
But that's all a long, long time ago.
-- Jerry
Paul Repacholi
...
> Various sources here comment on the fact that the KL CPU took too
> much power for what it did, to word-size, etc...
>
The original PS is a linear regulator. Hence, most of the input power help run
up the Air conditioning bill
~Paul