Conference 7.286::digital

    I dont completely buy the fact the sales of VAX is low because of Alpha
    announcements, example, every one knows that Intel 586 is coming out
    in production 6-9 months from now, and its is at least twice as fast
    as 486, but that do not stop people from going out and buying 486's and
    even 386's PC's, right?

    actually, VAX sales is not low, sales of VAX's are doing better now than 
    befor ..(price cuts?), I heared this in a meeting 2 days ago ..
    
    any ways, many other companies do announce more advanced versions of their 
    products long ahead of time (dont they?), but that ought not to stop their 
    customers to buy their existing products (right?) otherwise no one will 
    buy anything, because there will always be better/faster versions of the 
    products coming out the pipe..

    /nasser

    Remember when Osborne announced his new machine and went bankrupt in 3
    months? The new machine was not ready yet and the sales dived.
    Symbolics did this when they announced IVORY also. 
    
    One problem with VAX has been that the price/performance was has been
    so low, that people have been buying because they were forced to and
    had no alternative. FIgure this, VAX 9000 has recently been the only
    way to buy 90 vups and it costs $500,000. Well Alpha promises 100-150
    mips in a deskside system at an order of magnitude price reduction.
    What would you do?
    
    I'm glad the VAX sales are picking up. This is a good sign. There is
    the VAXstation 4000-60 at 12 vups on the desktop, pretty decent. And
    its followers are faster I am told. I hope we push these successfully.
    
    On the 486, you don't have toi wait for 586 because 486 is plenty fast,
    however, with the VAX, its been that the performance has been a dog
    giving you a reason to wait for both better speed and lower prices/mip.
    
    I hope we can market are newer vaxes that are around, they are fast now
    and thank goodness we have them now!
    
    Tom
    

    
    	re .0
    
    	I agree with some of what you say. However, it is SOP for companies
    	to announce products before they are ready...it happens all the
    	time, and everybody does it.
    
    	It would have been extremely difficult to keep Alpha under wraps
    	for any significant amount of time if only because of the great
    	amount of software work that needed to be done, such as working
    	with CSOs to port over existing applications. And Alpha is not
    	"just another VAX", it is an entire new architecture.
    
    	Has the announcement cut into VAX sales? Probably. How much?
    	Don't know. Is it the sole cause of headcount reduction? No.
    	Will Alpha solve all of DEC's problems? No.
    
    	This waiting around to see who will get cut next is demoralizing
    	and depressing. In my opinion, the company should just do it and
    	get it over with, so that people can get on with their lives
    	instead of having to live under the Sword of Damocles, as you
    	suggest. Waiting just adds to the stress and tension for everybody.
    	That layoffs will occur is a foregone conclusion; we can only hope
    	that fair and objective criteria are used to select those who will
    	be TFSOed.
    
    	john

    The downsizing of Digital did not start AFTER, but BEFORE the financial
    losses.  In 1985, Puerto Rico reduced its workforce to kick off the
    process.  In 1988 and 1990, we did it again.  And now we're closing up
    completely.
    
    The fact is that you can't blame our losses on anything other than
    bloated costs and reduced demand for our products.  No recent action,
    such as announcing ALPHA has affected either of those.  The bloated
    costs are being attacked by eliminating the cause of them, people whose
    jobs add cost, not value.  By the way, management is responsible for
    these unneeded jobs, not the people doing them.  If and when your job
    is eliminated, its your boss (and/or his boss at some level) who
    failed, not you.
    
    Apparently, nobody knows how to increase the demand for our products.
    
    /rab

    I heard on the morning news (a coupl edays ago) that DEC was rolling
    out a series of vaxen that were alpha compatible. This way customers
    can buy newer faster machines while waiting for Alpha.
    
    Gregg

    
    	Re: .1
    
    >I dont completely buy the fact the sales of VAX is low because of Alpha
    >announcements, example, every one knows that Intel 586 is coming out
    >in production 6-9 months from now, and its is at least twice as fast
    >as 486, but that do not stop people from going out and buying 486's and
    >even 386's PC's, right?
    
    	I'd like to have a 486-class machine for what I'd like to use it
    for, but I don't have to have it today.  Given what is probably going to
    happen to the prices of these systems once the 586 is out, I'm waiting.
    I, for one, don't mind being a generation behind if that saves me a lot
    of money and only incurs a wait of a half year or so.
    
    							-craig

re: .4, /rab

>    Apparently, nobody knows how to increase the demand for our products.

This is probably worth a topic all its own, but it probably belongs in
MARKETING, which I don't follow.

I've often wondered if the above isn't a big problem for DEC. When I was
an OEM 15 years ago, I always found myself selling to my clients on the basis
of what I could conceptualize for them above and beyond what they thought
they needed. Granted, customers are almost infinitely more computer literate
than they were in the mid seventies. But I still think that DEC may suffer
from a mode it developed in its formative years, where the equipment sold
itself and DEC didn't have to do much in the way of imagining new uses.
Yes - we've got lots of new innovative software and hardware products to
address a variety of horizontal and vertical markets. But do we have anyone
who bothers to think about how we could actually "create" demands for
our products?

-Jack

I have to agree with .0 in the assessment of Alpha on our current sales of Vaxen.
One of the most closely guarded secrets within IBM is new product data - simply
to keep from killing current sales.  They noticed a long time ago that when they
announced a hot new machine the current machines couldn't be given away.

Qualifier - I think this phenomena is well known throughout mid-frame and
main-frame makers.  I'm not so sure you can extrapolate to the desktop the
same situation - we're dealing with significantly different markets and
cost.

Do I think they could have kept alpha under wraps?  No.  But the really should
have had the *alpha-ready* program in place.

As the philosopher once said, Those who do not learn from hisotry are doomed
to repeat it.



Charlie

re .2:

>    the VAXstation 4000-60 at 12 vups on the desktop, pretty decent.

Actually, it's 10.5 VUPS.

    Not only is the fundamental premise of .0 flawed -- demand (in dollars)
    for minicomputers is not down because people are waiting for Alpha, but
    largely because price is falling faster than demand is rising -- but
    Alpha IS a product today.  From VTX PRICE:
    
     OPTIONS     PRICES AS OF:  10-JUL-1992         THE U.S. SYSTEMS PRICE LIST
     
     Model           Product
     Number          Description                    List            Standard
              
     21064-AA        64-bit Alpha CPU 150MHz        3,375.00         N/A
    
    Also, we need to get Alpha protos to ISVs, OEMs, etc., so that they can
    have software running on them before "general availability" of
    "Alpha-based computer systems".
    
    And this is a MARKETING question anyway, but I thought I'd refute the
    assumption that ALPHA (which is merely a chip) isn't a product yet.
    Heck, the 21064-SA (developers' kit) is in the new Sales Update.

.7>  ...do we have anyone who bothers to think about how we could
.7>  "create" demands for our products?

    No.

     ...do we have anyone who should think about how we could
     "create" demands for our products?

    Yes -- Marketing.

    

Lighten UP!

The February announcement was for the Alpha Chip. We are, for the first time
in my memory selling the CHIP to third parties. In the chip business it 
is customary to announce the chip shortly before availability of sample
quantities. Since we are a manufacturer of systems it is necessary for
us to make some comment about what the new chip would do to our product
line. The problems were caused by too little information be given out. I think
that if we were more forthcoming about VAXfutures for July and beyond
in February we would not have had quite the negative impact that it did.

There are also anti-trust implications for this. Microsoft has been sued
more than once for supposedly giving its internal product developers 
(Excell...) access to new versions of Windows
prior to public availability to the likes of LOTUS and Borland. They haven't 
lost any to my knowledge but the conservative legal department at D.E.C will 
no doubt be pushing for public annoucement and sale at a point that is 
defendable if in the future Alpha becomes widely used by third parties and we 
get sued. It sets a precedant for good behavior.

Regarding the Q3 loss. There was curious statement in the press release that
up to $200M of the loss was due to foreign currency fluctuations. This
is curious in that there is a group that is supposed to manage our risk
associated with foreign exchange. Guess they didn't do so well.

>re .2:
>
>>    the VAXstation 4000-60 at 12 vups on the desktop, pretty decent.
>
>Actually, it's 10.5 VUPS.
    
    	 What's a couple of VUPs between co-workers?   :-)
    
    	 A little more seriously though, the 4000-60 is a *very* nice 
    	machine. I received mine a few weeks ago. However, for those of 
    	us who own a 4000-60 and have co-workers that don't, it should 
    	be sold with some type of industial cleaner and a towel to take 
    	care of the finger prints and drool.
    
    	 Even more seriously, since the 4000-60 is such a great machine,
	I can only imagine what the 4000-90 can do. I believe these
    	machines should be an excellent example of hardware that "sells 
    	itself". I've heard established external customers complain about 
    	VAX worksation speed in general for several years now so the demand 
    	is out there. The real question (he asks aloud) is "will the 4000-xx 
    	machines will ever have the chance to become established in the
    	market?", particularly considering the marketing & sales value of
    	"word of mouth from a satisfied customer" and the amount of time
    	it usually take to establish that type of reputation.
    	
    	 If the demand for powerful "VMS functional" systems exists, the
    	supply exists and the cost per MIP/VUP/SPECmark (or whatever) is
    	"acceptable", what else would be considered a major factor in
    	dwindling sales? My bet would be stability, depending on your 
    	particular attitude it could be stability in the product OR it 
    	could be stability in the producer. Both are probably very major
    	factors in the current state of DEC business and I'm don't see
    	how arguing over a little nit here and there will fix that type
    	of thing. 
    
    	 My vote is for a well thought out, common sense, consolidated
    	plan for a future, *not* a mish-mash of messages, products and 
    	empire_building_power_hungry_chiefs that jump all over each other.
    
    
    	My 2 cents...
    
    	Glenn

    
    Note that the VAXstation 4000 model 90 has just been announced this
    week and is rated at 24 VUPS or 32.8 SPECmarks. I believe this to be
    the fastest workstation we sell.
    
    -Joe
    

re: .11

Thanks for the reassurance, Bill.
:^)
-Jack

    	Gee, it kind of boggles the mind, then, that even people working on
    Alpha are going to be laid off. Of course, this isn't expected to
    impact any V1.0 schedules. People will just be moved around to fill any
    voids without missing a beat. Everyone will just have to work
    harder...but with morale as high as it is, that shouldn't be a problem.

    Re .12 (Parasite):
    
>The February announcement was for the Alpha Chip. We are, for the first time
>in my memory selling the CHIP to third parties.  ...
    
    Well, you're perhaps a young-timer.  We did this for a while with
    PDP-11 chips.
    
    Steve Kallis, Jr.

re Note 1976.17 by HELIX::KALLIS:

> >The February announcement was for the Alpha Chip. We are, for the first time
> >in my memory selling the CHIP to third parties.  ...
>     
>     Well, you're perhaps a young-timer.  We did this for a while with
>     PDP-11 chips.
  
        And there was a PDP-8-compatible chip, not actually
        manufactured in our plants but I assume licensed by us (the
        later DECmate-n products used it).

        (I am assuming that they were actually sold, or at least
        offered, on the open market, and not just manufactured for
        our exclusive use.)

        Bob

	Speaking of PDPs, how did we manage the VAX announcement, and
	did it eat into PDP sales? I wasn't around then, is there
	anyone who was, and can tell us how that transition was 
	handled?

				Tom_K

    I cannot tell you how we managed the VAX Annoucement.
    
    However, no it did not eat up sales.
    
    PDP-11 was booming and a late-generation PDP-11 was taking off.
    
    VAX was like firing the second barrel, it went up as well.
    
    Back then, of course, DEC, the computer/high-tech industry, and the
    global economy were in more growthful shape.
    
    It hardly seems relevant to the present situation:
        -  VAX 'in decline' -- or at the very least perceived to be
    	-  DEC in financial stress
    	-  a mature[ing] industry
    	-  global economic stress -- recession/stagnation, emerging
    	     governments, Europe'92, etc.
    
      Ann
    

    Re .19 (Tom_K):
    
    When we announced the VAX-11/780 system (the first of the family), we
    announced that it had "native mode" and "compatibility mode" features:
    the latter was a PDP-11 m ode, where it would run RSX-11-based PDP-11
    software.  This was done because:
    
    a) There was little available native-mode VAX software; and
    
    b) because the PDP-11 was _immensely_ popular at the time, and the new
    computer could be considered an extension of it (the "-11" in the
    model name was to stress the "elevenness" of the new line).  
    
    If one was contemplating buying a new machine to do PDP-11 applications
    with no thought of developing new native-mode applications, the
    price/performance advantage wasn't there, and sticking with the older
    product was a better alternative.  Also, that first VAX computer was
    much bigger than any available PDP-11 system (the PDP-11/70 was the
    closest, and other factors were affecting its sales).  A large
    percentage of PDP-11 computers were OEM devices, which called for
    smaller UNIBUS and Q-bus based systems.
    
    In short, the new systems complemented the old, and were not a direct
    competition for most markets.
    
    Steve Kallis, Jr.

We ran into this with the VT100s.  For about 18 months there was a big
VT52 clone industry, between announcement of the VT100 and availability.

    
    I seem to remember that we did cancel the PDP-11/74 ( multiple 11/70
    cpu's in one cab ) because of fear it would eat into the new VAX 
    marketplace.
    
    There was also competition for market and resources between the Jupiter 
    machine ( TOPS ) and the large VAX of the moment (8600 maybe?).
    
     Chuck 

    
    	re .14 	"24 VUPS or 32.8 SPECmarks"..
    
    	This is the first time I've seen a VUP equate to a SPECmark. 
    	Can it be said then, that a SPECmark is approximately 1.3 VUPs?
    	I know that this is off the subject a bit, but I'd like to know
    	so that I can have a yardstick to understand how powerful these
    	new machines are.
    
    	Thanks,
    	John

    
    The VAX product line did kill the DECsystem 10's and 20's. Jupiter
    was the ultimate sacrifice. Both TOPS-10 and TOPS-20 were great
    operating systems. Remember the 'If you're not playing with 36 bits
    you're not playing with a full DEC' buttons? Was that really ~10 years
    ago... Time flys.
    
    		fred deady
    

    re .25
    
    Revisionist History, Rose Coloured Glasses or whatever :-).
    
    Many of the factors leading to the demise of the Jupiter were of its
    own making. Like the project being very late, over budget, not
    delivering the planned performance etc:. If Digital had been equally
    ruthless with engineering projects since then we might not have been in 
    such bad shape as we are today.
    
    Being part of the Corporate Sales Office in those days, we had to work
    out which customers had been given the 1100 or so Non Disclosure
    Presentations on Jupiter and how to reset their expectations. 

    
    SPEC is a VUP by definition.  Your run some set of tests than
    normalize to how that test ran on an 11/780.  The VUP is
    the suite of 99 benchmarks, the SPEC is a different set of tests.
    
    With NVAX products we added the KUCK (KAP) pre-processor and some
    of the tests took a big jump.
    
    Jon
    

    re .27
    
    SPEC is not a VUP by definition.  Yes, SPECmarks are relative to the
    CPU speed of an 11/780.  But that is where the similarity ends.
    
    VUPs are a DEC measure of CPU speed based on several internal
    benchmarks.  It compares VAX systems to other VAX systems.  It is a
    Digital measurement.
    
    The SPEC89 benchmark suite is a set of 10 benchmarks put together by
    the SPEC consortium.  This is a group of a bunch of vendors - including
    DEC - who work together, sort of, to come up with a set of benchmarks
    that measure CPU performance across architectures.  So you can compare
    the CPU performance of SUN, HP, DEC, brand-X systems based on
    SPECmarks.  
    
    Until a couple years ago, a SPECmark turned out to be roughly a VUP. 
    Then the vendors figured out a way to optimize some of the SPEC tests
    and everybody's SPECmark numbers inflated.  Except ours.  We did not do
    the same optimizing initially because these made the benchmarks less
    realistic.  But we started with the products announced in Oct 91.
    
    This is where the Kuck preprocessor comes in. 
    
    Meanwhile, the SPEC consortium got together and decided that they would
    change the SPEC suite of benchmarks because everyone had optimized the
    initial set to the point where they were misleading.  So the original
    SPEC suite is now called "SPEC89" and the new suite is "SPEC92".  
    
    But everyone does not report SPEC92 numbers yet.
    
    - Greg

    And on the subject of announcing and talking about Alpha . . .
    
    Remember the situation a couple years ago?  The customers and sales
    reps around here were asking some tough questions about the future of
    our products.  You saw all the analyst coverage.  RISC performance was
    going to go straight up over the next few years, CISC performance would
    not come close to keeping up.  UNIX (somebody else's!) was going to
    take over the world.  Price/performance, raw CPU performance, and OPEN
    became the popular buzzwords.  They still are today.  And all of a sudden, 
    every VAX out there was positioned as a dinosaur.  
    
    People asked me all over the place, why should I buy DEC stuff today
    when your future looks like you will *never* be competetive?  From
    where I sit, we were looking out over the edge of a really nasty cliff.
    
    I know times are tough today, but I still believe times would be *lots*
    tougher had we not gone public with Alpha.
    
    Just my opinion.
    
    - Greg

    Oh dearie dear. If vaxen is used and people pronounce DEC as "deck",
    then we're all in deep trouble.
    
    That's deep trouble people. Very deep indeed. It's the end of the world
    as we know it.
    
    How absurd.

re: .19, .20, etc.

It was my understanding that the 11/74 was a bit of a dinasour.  Sales never
got off the ground and dual processing wasn't realy in yet....

The PDP-11/60 came out about the same time as the 11/780.  It was a moderatly 
priced, high end PDP-11, and well packaged system in a wide semi-low-boy cab.
The problem was, people bought VAXs instead!  As it turns out, Ed. Services was
the largest single customer both internally and externally!!!

>It was my understanding that the 11/74 was a bit of a dinasour.  Sales never
>got off the ground and dual processing wasn't realy in yet....

    The 11/74 was never offered for sale. Though a few were built and I
    believe there was a field test. The RSX development people shut theirs
    down only a year or so ago I believe. I believe it would handle quad
    processors not just dual. And it was SMP which is why I was surprised 
    that the 11/785 was ASMP. It was fast and would very likely have hurt
    VAX sales.

>The PDP-11/60 came out about the same time as the 11/780.  It was a moderately 
>priced, high end PDP-11, and well packaged system in a wide semi-low-boy cab.
>The problem was, people bought VAXs instead!  

    The big edge the 11/60 had was that you could write your own micro code
    (ie add instructions) for it. There were not that many people who could
    or who wanted to take advantage of that. Digital used it internally to
    emulate the PDP-8 for PDP-8 software development BTW. If we'd had some
    good micro code packages that accelerated specific applications or types
    of applications it might have sold better.

    			Alfred

>The PDP-11/60 came out about the same time as the 11/780.  It was a moderatly 
>priced, high end PDP-11, and well packaged system in a wide semi-low-boy cab.
    
    As I recall, the 11/60 was due out well before the 11/780, but was late
    so came out at around the same time.   It had an internal tag of "PDQ"
    attached to it - a pretty d****d quick machine - and there was some
    comment around that the implementation project was also supposed to be
    pdq but wasn't.   It had some cache and user-programmable writeable control
    store, but only 128K of memory, max.   Ed. Services probably bought them
    to be able to host courses on wcs.
    
    Then, though, Digital was prospering enough (and there was little enough
    competition in the market) that people could laugh the 11/60 off.   
    The aspects of innovative technical spec. but late delivery and failure
    to match main trends in the market, sound familiar, though...
    
    On .19 and .20 and how the VAX announcement was handled, I seem to
    remember (working for a customer by end 1978) that the announcement was a
    bit confused, and caused comment that "Digital had got it wrong" (high
    cost, limited memory, limited storage, limited applications migration,
    limited compiler availability, small page size for VM, etc).   Of course,
    Digital got a bit larger after that.   But some of the comments were
    probably accurate all the same.   Not sure what that shows, apart from
    pundits having limited relation to the marketplace, and excellent
    engineering being pushed into products that were sometimes less than
    excellent, but succeeding anyway through market presence.

    
    
    I used to use variations to the following argument many times
    before:
    
    If waiting for Alpha is what really hurts VAX sales, why 
    aren't there more ALPHA licencees queuing in to get a bit
    of the cake (there's still only three - minor - firms
    Cray, Kubota and Olivetti ) ???
    
    						helmut
                                                 

    
    The ALPHA announcement only exacerbated the real problem: the
    trend in the industry called downsizing. People are moving to 
    cheap, industry standard platforms, linked together in client
    server arrangements. VAXes aren't part of this trend. 
    

RE:        <<< Note 1976.24 by TEXAS1::SOBECKY "It's all ones and zeros" >>>
                            -< VUPS vs SPECmarks? >-

    
>    	re .14 	"24 VUPS or 32.8 SPECmarks"..
>    
>    	This is the first time I've seen a VUP equate to a SPECmark. 
>    	Can it be said then, that a SPECmark is approximately 1.3 VUPs?

SPECmarks are computed by comparing current benchmark times with 1989 reference
times from a VAX-11/780, which probably would rate at 1.2 or 1.3 today due to
compiler improvements.

VUPs are calculated by comparing the new machine with a VAX-11/780, both with
the most recent software.

/d

>    The 11/74 was never offered for sale. Though a few were built and I
>    believe there was a field test. The RSX development people shut theirs
>    down only a year or so ago I believe. I believe it would handle quad
>    processors not just dual. And it was SMP which is why I was surprised 
>    that the 11/785 was ASMP. It was fast and would very likely have hurt
>    VAX sales.

The 11/74 had two changes over the 11/70. It supported 4 processors, and it had
microcode and data path support for the PDP-11 Commercial Instruction Set. CIS
did not buy much over a better compiler for the base machine. The MP hardware
had some technical problems due to the physical aspects of connecting 4 
processors to shared memory.

The 11/785 was a Schottky TTL version of the 11/780 with the cycle time down to 
133 ns. You must mean the 11/782, which was sold as an ASMP. The hardware was
symmetric in that all memory (MA780) was accessible to both processors, but
each processor had its own SBI. VMS did not support SMP at that time. The
interesting thing about the 11/780 is that the SBI would have supported another
processor, but the 780's SBI interface logic was designed for it to be bus 
master, therefore you could not plug 2 CPUs into the SBI.

Dileep

	RE: .38 You're right, I did mean the 11/782. Thanks.

			Alfred

RSX-11M-PLUS was designed as the operating system for the multi-processor 
11/74.  The 11/74 was no dinosaur.  In fact, it blew the doors off the
"11"/780.  The decision to cancel was very upsetting.  It was my melancholy
duty to remove all references to multi-processing from the operating system
documentation.

The multi-processor wasn't very fancy.  Instead of one processor on the 
Unibus, there were four.  (simplified explanation)  It worked fine and would 
have materially affected VAX sales.  I believe we made the correct decision 
in not selling it, although at the time the excuse was that training and 
field service couldn't ramp up to supporting both the VAX and the 11/74 at 
the same time.

There were multi-processor hooks in the J11 chip and we sometimes fantasized
about that.