Conference 7.286::digital

    
    If the graphics work on the clones, and the clones have a competetive
    advantage of leveraging the reference design, lower overhead, and
    clever engineers that tweak raw performance -- doesn't it call into
    question the reason for our worskstation group?
    
    The workstation group will soon be killed by their own strategy.  By
    concentrating on NT, and reducing their UNIX and VMS investments, they
    will be killed by the commodity business... because they will not be
    able to differentiate themselves in a box business.
    
    

	Couldn't someone just by the Pyxis boards? That is what
	a Powerstorm cards is, isn't it? 

							mike

    
    And didn't we just buy a 3d design company on the left coast to design
    graphics boards for Alpha.  Don't they compete directly with the APS
    graphics design group.
    
    

The following are JMHO...

re: .0

You (slightly) misread the page in question.  The Digital 500a won, hands down, on
a pure performance basis.  The NTSI clone had better price/performance, which reflects
the fact that our clone competitors don't have Digital's overhead to pay for.

Powerstorm 4DT was never supposed to be available on any clone.  It was supposed to be
the differentiator, the reason that someone might pay more for an honest-to-goodness
Digital workstation instead of a clone.  Apparently some clone vendors have done an
end run around us, and this is what is to be discouraged.

The theory is that it is better to keep some of the graphics proprietary, and leverage a
billion or two in workstation sales, than to sell the graphics in the open market,
and only make a few million.


re: .1

If we allow our efforts in graphics hardware and software design to be handed over to
the clone vendors at no cost, then you are correct -- Workstations will have a hard
time differentiating ourselves from the clone vendors.  Hence the desire not to sell
our "differentiator" to the clone vendors.


re: .2

No.  Pyxis is a chipset used on a motherboard as the interconnect between the CPU,
memory, and PCI bus.  The Powerstorm cards use a completely different set of chips.


re: .3

You are somewhat misinformed.  Yes, we acquired a graphics company (Megatek) in San
Diego to design graphics boards for Alpha workstations.  However, their work is more
complementary than competitive to the work being done in PKO.

    <<< Note 5205.5 by WRKSYS::mccasa.eng.pko.dec.com::DUTTON "There once
    was a note, pure and easy..." >>> reformatted for 80 columns

    The following are JMHO...

    re: .0

    You (slightly) misread the page in question.  The Digital 500a won,
    hands down, on a pure performance basis.  The NTSI clone had better
    price/performance, which reflects the fact that our clone competitors
    don't have Digital's overhead to pay for.

    Powerstorm 4DT was never supposed to be available on any clone.  It was
    supposed to be the differentiator, the reason that someone might pay
    more for an honest-to-goodness Digital workstation instead of a clone. 
    Apparently some clone vendors have done an end run around us, and this
    is what is to be discouraged.

    The theory is that it is better to keep some of the graphics
    proprietary, and leverage a billion or two in workstation sales, than
    to sell the graphics in the open market, and only make a few million.


    re: .1

    If we allow our efforts in graphics hardware and software design to be
    handed over to the clone vendors at no cost, then you are correct --
    Workstations will have a hard time differentiating ourselves from the
    clone vendors.  Hence the desire not to sell our "differentiator" to
    the clone vendors.


    re: .2

    No.  Pyxis is a chipset used on a motherboard as the interconnect
    between the CPU, memory, and PCI bus.  The Powerstorm cards use a
    completely different set of chips.


    re: .3

    You are somewhat misinformed.  Yes, we acquired a graphics company
    (Megatek) in San Diego to design graphics boards for Alpha
    workstations.  However, their work is more complementary than
    competitive to the work being done in PKO.

    
    How can I be misinformed when I only asked a question.
    
    I didn't make a statement I asked a question!!
    
    /art

    
    
    >The theory is that it is better to keep some of the graphics
    >proprietary, and leverage a billion or two in workstation sales,
    >than to sell the graphics in the open market, and only make a few
    >million.
    
    
    Statement:
    
    Good theory, a billion or two in workstation sales, I like that!
    APS workstations are off 39%.  See not 5187.0
    
    Question:  Where are the billion or two in sales.  On some forecast
               sheet I suppose?   
    
    
    I sound more like Kratz everyday!  I don't know what to make of it. :>)
    
    /art

re:.7

from your .3:
>    And didn't we just buy a 3d design company on the left coast to design
>    graphics boards for Alpha.  Don't they compete directly with the APS
>    graphics design group.

Your "questions" make a statement and ask if it is true.
I responded to them as such.  No offense was intended.

>    Statement:
>    
>    Good theory, a billion or two in workstation sales, I like that!
>    APS workstations are off 39%.  See not 5187.0
>    
>    Question:  Where are the billion or two in sales.  On some forecast
>               sheet I suppose?   

One can engage in as much finger pointing as one likes on that subject.
I'm not interested in playing;  I'd rather go back to getting the next
kick-ass graphics chip out the door -- at least I have some control over
that!
   
>    
>    I sound more like Kratz everyday!  I don't know what to make of it. :>)
>    
>    /art

Every company needs people willing to occasionally remind the emperor
that he's naked....

    This discussion reminds me of Digital in the 1980s, when the VAXBI bus
    spec was withheld from add-on equipment vendors who manufactured
    peripherals that Digital also produced.  On several occasions, Digital
    sued vendors such as EMC and ClearPoint for daring to interface their
    vastly cheaper and better-performing disks and memory to the VAXBI. 
    This was done purely to protect Digital's captive disk and memory
    businesses, and it succeeded in driving most of the third-party add-on
    vendors, as well as customers reluctant to pay Digital's inflated
    prices, into the arms of the competition.
    
    M 

I don't see the similiarity...  we're not stopping anyone from
adding *their* PCI-based graphics into our systems, or their own
clone systems, and selling them.  We're simply objecting to them
hijacking our hardware/software, and selling it.  

If the clone vendor was selling systems with VMS or Digital UNIX
installed without having paid us the licensing fees, we'd object,
yes?

    Do we sell PowerStorm card?  If so, why can't clone vendor buy it and
    put it in the system?
    
    It would be dream come true if thousands of clone vendors were to sell
    VMS or Digital UNIX systems to millions of customers!
    
    - Vikas

> We're simply objecting to them
> hijacking our hardware/software, and selling it.  

> If the clone vendor was selling systems with VMS or Digital UNIX
> installed without having paid us the licensing fees, we'd object,
> yes?

BUT, if they buy the graphics hardware, that specific purchase should
accurately reflect the true costs/profit/etc. involved (including all
appropriate overhead) right?  That should also be true with any software
costs (e.g. VMS on a "clone".)  If this isn't true, then why are we
effectively subsidizing the prices of these options from other places?

I find it - as ususual - relatively absurd that (once again) we appear
to be trying to invent engineering solutions to pricing structure problems
(i.e. spending real money trying to figure out how to sell fewer units
of something.)  I really wish we would concentrate as much on selling
*more* to improve our long-term postition as we seem to spend on trying
to sell less at higher margins, which is a long-term self-defeating strategy
(as Apple is showing all too well.)

FWIW - I have to agree quite strongly with the BI analogy (that was
the first thing that came to mind for me too...)

>    Do we sell PowerStorm card?  If so, why can't clone vendor buy it and
>    put it in the system?

As I said previously, no, we're not in the commodity graphics business, so
we don't want to sell the card(s) in the open market.  However, I suspect that
the cards are orderable as "spares", and that this is how the clone vendors
obtain them;  the NT drivers for them are then easily hijacked.
   
>    It would be dream come true if thousands of clone vendors were to sell
>    VMS or Digital UNIX systems to millions of customers!
>    
>    - Vikas

True, but only if they paid us the licensing fees.

Again, if they are orderable as "spares", this should *not* be a problem
as long as the costs are accurately reflected (including all driver
development costs and overhead.)

We may not "want" to be in the "commodity graphics business", but Alphas
are not (yet, anyway) "commodity" platforms.  From a corporate perspective,
given the relatively limited graphics support for Alphas due to various
issues, it seems like it is "reasonable" to expect that they should be
made available, again *provided* all costs are actually accurately accounted
for.

(If Alpha doesn't succeed overall, a lot of other things are going to look
pretty irrelevant...)

    
   > the cards are orderable as "spares", and that this is how the clone
   > vendors obtain them;  the NT drivers for them are then easily hijacked.
    
    Even if they can order it as spares, it should still reflect the true
    cost.  Why you worry about hijacking the drivers?  Most of the
    companies has driver on the WEB page for customer to down load for
    upgrade.
    
    The mentality is still in the old VMS age, you tried to guard VMS
    as the most profitable OS and eventually lose the OS market.
    

    re .16: Process warning: buying as SPARES doesn't (shouldn't?) reflect
    software costs. It's assumed that it's the hardware that's broken and
    that a new licence is not needed, just new hardware. Buying as
    "add-ons" is different. A licence may be needed, e.g. as *used to* be
    the case if you added an OpenGL-capable card to an NT system.
    
    Now returning to your scheduled discussion... and by the way, I think
    this sounds like BI too, but ...
    
    regards
    john

>  <<< Note 5205.18 by BBPBV1::WALLACE "john wallace @ bbp. +44 860 675093" >>>
>                 -< spares => replacement HARDWARE cost only >-
>
>    re .16: Process warning: buying as SPARES doesn't (shouldn't?) reflect
>    software costs. 

    Have you bought spare parts for your car lately?
    My experience you pay DEARLY for spares.

    We WANT Alpha clone vendors to be successful (no?).  So lets
    sell these things openly and charge what they're worth.  Haven't
    we learned YET that this protectionism just does not work?

    I cant believe anyone is Digital can use this justification with a
    straight face anymore.
    bob

Sigh.  I don't really want to continue down this rathole, but...

> BUT, if they buy the graphics hardware, that specific purchase should
> accurately reflect the true costs/profit/etc. involved (including all
> appropriate overhead) right?  That should also be true with any software
> costs (e.g. VMS on a "clone".)  If this isn't true, then why are we
> effectively subsidizing the prices of these options from other places?

If we were in the commodity graphics business, and intended to sell these
cards stand-alone, then yes, the price that we charged for them would reflect
the true costs/profit involved, both for the hardware and the software driver
development effort.  But that's not what we're talking about here.  
 
> I find it - as ususual - relatively absurd that (once again) we appear
> to be trying to invent engineering solutions to pricing structure problems
> (i.e. spending real money trying to figure out how to sell fewer units
> of something.)  I really wish we would concentrate as much on selling
> *more* to improve our long-term postition as we seem to spend on trying
> to sell less at higher margins, which is a long-term self-defeating strategy
> (as Apple is showing all too well.)

I can appreciate your position here.  In this particular case, I don't believe
that this is what is going on.  

> FWIW - I have to agree quite strongly with the BI analogy (that was
> the first thing that came to mind for me too...)

I still don't see the analogy.  

>    We WANT Alpha clone vendors to be successful (no?).  So lets
>    sell these things openly and charge what they're worth.  Haven't
>    we learned YET that this protectionism just does not work?

Do I, as a good corporate citizen, want Alpha clones to succeed?  Of course.
Increasing Alpha volumes can only be good for us overall.

Do I, as a workstation employee, want to see our hard work GIVEN AWAY to someone
else to make money?  Hell no.  And if we can make more money for Digital by
only selling our internally developed graphics on our workstations, then let's do it.

    >>>Have you bought spare parts for your car lately?
    >>>My experience you pay DEARLY for spares.
    
    	Sorry, I can't resist this one, but amazingly it is relevent.
    
    	If you purchase a "proprietary architecture" when you buy that
    new/other car, you will find the price for the spares valued
    accordingly.  This is one of the reasons I don't own a BMW/Lexus/Nissan/
    anything British (sorry 'bout that for you folks on the other side of
    the pond)/etc.  When there is *one* source for the parts, the price has
    no bearing in reality.  I quote dealer pricing for my wife's $75 clutch 
    disk for a Nissan, the $45 brake pads for a Suzuki Quadrunner (only
    place to get 'em) and contrast that to $17 for brake pads for my Plymouth 
    Voyager down at Pep Boys.  The latter is pretty much what they are
    actually worth, the others more like "you want 'em, we're the only
    ones with 'em so pay up or shut up".
    
    	Leo (who still enjoys working on vehicles, objects to some of the
    		pricing)
    
    

>I still don't see the analogy.  
    
    Perhaps another analogy will do.  We sell Alpha chips to clone
    manufacturers to run NT.  Your argument says we shouldn't because doing
    so doesn't force a customer to buy our own workstations to get the high
    end performance that Alpha provides.
    
    If it's good for the chips isn't good for the components?

    .3 asks if we didn't just buy a 3D graphics company (Megatek) that
    competes directly with the APS graphics design group.
    
    The answer is that they don't compete at all with the APS graphics
    design group.  They design graphics options with significantly
    higher visual quality (and price) than the mid-range options that
    we are designing on the east coast.  They are also actively 
    encouraging migration of their higher visual quality technology 
    into the lower cost graphics options that we are designing out here.
    
    Also (and this is very important), Digital did not repeat the
    mistakes of the past by having graphics design done in widely
    disparate parts of the company.  Brian Croxon (my boss' boss)
    is in charge of both the east coast graphics designers and the
    west coast graphics designers.  So it's pretty certain that we're
    going to keep working cooperatively.  
    
    All this is unrelated to the main issue, of whether we should or
    should not allow clone vendors to sell our internally-designed
    graphics options.  There is a clear stratgey to not *market* our
    internally designed graphics on non-Digital machines, and I think
    that makes a lot of sense.  Whether to spend money and software
    development effort to *prevent* the Alpha-clones from using our
    internally designed graphics is a murkier question, IMO.
    
    	Regards,
    	Larry Seiler
    

    May be that's NOT a good analogy, because we *allegedelly* sell chips
    to clone-vendors at much cheaper price than we sell to other internal
    Digital customers.

Ok, so the spares should have two prices:  with and without a tradein. 
The without tradein price should reflect the licensing cost.   That makes
sure we get our money and the clone vendors can sell whatever they want.

(Yes, I know this means you have to deal with paperwork on returned boards
but I'll bet that mechanism already exists in this company).

    The Standard price listed for the option PBXGI-AA  POWERSTORM 4D40T 3D
    graphics is 2,495.00 which certainly is not what one would call a
    "commodity" card.
    
    - Vikas

    There are three major factors behind this strategy:
    
    First, the Workstation group decided (two years ago!) to make graphics
    a core competency and one of (if not the) major differentiator for
    Digital workstations and a competitive advantage.  The workstation group 
    has made, and is continuing to make, major investments in graphics.  
    Today, over half of the engineering investment in workstations is in 
    graphics.
    
    We are following a multi-prong strategy for graphics product development,
    including product buy-outs, partnerships, internal development and even
    acquisitions of other companies.
    
    Second, we take support and product quality seriously.  Our most recent
    product, the PowerStorm 4DT family, was developed in conjunction with
    key software partners, and optimized and tested with key applications
    from the very beginning.  As a result, the 4DT production introduction
    and rollout was the cleanest graphics introduction we have seen in
    several years -- while achieving industry leading performance (we are
    beating the best SGI workstation graphics, the Maximum Impact, in 3 out 
    of 5 of the industry standard OpenGL benchmarks -- at 1/2 the cost).
    
    We have tested and qualified the 4DT's on each of the platforms they
    are supported on.  Typically, we have found some bugs with each new
    platform we support.  We have no access to the clone platforms, so we
    have no way to test or qualify on the clones, and no way to support the
    clones.  Experience has shown us that attempting to provide customer
    support on clones is a nightmare.
    
    Third, the workstation group does not have a market and channel plan
    for supporting Alpha clones.  Experience clearly shows that success in
    any channel requires well developed, well executed strategy and plans!
    
    Also note that the PowerStorm graphics support both Intel and Alpha
    platforms, so any channel strategy would need to address both Alpha
    clones and Intel clones.

    Re: 27
    
    The list price on a PowerStorm 4D40T is $3,995 (not $2,495).
    
    The 4D50T is $7,995 and the 4D60T is $11,995.

>    May be that's NOT a good analogy, because we *allegedelly* sell chips
>    to clone-vendors at much cheaper price than we sell to other internal
>    Digital customers.

Where do you people hear these things?

    RE: .28
    
    	Russ, well spoken!
    
    >>The workstation group has made, and is continuing to make, 
    >>major investments in graphics. 
    
    This tends to say:
    	Workstation group + their limited R&D = Great Graphics!
    
    	Semiconductor group + their limited R&D = Great Processors
    
    
    I'm sure I could build a really great equation from all this, but would
    it not make sense to move Semi R&D$$ to WS R&D$$ and ensure that both
    get sold and help towards DIGITAL$$?
    
    I understand the "ownership" of the effort done, most certainly under
    some of the constraints (hmm, did someone mention politics?) of the past 
    year or so, but your product is likely to be leadership for only a year 
    or two, sitting on it provides no measureable anything aside from being
    able to say that you sat on it for a year.
    
     But I digress, provide the means to sell it at the needed return.  If
    DIGITAL Semi has to OEM it with the chips, make that possible.  When
    you get a few hundred thousand installations depending on it and you're
    funded after the year 2001 remember this note.
    
    	Leo
    

    re .30,
    
    	I'm wondering the same thing myself, since I doubt its true.
    Pricing is usually based off of quantity purchased. So theoretically
    a clone-vendor could order a large amount and get a discount. Which
    would make the price cheaper then internally. Of course if internal
    bought the same quantity they would get the same price.
    
    Ron

    Who is gonna care in two years?? This is one more "protect the digital
    difference" bad decision. It didn't work in the 80's and it wont work
    now. I do however get a kick out of the DEC corporate defenses on such
    foolishness. My mom used to say "go ahead, cut off your nose to spite
    your face". I am glad to hear this petty decision will protect our
    dwindling workstation business, who would want to gain market share
    with competition after all

Hmmm, that's not what I thought I was saying...

I think it's fine to sell Alpha chips to clone vendors. It increases
the volume of Alpha systems in the world, and maybe creates some
pull for Alpha native applications.

What I'm saying is that we have a right to produce value add options,
like 3D graphics accelerators, that we sell only on our systems.
If someone else wants to build a 3D graphics accelerator (and write
the drivers for it) and sell it on Alpha systems, great!  Nothing
is being done to prevent that kind of activity.  That's why this is
NOTHING like the BI situation, where we actively discouraged other
people from building add-ons for our systems.  I don't know how to
say it more clearly than that.

Now, I agree that we could adopt a different model, and sell the 
3D graphics option as a component to clone vendors.  Its possible that
we could make more money through high volume sales of such an option;
I ask you to consider whether the other position is also possible, that
there is more money to be made restricting that option to our own
workstations only.  I'm sure a lively debate will ensue... :)

re: .31

>    I understand the "ownership" of the effort done, most certainly under
>    some of the constraints (hmm, did someone mention politics?) of the past 
>    year or so, but your product is likely to be leadership for only a year 
>    or two, sitting on it provides no measureable anything aside from being
>    able to say that you sat on it for a year.

Who said we're sitting on our laurels?  3D graphics is an area of *continuing*
R&D investment.  We have leadership now, and we're working to hold onto it.

    I don't know the right answer on this one.
    
    But I *do* think that Todd Dutton deserves thanks for answering
    honestly, engaging in the debate, and being clear about his
    perspective.  Thanks, Todd.
    
    One point that may not have been clearly heard in some of the previous
    responses - why not charge the clone makers a fair price, rather than
    closing the business off altogether?  Selling at a fair price gives
    both the appearance and the reality of allowing more channels; refusing
    to sell gives at least the APPEARANCE of predatory, proprietary
    see-figure-one business practices.
    
    	/john

    	Let me echo another thanks to Todd.
    
    	Let's get the right return.  If this is (and you say it is) one of
    those kick-** that can do it,  get the word out and get the return on
    your efforts.  Two years of engineering with a stellar result should 
    be worked to the fullest.  
    
    	Again, thanks, Todd
    

    
    The BI analogy doesn't compare.  We attempted to "license" the
    rights to build BI devices.  Pay a license fee, and buy the BI
    chip from us.  BTW, we have this nice bare board with the chip
    already on it that you can buy.  It was clearly an attempt to
    find a way to recover our R&D for the cost of developing a new
    bus architecture.
    
    In this case, anyone can design a PCI card, write the drivers
    for it, and sell it to Alpha clone makers.  Business is
    business, we have the right to sell a product only through our
    own channels for systems, if the SBU deems that Digital's
    systems can compete on value, instead of price, by doing so.
    
    Funny how so many can Monday-morning quarterback a business
    decision, when they don't hold the profit and loss responsibility
    for that decision.
    
    Dennis
    

As manager of the software team that produces the great NT drivers for our
Powerstorm cards,  I forwarded the base note to the Workstations group
VP, Philipe Ribeyre.  He asked me to post this response:

		
"We are making a LARGE (tens of million $) investment into graphics.  We will 
get a return on our investment only if we get revenue for graphics boards and
systems.  Graphics are out key differentiator and we intend to keep it
this way." [Philip Ribeyre, VP Workstations]

Garry Poegel 
Manager NT 3D Graphics

  >  Funny how so many can Monday-morning quarterback a business
   >     decision, when they don't hold the profit and loss responsibility
    >    for that decision.
    
    The individuals making the business decisions for this company have
    educated me fully in the category of "loss". Could someone point out
    how these same people have addressed the concept of "profit" in the
    last several years????

    
    Much as I have little love for the workstation group, and their
    "strategy", I have to agree with them on this one.  As they believe
    that the 3D graphics that they develop provide their differentiation,
    and because their investment in 3D graphics isn't reflected by the
    pricing/volume of options, they have every right to require that the
    option only works on their workstations.
    
    Of course, it's my opinion that they are on a course to oblivion.  They
    have hitched their wagon firmly to NT, and their strategy, while it was
    good for the UNIX and VMS markets will fail in the NT market.  They
    just can't compete on price, and they underrate the ability of graphics
    options producers to provide competetive performance in OpenGL/NT 
    graphics, but at significantly lower prices.
    
    The VMS graphics has been shrunk to the TGA2 chip (oh, and we will sell
    the PixelVision until we run out - but not on any EV past the EV56). 
    There are *no* new VMS graphics planned *ever*, and the commodity
    graphics were taken over, and then dumped (the last one, the S3 card,
    went to EOL on workstations, no replacement).  And UNIX is probably 2
    years behind VMS on the same path of being marginalized.  We won't sell
    VMS on the newer workstations because the VMS customers are being used
    to soak up the inventory that they can't give away to NT customers.
    
    So, given the NT focus, they would be better off to shut down the
    workstation group, and move the work to Hudson where they would design
    and build systems and graphics using the commodity model.  Cut to the
    chase.
    
    
    

    
    FWIW .. I agree with .41
    
    - establishing marketshare is what it is all about these days. This
    means getting everyone on your side ..dealers and Customers.
    - sell graphics cards to dealers at a price that ensures we will make
    a good profit on it, otherwise, dealers will find another graphics 
    vendor and defacto support for Alpha graphics market will become
    something else than the PowerStorm ..
    
    Boy, does this thread's discussion sound like someone trying to justify 
    BETA tape technology over VHS ?
    
    Key is marketshare.
    
    / Kerry
    
    

    
    .42  Thanks, but it's not BETA vs VHS.  It's "Shaws Brand" (a local
    supermarket) versus the name brand.  We believe that we have the name
    brand, and it's differentiation is our own graphics.  But another part
    of the company design and sell all the parts that are used to create
    the store brand.
    
    The real question is how many people will pay a premium, and how large
    a premium, for a DEC workstation.  Historically, the VMS folks do, but
    we don't care about them anymore, the UNIX folks are more likely to
    consider buying the non-DEC workstation, but by-and-large, those
    customers are just as likely to go to HP or Sun.  Our focus is on the
    NT customer, and here you have a very, very, price sensitive customer,
    with no loyalty to a hardware vendor.
    
    Hudson has layed out the plans for a $2500 NT workstation.  Will the
    workstation group be able to compete on price, performance, and
    graphics differentiation?
    

The issue is not _just_ market share. It's also to what degree you can compete 
with your own customers and still have them as customers. 

Holding back key components from our clone-building partners in order to 
differentiate our offering is a sure way to reduce their loyalty. I understand 
that the Workstations group isn't chartered to increase Alpha volumes, but if 
Alpha doesn't succeed, we won't have a workstations group (or any other group, 
for that matter).

Logic would suggest that we either spin off Semiconductor or reorient our 
workstation efforts.

But then again, when did logic ever prevail around here?

\dave

    A couple of key points:
    
    First, the workstation group is focusing on graphics as the competetive
    advantage and competitive differentiation -- and this graphics focus is
    on the capabilities of the total system, not just the graphics board. 
    Nothing in this strategy precludes the use of alternative technology,
    including (potentially) Alpha motherboards from Hudson -- or even Intel
    motherboards from Intel!  It all depends on the platform strategy, and
    we can make specific sourcing decisions on a case by case basis.
    
    We are sourcing graphics hardware both internally and externally. 
    Currently, a lot of our value add is in the Alpha drivers -- with the
    high-end 3D graphics board, we are getting over twice the graphics
    performance out of it that the board vendor is!  In the PowerStorm
    family, we have PowerStorm 3D10 (buyout), PowerStorm 3D30 (build),
    PowerStorm 4D20 (build), PowerStorm 4D40T/4D50T/4D60T (buyout).  Future
    products include both build and buyout -- we are staying on the leading
    edge of exploiting new graphics technologies as they become available.
    
    Obviously, a key component of this strategy is ensuring that
    applications work across the different generations of graphics devices!
    
    Second, offering PowerStorm graphics to clone vendors has cost. 
    Qualifying graphics on each system costs (including any software
    changes required).  EXPERIENCE has shown us that support on clones is
    extremely expensive; we aren't involved until things have blown up and
    a customer situation is critical, we don't know the whole
    configuration, and we haven't been involved in configuring the systems.
    
    SUCCESSFULLY offering graphics in the clone market -- either Alpha or
    Intel -- requires a different business model.  It requires a channels
    strategy, development, support and qualification mechanisms,
    promotional activities and programs.  This is the familiar OEM market
    -- it can be done, and Digital has done it successfully in various
    areas.  It is NOT, however, simply allowing people to randomly grab
    options out of our catalog and stick them in an arbitrary box.

    re: .41
    
    I also agree...when the NT takes all strategy was developed and rolled
    out in the WS segment, we outlined our concerns, and outlined our
    expected results, which have come to pass with flying colors.  Since we
    are now presiding over the demise of the WS business, the question is
    should we cut our losses as suggested, or try to rebuild the Digital
    UNIX and OVMS WS business and hang on for a few more years?
    
    Mark
    

re: .41

>    Of course, it's my opinion that they are on a course to oblivion.  They
>    have hitched their wagon firmly to NT, and their strategy, while it was
>    good for the UNIX and VMS markets will fail in the NT market.  They
>    just can't compete on price, and they underrate the ability of graphics
>    options producers to provide competetive performance in OpenGL/NT 
>    graphics, but at significantly lower prices.

We'll have to agree to disagree on this one, Fred.  I think it's possible
for us to produce Intel & Alpha workstations that can compete, when you
start comparing feature-to-feature.  As far as other graphics options
vendors are concerned, we are *very* aware of their capabilities. As Russ
Doty noted previously, we only choose to build our own 3D graphics when
we see clear advantadges to doing so -- the graphics strategy is biased
towards buying out options from third parties in areas where we can't
compete, such as "low-end" options like Matrox.  

But if you've seen any of the output from the graphics options produced
by Megatek, you know that the quality of the images produced by our
high end options far exceeds anything that commodity 3D vendors are
able to do today.  In terms of full scene antialiasing, transparency,
shadows, etc, the difference in image quality is night and day.  The
difference in performance levels is also night and day.  These are displays
equivalent to what SGI offers in their Infinite Reality -- at *far* less
cost to the customer.  

>    So, given the NT focus, they would be better off to shut down the
>    workstation group, and move the work to Hudson where they would design
>    and build systems and graphics using the commodity model.  Cut to the
>    chase.

Hudson used to be in the commodity graphics business... and got out of it.
I think they're having a hard enough time delivering on their existing
commitments with their existing staff -- why do you believe they'd be
successful taking on the commodity graphics space?
 

re: .43

>    The real question is how many people will pay a premium, and how large
>    a premium, for a DEC workstation.  

Yes, that *is* the $64 question! :) 

Even in the PC space, which is the epitomy of cost-sensitive buyers, there
are people that will pay a premium for what they perceive to be a higher
quality product.

When you start talking about "personal workstations" with customers, one
thing you hear is that they want dependable, reliable machines.  They're
not interested in PCs that they have to reboot every day or so in order
to resolve hangs.  They're willing to pay a premium for the kind of
quality and stability for which Digital has a reputation.  

Granted, these are a particular set of market niches, but that's what 
"personal workstations" are aimed at -- general purpose business users
*are* going to buy traditional PCs, and I'm sure the PCBU is happy to 
sell them!   

>    Historically, the VMS folks do, but
>    we don't care about them anymore, the UNIX folks are more likely to
>    consider buying the non-DEC workstation, but by-and-large, those
>    customers are just as likely to go to HP or Sun.  Our focus is on the
>    NT customer, and here you have a very, very, price sensitive customer,
>    with no loyalty to a hardware vendor.
>   
>    Hudson has layed out the plans for a $2500 NT workstation.  Will the
>    workstation group be able to compete on price, performance, and
>    graphics differentiation?

Well, time will tell.  Obviously, I believe yes; clearly, you have your doubts.

    
    Yup.  It is the question.  And only time will tell.  Early returns, if
    the press results of the decline in WS sales are not encouraging.
    
    Between ever increasing performance in the Intel space, the entrenched
    perception (if not reality) leaders on the performance side (SGI), and
    the masters of the distribution channels like HP.  Can we win a niche
    in the NT market?  At the low-end pushing up are the razor thin margins
    of the PC space, dominated by Intel, binary compatability, shrink
    wrapped software, and hardware vendor independence.  At the high end is
    the application and performance driven market, best characterized by
    SGI.  Pushing down, still Unix dominated, but which may start to fall
    apart with Merced, and slowly shifting to the PC/NT space, because the
    cost is so much lower.
    
    I'm not happy about the workstation strategy, because it trashes the
    O/S that I work for.  It uses it only to backfill revenue
    shortcommings.  While it may have been written off by the company (the
    OVMS workstation/desktop/technical user), these systems have been
    traditionally used by ISVs, and customers, as small servers, and
    development machines.  They are hostages, are getting screwed, and
    pissed off.  Which spills over to the perception of the server space
    (customers don't differentiate between our internal organizational
    structures).
    
    I give the workstation strategy low odds of working.  But it's your
    bet, even though it negatively impacts the viability of my O/S.
    
    BTW - the TGA/TGA2 could have/should have been a killer in the
    commodity space... except for the fact that it required high-cost VRAM,
    and we didn't market it for the Intel space like say, a #9 video card.
    The commodity graphics chip they did, IMHO, made some poor choices
    (like the lack of a plane mask), and we were not prepared to build and
    market the card directly... or even use it internally.  That's too bad. 
    And I'm sorry that Hudson is getting out of the game.  Especially since
    I have no idea what graphics will be available for OVMS when TGA2 goes
    away... I hear NEON is the successor, but it's not being done for VMS.
    

re: .49
>    
>    Yup.  It is the question.  And only time will tell.  Early returns, if
>    the press results of the decline in WS sales are not encouraging.

As I previously noted, there's plenty of opportunity for fingerpointing on
that issue.  And you see some of it going on in other notes in this file,
so I'm not interested in going there.  
    
>    Between ever increasing performance in the Intel space, the entrenched
>    perception (if not reality) leaders on the performance side (SGI), and
>    the masters of the distribution channels like HP.  Can we win a niche
>    in the NT market?  At the low-end pushing up are the razor thin margins
>    of the PC space, dominated by Intel, binary compatability, shrink
>    wrapped software, and hardware vendor independence.  At the high end is
>    the application and performance driven market, best characterized by
>    SGI.  Pushing down, still Unix dominated, but which may start to fall
>    apart with Merced, and slowly shifting to the PC/NT space, because the
>    cost is so much lower.

The high-end performance market, as characterized by SGI, is indeed migrating
to NT.  In fact, it is doing so at a surprisingly fast rate.  It is not slowly
shifting to PC/NT, the ISVs are running as fast as they can towards NT!  Given
Merced's delays, Digital has a tremendous opportunity to capitalize on this
transition, to be associated with high performance graphics on NT the way SGI
is associated with it on UNIX.
 
>    I'm not happy about the workstation strategy, because it trashes the
>    O/S that I work for.  It uses it only to backfill revenue
>    shortcommings.  While it may have been written off by the company (the
>    OVMS workstation/desktop/technical user), these systems have been
>    traditionally used by ISVs, and customers, as small servers, and
>    development machines.  They are hostages, are getting screwed, and
>    pissed off.  Which spills over to the perception of the server space
>    (customers don't differentiate between our internal organizational
>    structures).

If we're failing as a company to deliver essentially "entry-level" servers
to the OVMS ISVs and customers, it's not clear to me that this is a failure
of the *workstation* strategy...  
   
>    BTW - the TGA/TGA2 could have/should have been a killer in the
>    commodity space... except for the fact that it required high-cost VRAM,
>    and we didn't market it for the Intel space like say, a #9 video card.
>    The commodity graphics chip they did, IMHO, made some poor choices
>    (like the lack of a plane mask), and we were not prepared to build and
>    market the card directly... or even use it internally.  That's too bad. 
>    And I'm sorry that Hudson is getting out of the game.  Especially since
>    I have no idea what graphics will be available for OVMS when TGA2 goes
>    away... I hear NEON is the successor, but it's not being done for VMS.

While Hudson owned the TGA design, the graphics group here in workstations 
owns the credit/blame for TGA2 (which is still one of the fastest, if not the
fastest, 2D cards in the world).  As far as I know, TGA2 is not going away
anytime soon, at least in the OVMS world.

I can't comment here on products that may or may not be under development :),
but I can say that I haven't seen any requests for support for any new graphics
devices on OVMS.  Of course, they may not make it all the way down the command
chain to the trenches I live in... :)
 

Just for the record...

The NTSI machine performed better than most of the other Alphas in the PRO/E
magazine test because THEY CHEATED. The NTSI machine had 512 Mbytes of mem vs.
256 Mbytes for every other machine in the test. The text of the article explains
that the memory could not be depopulated because it was built from a minimum
number of large modules, so the ran the tests with the large memory.

--RS

P.S. The PWS 500a ran faster with 256 Mbytes than the NTSI machine did with 512
Mbytes!

    They cheated...
    
    It wasn't fair...
    
    The benchmark was rigged...
    
    where's the appeal processes if this type of travesty actually occured
    outside the bounds of propriaty...
    
    Where's the press release repudiating the benchmark process...
    
    Don't tell us.. make a stink about it to the bodies responsible
    for the auditing...
    
    JMHO
    
    John Wisniewski

>From http://www.imc.das.dec.com:9015/announce/bench97.htm
>
>	NTSI's Alpha 500 had the best price/performance of all the reviewed
>	machines at a cost of $13,500, which included our own DIGITAL 
>	PowerStorm 4D40T graphics as well as 512MB of RAM. Work is currently
>	underway to ensure that the use of our PowerStorm 4DT series graphics
>	on Alpha clones is discouraged legally or better yet, disabled.

I assume the price of $13,500 accurately reflected the additional
money?  If so, and they still won price/performance....what
was our price?

Also....are they going to be able to provide our graphics
in volume?  Sounds like they may have done a great job
creating demand that they wont be able to fill.
bob

This mail came earlier this week.  Since the author desires for the news
to be spread, I am interpreting it as being appropriate to post in this
notesfile.  Moderators feel free to take action, if you disagree.

I wish they had referenced their Dataquest source more accurately, so
we could interpret the numbers quoted better (a Web reference would have
been nice).  Based on my perspective on MCAD workstation sales in my
field territory, I have a hard time believing the claims.

Ram


--Boundary (ID TIdo3seNRaoQhCBpDHbnsA)
Content-type: MESSAGE/RFC822

Date: Wed, 26 Mar 1997 00:51:57 GMT
From: Kenneth Goldner <[email protected]>
Subject: FW: Dataquest Says Digital is #1 in Mechanical Unit Volume for 1996
To: {TOO MANY PEOPLE}
Message-id:
 <c=US%a=_%p=Digital%[email protected]>
MIME-version: 1.0
Content-type: MULTIPART/MIXED; BOUNDARY="Boundary (ID /2XmR+zQbyowTq95TKx0HA)"
Delivery-date: Wed, 26 Mar 1997 00:55:00 GMT
Posting-date: Wed, 26 Mar 1997 00:49:40 GMT
Importance: normal
A1-type: MAIL


--Boundary (ID /2XmR+zQbyowTq95TKx0HA)
Content-type: TEXT/PLAIN; CHARSET=US-ASCII

Great News!!  Let's spread the news!
Ken
>----------
>From: 	Ty Rabe
>Sent: 	Monday, March 24, 1997 5:23 PM
>To: 	Dan Innes
>Cc: 	Philippe Ribeyre; Ronald Locklin; 'Ron Beck'; 'Mamoru Nakano'; Kenneth
>Goldner
>Subject: 	Dataquest Says Digital is #1 in Mechanical Unit Volume for 1996
>
>Dan, could you publish these results in the NewsFlash?  They arrived today:
>
>Dataquest recently identified Digital as the number 1 computer systems vendor
> to the worldwide Mechanical CAD/CAM/CAE market based on calendar year 1996
>unit shipments.  Here is a summary of their report:
>
>Category		Result		Share	 	% Unit Growth
>
>Total CPU shipments	68,715		  18%		         24%
>
>UNIX shipments		18,364		  13%		         17%
>
>NT shipments		 4,155		  29%		       106%
>
>Other Shipments	 3,340		  NA		         11%
>
>Digital was ranked #1 in all categories but UNIX shipments, where we were #5.
> However, even in the UNIX category, our unit shipments grew 17% and our non
>PC unit shipments grew 25%, according to Dataquest.
>
>SOURCE:  Mechanical CAD/CAM/CAE Worldwide Market Share, Market 	    	
>Statistics, Dataquest, March 10, 1997 	
>

--Boundary (ID /2XmR+zQbyowTq95TKx0HA)
Content-type: MESSAGE/RFC822

Date: Wed, 26 Mar 1997 00:55:16 GMT
MIME-version: 1.0
Content-type: TEXT/PLAIN; CHARSET=US-ASCII
Importance: normal
A1-type: DOCUMENT

RFC-822-headers:
Received: from pkohub1.athena.pko.dec.com by gcpmpk.pko.dec.com
 (PMDF V5.1-6 #21004) with SMTP id <[email protected]> for
 [email protected]; Tue, 25 Mar 1997 19:49:46 EST
Received: by pkohub1.athena.pko.dec.com with SMTP
 (Microsoft Exchange Server Internet Mail Connector Version 4.0.994.63)
 id <[email protected]>; Tue,
 25 Mar 1997 19:52:28 -0500
X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.994.63

--Boundary (ID /2XmR+zQbyowTq95TKx0HA)--

--Boundary (ID TIdo3seNRaoQhCBpDHbnsA)--

    
    Boy - last posting is what is really needed - can we get official
    pointer on Net where this might be posted ? Perhaps we could get the
    powers that be to make this a flash on our DIGITAL home page ?
    
    On a related topic:
    
    Perhaps someone could comment and/or provide me with some info on how
    to respond to this posting on CompuServe ? Does anyone know how
    this referenced graphics card compares to the PowerStorm card ?
    
    "Evans & Sutherland has released a chipset called Real Image Technology
    which addresses the mainstream graphics market for the first time!
    
    It has been announced that the chipset is to be used in Diamond
    Multimedia's Fire GL 4000 board and in Excel Graphic's Ecclipse.
    
    As an Alpha user I hope that this chipset finds its way onto an
    Alpha-compatible graphics adapter soon.  We use Alphas because of 
    their superior performance and we need E&S graphics because of the 
    same reason!
    
    I know we Alpha users are somewhat smaller in size to the Intel users
    but if anyone knows how to effectively lobby board manufacturers for 
    this Alpha-compatible E&S graphics adapter product please let me know."
    
    

re: .48

> When you start talking about "personal workstations" with customers, one
> thing you hear is that they want dependable, reliable machines.  They're
> not interested in PCs that they have to reboot every day or so in order
> to resolve hangs.  They're willing to pay a premium for the kind of
> quality and stability for which Digital has a reputation.  

SLOAN is a workstation running VMS 5.5-2.  It typically runs for about
3-6 months before a reboot is required.  More often than not, it's
because of a power hit.  My pc running Windows 95 would be lucky
to run for 2 days without a reboot.

My workstations now is an Intel PC running WNT V4.0. From 3 months of usage,
it appears to be as stable as VMS.  Seems like the guys who designed
WNT had some experience with VMS. :-)