Conference 7.286::digital

    Doesn't matter what they're trying to do in Note 4568 e.g., The New
    Partnership we'll be out of business before we can establish a new
    culture.  
    There's two sides to every story.  Anyone know ours (Ok, is Dave still
    around)?  Anyone involved in the request care to comment?
    This type of Marketing really establishes Customer Confidence. 
    Reproductions of this article will end up at every Customer that both
    we and HP are involved with.  Once again we missed the opportunity to
    allow a magazine to compare our systems against the competition.  I
    can only speculate that our systems must be performing poorly as
    compared to the competition or we'd be pounding down the doors or else
    we're fools.

    	Regards,

      
    
    re .0
    
    the only thing that amazes me about the article is that you're
    surprised by it.....
    
    6 years ago we all would have been surprised..shockes, dismayed, 
    mad as hell. 
    Today? ..... so, yeah????
    

    I have participated in several successful magazine benchmarks - see
    cover story of UNIX Review October 1995 for an example.  Since funding
    to cover benchmarks is generally VERY hard to find (I can cite at least
    5 organizations who declined to fund such activities in the last year),
    if there is anyone out there who is still doing magazine benchmarks
    please contact me.  I will be happy to share with you my checklist of
    what to do to improve the probability of success of a magazine
    benchmark.
    
    Hint #1: Doing a high-quality job on the preparation for a magazine
    benchmark requires more technical effort than what managers are usually
    prepared to allocate.  It takes time, $, machines.
    
    Hint #2: the higher quality the job that is done, the less visible it
    is.  A perfectly prepared benchmark is a benchmark that has NO
    SURPRISES, that is, one for which all the effort becomes invisible.  
    
    Hint #3: what is the return to Digital from an article such as the UNIX
    Review cover story (which cost about $50K in time & machines to
    prepare) vs. the return to Digital from $50K spent on advertising?
    
    I would be delighted to address the staff meeting of any appropriate
    group regarding either the funding and planning issues referenced above
    or the technical issues.
    
    /John Henning
     CSD Performance Group
     Digital Equipment Corporation
     henning@zko.dec.com
     Veteran of benchmarks: Advanced Systems, Federal Computer Week, Sun
                            Expert, Digital News & Review, Open Systems,
                            DEC Professional, UNIX Review
    
    PS In Dave's defense, not all magazines are equally easy to work with.
       Some are very hard to work with, and this may be one of them.  Also,
       I don't believe it was his decision to decline to fund the technical
       efforts that are urged above.

    sad commentary, .2, but true indeed...
    
    best unkept secret is about a dual processor NT machine of a
    competitor's that not only runs circles around the same config'd
    Prioris, it's about 1/3 less list price...won't be a secret much longer

    .3
    
    the biggest complaint I hear from Digital customers is how hard it is
    to work with Digital, get someone to return a call, email, etc...the
    tfso's were done but processes never improved where a fewer number can
    manage the existing business which remained, much less try and grow the
    business...this is this management's achilles heel and digital's
    possible demise

    I cannot dispute Steve's original memo in this note stream.  I am a
    faithful reader of a number of industry magazines, including PC Week,
    and was as disappointed as he was with the story.  
    
    But all is not completely blue.  In InfoWorld's June 17th issue, they
    ran benchmarks also on a Pentium Pro, and included other machines such
    as the Alpha XL 366, Power Mac, and an SGI Indy R5000.  While the
    Pentium Pro did well in the report, the overall leader was the Alpha,
    which took first place due to setup and installation, expandability and
    system design, as well as speed.  (Surprisingly, the SGI box came in
    dead last, since it excelled in no categories, and was the most
    expensive machine in the study.)  
    
    But we only wound up with a Satisfactory score in the overall area
    because we failed miserably in the Technical Support area, a
    traditional strongpoint of our's in years past.  This is what InfoWorld
    had to say:
    
    "Although the Alpha is a very new product, the fact that two of the
    people we spoke to at Digital hadn't heard of it wasn't very
    reassuring.  After many calls, we got the impression that the whole
    support process was mired in bureaucracy.  We got an answer to one of
    the problems, but it took way too much help from us.  On another
    problem, we had to make two calls, and after speaking with four people,
    we were told they would get back to us.  We're still waiting."
    
    To receive a 0 score for technical support does not make new customers
    (who will decide the ultimate fate of Alpha, after the
    potential base has been saturated) feel comfortable with Digital. 
    Whether it is because TFSO's have cut too deep into the organization or
    whether it is something else, this area needs to be fixed.
    
    Keep your heads up, though - it's not all gloomy out there.
    
    Chuck
    

    re .3 and David Berlink (Reality Check)
    I think you are mistaken.  I remember, awhile back, Digital announcing 
    our New Business Practices.  We don't do a lot of that stuff any more. 
    We're responsive to the Customers these days in the NEW DIGITAL.
    
    When we announced our New Business Practices I heard a Customer jokely
    say:
    "the problem with the New Business Practice is that individual on
     the other side of the desk hasn't changed, therefore, nothing has
     changed at Digital."  
    
    I guess with the downsizing there's less of us to do it.
    
    We worked in Base Product Marketing for Workstations at the time of the
    MIPS-based workstations.  We dealt with magazine like Personal
    Workstation, I don't believe that it's around anymore, and afew others. 
    A number of the things that David Berlind mentioned were true.  Our
    main issue was that the magazines all wanted the LATEST-AND-GREATEST 
    and it was hard getting these seed unit from Engineering, moreover, 
    seed unit were EXTREMELY expensive as compared to a revenue stream unit.
    The products were just NOT available which was surprisely true.  And
    those were the days when we were rolling around fine.
    
    	Regards,

Absolutely amazing.

Why does Digital seem to be so inept at marketing it's products?

We somehow believe that people will understand the technical
superiority of our products just because they are technically
superior, which is not always true.

Technical superiority is only one small piece of competing.  Remember
VHS vs Beta, etc.  Repeat, ONLY A SMALL PIECE.

We sit here and complain about how HP or whoever is talking about
products that will come in the future (processors, speeds, etc) and
are not actually shipping today.  We doubt their ability to actually
deliver on their statements and somehow believe it is unfair that they
are making those statements.

They are marketing their products (even future products) and their 
brand name and gaining mind share.

We seem to treat our future direction with processors, speeds, etc.
as top secret information.  We wouldn't want a potential customer
to know how powerful we expect our systems to be a year from now
and above all we wouldn't want to give optimistic estimates,  even 
if they were identified as estimates.

Bob Palmer was talking about how there will be continued distress in 
the PC industry because there are so many vendors with virtually 
indistinguishable products.  I believe this is becoming true of larger 
systems as well.  

What do you think creates a key differentiation in the customer's mind 
and makes some companies successful in that type of environment?

It is marketing; brand recognition.

We have tremendous products that we can't seem to sell.  At least not at 
a rate that is on par with industry growth rates.

Perhaps our problems have not allowed us to appropriately invest in
marketing.  The example of not capably providing machines for 
benchmarking by magazines.  Absolutely ridiculous, but I can here the 
discussions: "It's not coming out of my budget... we need to 'control 
expenses'"; "The magazines don't like us anyways..."

    In all fairness to Digital, I gave this article to my wife to read
    (Kathleen Connors), who was a Product Manager for uPDP-11, uVAX, and
    the VAX4000 for approximately 5-7 years.  She indicated that Digital 
    had a policy of not providing systems for these request as the systems 
    being requested were recently development efforts and had either hardware 
    or firmware bugs.  We would provide the systems with the understanding 
    that they should NOT be used for benchmarks because of these issues. 
    They would go ahead and benchmark them anyways and publish the results. 
    The intent of providing the system was for the magazine to understand
    the architecture and features of the new systems.
    
    We've had 64-bit for acouple of years now and can't take advantage of
    it.  The newly announced UltraSPARC is being very well received in the
    market.  I know customers that were limping along with the old systems
    and clones e.g., Solburn and have now ugraded to the UltraSPARC.  They
    wouldn't even consider the Alpha.  They use to all be VAX customers.  When
    Sun gets a 64-bit OS and Apps and the customer's are breaking the door 
    down to get a piece of the action we're going to be sitting back and 
    wondering what the "H_LL" happened.  And we'll be saying: "BUT we've had 
    that for over 4 years (as we downsize for the last time)."

    We (PCBU) supply magazines with systems to benchmark all the time.
    90% of the time we can supply what they're looking for.  Only
    occasionally do we get caught in a product transition where
    the lead time of the magazine extends beyond the availability of
    the current product and the next generation isn't quite ready yet.
    
    The basic problem here is that Alpha can't compete with P6 servers.
    It's way overpriced for the performance, or woefully slow given
    a price target.  We should have never sent in a system for review
    to PC Week (or any Ziff pub).  Stick to Windows Magazine (they have
    two editors, Ruley amd Heller, that like Alpha, plus they have no
    internal server benchmarks), Byte (where we worked closely in
    optimizing their Bytemark benchmark for Alpha), the Unix rags (where
    we typically compete against Sun/SGI), and the Digital specific rags.
    Anything else we should stay away from: just do the Digital Semi
    shuffle (close your eyes, click your hells, and just claim Alpha
    is faster), but NEVER submit a machine for benchmarking.
    Kratz
    
    

     My friends at Compaq claim that they can usually beat us in a
    benchmark. The reallity is that benchmark results really don't matter
    to them as long as they continue to ship a truckload of stuff for every
    Alpha system we sell. 
                          

    - David Berlind
    
    ZD Labs is not only a home away from home for PC Week Labs' 
    networking  team, it is also the largest independent testing 
    facility in the world.
    
    Within its walls, a 300-node LAN lab combines with the world's 
    only cross-platform, distributable and freely available benchmarks
    --the Ziff-Davis Benchmark Operation's ServerBench and NetBench--to 
    make the consummate testbed for testing servers, NOSes and network 
    infrastructure.
    
    This lab and these benchmarks allow PC Week Labs to
    continually be the first to benchmark the latest that server and
    network operating system vendors have to offer.
    
    So, you'd think that when we were ready to perform an
    exhaustive test of multiprocessing Pentium Pro servers, Digital
    Equipment Corp. would be banging down our doors to set up
    a showdown between its Alpha and the Pentium Pro. This
    comparison would be the first opportunity for Digital to back
    up what it has been saying all along about the Alpha.
    
    We invited Digital to the party, and they came. But they didn't
    come with much. We weren't all that surprised because the
    Digital Alpha group's history of responding to PC Week Labs
    has been checkered, to say the least.
    
    A couple of years ago, when the first Alpha-based PC running
    NT was released, we requested a unit. A person whose
    identity shall remain anonymous (OK, his name was Dave)
    assured us that we'd have a unit the next day. A day later, no
    unit. Three days later, no unit.
    
    A trace was put on the shipment. A few days later, we learned
    that the unit had mysteriously appeared on one of Digital's
    shipping docks.
    
    We then were told that the unit would be in our lab in a day.
    One day later, no unit. One week later, no unit.
    
    We had a virtual repeat of this experience as we embarked on
    our supertest of super servers. This time, Digital managed to
    get us a machine, but the machine's limited expansion capability
    made it impossible to equip the unit with the four processors
    and 256M bytes of RAM mandated by our standard test
    methodology. There weren't enough slots.
    
    Questioning the machine's scalability, we settled for three
    CPUs. When we contacted Digital to say that the machine
    performed below expectations, they said they'd get back to us.
    Fully expecting them to forget us, we were pleasantly surprised
    when they called. However, the news that the machine we
    tested had a buggy motherboard was not so pleasant.
    
    This meant we'd have to invest hours in testing another
    machine, provided that Digital could furnish us with one. A new
    machine was promised to us by a certain date by someone
    who will remain anonymous. (OK, it was Dave again.)
    
    When the machine didn't show up, we called. We explained
    that if we didn't get another machine, we'd have to assume that
    a different machine didn't exist and would have to go to print
    with numbers we had.
    
    Since then, we've received no machine, and none of our many
    desperate phone calls have been returned. We can only
    assume that Digital could care less about its battle with Intel.
    

I was involved in the tail end of this fiasco and tried in my own 
little way to salvage the situation.

All I will say in public is that we deserve every bit of bad press 
that we got in this case.

If someone gets upset because I say this, too bad.  We blew it.

Ken Berkun
DECwest

re .-2

	Got to admire the honestly in .-2. Sometimes it seems that folks
	are not able to admit they did something wrong. In this case it
	seems pretty obvious that the author of .-2 didn't do anything
	wrong personally (unless he's a VP :-) )

re .-1	Why not no admit that we (Digital) are currently screwing up a lot
	of things. I personally wouldn't like to hear "Thanks, job well
	done" stuff from my boss if I had screwed up badly.

        Well, I'll be history in 60 days anyway ...


        	Tero

RE: .14

	DECwest is a site in the greater Seattle area near the 
	Microsoft campus. Primarly focus of the main group out 
	there is Windows NT work. There are other groups as well.

							mike

    Sad part: the Alpha folks have run the same benchmark, knew that
    P6 servers are faster, and PC Week got exactly what was expected.
    
    I did get a kick out of the "bad motherboard" tho.  Too bad they
    called our bluff.
    Kratz 

    What's up Kratz ?
    
    No 'We Lied' for a few weeks !
    
    
    Aw

    
    We made the tv news again last night. NBC with Tom Brokaw.
    Special report on "dumbsizing". We made the list, along with
    Apple and others.
    
    Wasn't a favorable report. It just doesn't work over the long
    run.
    
    Heck...I'm hearing all kinds of complaints lately from friends
    who do business with us. "Dave...nothing is getting done, where
    are the people that are supposed to handle the business?".
    
    
    Yep..and they also tell me...the business is going elsewhere.
    And these people support us, but can't help us, since we are
    doing to ourselves.
    
    
    
    

    Forgetting, for the moment, the fact that this matter was obviously
    handled poorly, there seems to be this ongoing internal debate over the
    relative performance of Alpha vs. P6 in the NT environment.
    
    Can someone who  has actually seen benchmarks for the two comment on
    what configurations of each is really faster and comparing approximate
    prices.  No guessing please.  Someone must really know, was this
    benchmark a fluke or not?

              Does anyone know if the benchmark was done on a 2100 or
    2100A? We did have some silicon (not Alpha) problems initially on
    this motherboard. I can state that you can have 4 cpu's and 500mb
    of memory. So unless they were only supplied 64 or 128MB memory
    boards then they wouldn't have been able to fit 500MB. There were
    some problems with the 2100A with NT related to the combo chips used
    for the keyboard and floppy. This has been addressed recently and
    may or may not have been the bug?? This bug never showed up  in
    UNIX and OpenVMS testing. FWIW if I'm not mistaken NT isn't a 64
    bit os. So it's not terribly surprising that a benchmark could be
    skewed one way or the other. Seeing as the 2100 group has been 
    disbanded you may never know who Dave is...
    
       Bob Brower motherboard designer for the 2100A

Is anybody else uncomfortable with this talk that a P6 outperforms Alpha?
If you need to go to 64-bit addressing for there to be better performance,
doesn't that imply that our market will likely narrow to VLM systems?

Jim

    There is some truth in the Alpha=VLM-only story. But there are also
    some applications where Alpha continues to outstRIP Intel. The
    pre-press market, for example, is one I know a bit about. They'll use
    all the power they can get - I/O bandwidth, memory bandwidth, whatever.
    They like multiprocessor systems that work (cf P6 cache coherency).
    [When you're producing the printing plates for a newspaper, every
    second counts, and you can't afford to be unreliable]. The rewarding
    end of this market has traditionally used whatever hot boxes it could
    get its hands on; Motorola 68k with Unix, Sun with Unix, now Alpha with
    NT.
    
    If Digital's PC channels strategy made sense the salesfolks could also
    sell lots of worthwhile PCs for the desktop and low-end server which
    currently go to Dell, GW2K etc.
    
    However, man cannot live by prepress alone. And although there may be
    other equivalent areas, are there enough ? 
    
    I still don't understand why, if P6 is worrying us so much, it isn't
    also worrying HP and Sun a lot more (apart from the minor detail that
    they have loads more Unix applications than our Unix does).
    
    regards
    john

>    I still don't understand why, if P6 is worrying us so much, it isn't
>    also worrying HP and Sun a lot more (apart from the minor detail that
>    they have loads more Unix applications than our Unix does).
    
    HP and Sun have a strategy and vision that they can effectively
    communicate to their customer.  Digital is just selling everything off
    and only has Alpha that can't beat the recently announced P6 products. 
    We continue to flounder about in the market placing looking for
    something.  The only thing we seem to be consistent in is shooting
    ourselves in the feet.

    Are we gaining any new market share for success or are we surviving due
    to the installed base that appears to be dwindling away.  We don't seem
    to have any momentum

    At the chip level, P6 does not beat Alpha.  Not then, not now, not by a
    long shot.  What happens at the system level (which, of course, is the
    only place applications can run) may be another matter.
    

When you say that Alpha is faster than P6, is that for combined integer and 
floating point operations, or integer alone?

Every benchmark I have seen shows that the Alpha is superior in floating 
point operations.  So, of course, it is an outstanding CAD platform.

However, the integer benchmarks of a 5/300 vs 200MHz P6 look pretty close. 
 And if I'm looking into purchasing an NT server for Notes or Exchange, 
then floating point speed isn't much of an issue.

Today I can buy a 5/300 Alpha system.  I can also buy a 200MHz P6.  As a 
server, which is faster, and by how much?  That is what customers are 
asking.

john

CPU performance measurements for most current and future processors is 
available at:

http://infopad.eecs.berkeley.edu/CIC/


						-Paul

    Re .27:
    
    You can also buy Alpha systems at 366MHz and 400MHz.
    
    Pentium Pro-200	8.1 SPECint95	6.0 SPECfp95
    
    Alpha 21164-366    10.6 SPECint95  14.8 SPECfp95
    
    Alpha 21164-400    12.0 SPECint95  17.0 SPECfp95
    
    The 300MHz Alpha 21164 was rated at 7.3 SPECint95 and 11.6 SPECfp95.
    

    > Alpha 21164-400    12.0 SPECint95  17.0 SPECfp95
    
    Part number?
    
    				-John

    >> Alpha 21164-400    12.0 SPECint95  17.0 SPECfp95
        
    >  Part number?
    
    21164-FB   Source:  DS online product catalog.
    
    Sorry not to be more explicit.  Part numbers aren't my business (which
    I should be doing now...)

    I thought I heard recently that our Alpha-based webservers really kick
    industry butt.  Isn't there something (probably a developing standard)
    called a "WEBstone" at which Alphas excel?
    
    Dan$HopefulYet
    

RE: .32

	You probably want to ask Marc Slaters group

	Not to jump on Kratz' bandwagon (not by a longshot), but the P6
	numbers and the Alpha numbers are close enough that the cost
	difference between the two is not justified. I'm sure there
	is plenty coming down the pike with EV6, but at what cost?

							mike

	It seems to me that it would be in our best interest to prod 
Microsoft to get NT to 64 bits QUICKLY... BEFORE intel can make the 
Pentium a true 64 bit chip. I've heard anywhere from the end of 'CY97 to 
the year 2000 for a 64 bit NT. We may already be doing this with our 
pushing VLM in the UNIX database server space. It's becoming obvious that 
the Alpha story in the NT space is loosing its credibility as the P6 and 
its successors ramp up.

	JMHO,
	Wayne
 

>	It seems to me that it would be in our best interest to prod 
> Microsoft to get NT to 64 bits QUICKLY... BEFORE intel can make the 

And why would Microsoft want to do that quickly. I'd be willing to bet that
98% of the NT market is Intel and it certainly ISN'T in Microsoft's best 
interest to make their cash cow a second class citizen.

tim m.

    my thoughts exactly
    
    and beyond that - it is quite clear that Microsoft has the upper hand
    in this partnership.
    

MicroSoft wants to be Database biggie. 

But the Oracle's VLM.....no chance unless they play the 
64bit game. Otherwise its just a toy..

    re .37 
    
    Aha, isn't this the point. My understanding of the debate is this
    
    Alpha is 64bit, fast, and in one sense expensive, expensive in that for
    some tasks you may have to buy more than you need, and you can get a
    much cheaper 32 bit box from SUN/HP etc. Now HP have come along with a
    very fast 32 bit box and are using it to rubbish alpha. HP, IBM, Sun
    and way behind us in the 64 bit race and are fighting to plug the gap
    and catch up at the same time. They say, you don't need 64 bit yet, but
    by the time you do , we will have it, and of course it will be cheaper
    than alpha. We say, first of all, the Mandy Rice-Davies quote: "They
    would say that about 64 bit and alpha wouldn't they?" They haven't got
    it, and when they get it they will have a lot of headaches which we
    have tackled. But 64 bit is the future. Slowly but surely it will take
    over from 32 bit and our lead will become more and more important. So
    by investing in Alpha you are investing in the future and getting
    yourself a medium term competitive advantage. And as time goes on you
    will notice this more and more. Oracle VLM is just the start. You will
    see more and more applications that need 64 bit as time goes on.
    
    My understanding is that the Benchmark mentioned here is for a 32 bit
    os, so the alpha advantage can't show.  Another point is that HP are
    using out of order processing to increase speed, but this is a short
    term design fix which shows they cannot make the move to 64 so quickly.
    I wouldn't claim the expertise to be able to judge these issues, but it
    would be good if those who do would speak up. There has been a tendency
    is the past few weeks here for people to lose confidence in Alpha,
    which is not good news, but the worst part is that the arguments for
    Alpha are not being made. 
    
    Okay, I accept we shot ourselves in the foot again, and we have a lot
    of organisational, mindset and marketing problems. But I do believe
    Alpha is a medium term winner, because it is 64 bit and we have it now. 
    The others haven't got it yet, and when they have it, they will realise
    what we have achieved in the last 4 years. Also with the 64-bit WNT
    movement, aren't we well placed with the affinity programme ? It seems
    to me that BP, for all his faults regarding chainsaw-style downsizing,
    has placed some good bets for the future. Or am I being wishfully
    optimistic? Can we have more substance and less "mood" into this debate?
    
    Kevin
                            

    Re: Alpha vs. P6
    
    is the speed performance difference enough to justify the cost?  

Further to .32 and .33...

Webstone benchmarks are run by the CSD Performance Group.  See
http://www.digital.com/info/alphaserver/news/webff.html for more
information.

For information on other industry standard benchmarks see the CSD Performance
Group Web page at http://sdtad.zko.dec.com/.

For a (hopefully) more organized view of the commercial performance landscape
see the Frequently Asked Questions (FAQ) at 

	http://yesmom.zko.dec.com/html/faq.htm

    Sorry to be so retro as to ask for a summary in DECnotes Marc, but is
    there a quick-and-dirty answer to the general question that can rally
    the troops reading here?
    

    Another good site to get performance results from is:

         http://www.tpc.org/

    This is the Tranaction Processing Performance Council site.

    - Bob

    The ALPHA is better. Infoworld June 17, 1996, www.infoworld.com.
    Performance isn't the barrier, it is the PRICE. The Digital ALPHA XL
    366 was twice (two times, 2X) as expensive as the Dell Dimension XPS.
    Who can justify that? 
    If the huge price difference wasn't enough to point out the new digital
    difference the magazine discovered our crumbling corporation can't
    support the product even at twice the price.
    Of course with the company paralized waiting to find out which VP's
    will survive the ABU/SBU merge, who is gonna care?

    re .41
    You won't see any Netbench, TPC/C, or BackOffice benchmark
    results between Alpha (32 bit or 64 bit OS) and P6 servers,
    if that's the "rally" you're looking for.
    
    SPECfp we got lots of tho.
    Kratz
    

    There are two problems with the ALPHA;  PRICE and Marketing (lack of). 
    It's just that simple.

    Number 2 (marketing) becomes a HELLUVA lot easier once number 1
    (price) is fixed, too.  Try marketing a hamburger stand that
    sells $40 hamburgers.  Alpha's pricing is Charlie Christ's
    problem now; for those that wish to send your cards and letters.
    Kratz
    

    Businesss Partner's understand that the AlphaStation and AlphaServer
    has a premium associated with it as it runs three operating systems,
    however, they were under the assumption that the Alpha/Celebris XL
    systems were suppose to be price competitive since it only runs NT?
    
    Are we not seeing this with these systems?
    
    	Regards,

>    Businesss Partner's understand that the AlphaStation and AlphaServer
>    has a premium associated with it as it runs three operating systems,
>    however, they were under the assumption that the Alpha/Celebris XL
>    systems were suppose to be price competitive since it only runs NT?

Of course, running multiple OSes is only interesting to the (relatively
small) set of customers that are actually going to use it, or think they
may use it.  Why does anyone think that people that aren't interested
in this "feature" will pay a premium??  Note that x86s can also run
DOS/Windows/Windows NT/Unix (more OSes, supported better, more native
apps, no hassles about separate consoles/PALcode, etc.)

As far as I'm concerned, this is nothing but another in a long line
of *excuses* for keeping prices high, which in turn have limited market
share.

Industry *revenue* growth last year was about 20% higher than Digital's.
(Adjusted for inflation, that borders on infinitely higher.)  Where is
"no excuses" management?  "Going for growth"?  etc.  (I was extremely
disappointed that it sounds like we're *planning* for sub-industry-standard
growth next year, at least in some areas.  What ever happened to benchmarking
against leadership competition??  (Or do we only do that when it matters
for costs, rather than for metrics like growth, or maybe revenue/VP?)

> Note that x86s can also run
> DOS/Windows/Windows NT/Unix (more OSes, supported better, more native
> apps, no hassles about separate consoles/PALcode, etc.)

But Intel can't run OpenVMS!  And that's all that matters ;)

    Re: -2
    
>Industry *revenue* growth last year was about 20% higher than Digital's.
>(Adjusted for inflation, that borders on infinitely higher.)  Where is
>"no excuses" management?  "Going for growth"?  etc.  (I was extremely
>disappointed that it sounds like we're *planning* for sub-industry-standard
>growth next year, at least in some areas.  What ever happened to benchmarking
>against leadership competition??  (Or do we only do that when it matters
>for costs, rather than for metrics like growth, or maybe revenue/VP?)
    
    I recently did a financial ratio analysis for DEC (compared to 10 other
    major firms in the industry) as part of a school project.  Asset 
    management and profitability are our glaring weaknesses (big surprise!).  
    I'm curious about the '96 annual report, but it appears that things haven't
    changed.  Even when Alpha's are 2x the cost of the competition
    our margins still suffer.  Of course, management should be aware of
    these ratio problems (not that management by ratio manipulation is
    always recommended).
    
    What is even more shocking is when you compare recent performances to
    the peak of 1987.  It shows that you can't hide margins in best-selling
    products like we used to.
    
    Mike

    re .50
    
    Hey ! what a coincidence, I too a did similiar ratio analysis, as a part
    my class project(Graduate program in Finance).I did the analysis on 5 major 
    companies. The information was arrived at based on publically available 
    information. Whenever company's revenue growth is in Sync with the
    market demand, it maaintains the market share. When the company over
    the industry growth, it gains market share- so it will have excess cash
    to acquire firms, pay dividends, buy back shares or decrease long term
    debts. When the company is NOT growing as rapidly as the industry
    growth typically the company will have problems such as Decreased
    Liquidity, Increased Liability, Decreased dividend, Asset sales and
    equity issue ( probably  preferred stocks ) etc etc.. 
    
    Ratios given below explains the previous noters (4673.50) comments.

    ( SOURCE : Balance Sheets of the 5 companies ).

    Jay
    
    ----------------------------------------------------------------------------
Revenue for past 5 years

        Digital	H.P.	IBM	SUN	SGI
	
1991    13911	14494	64766	3221	736
1992    13930	16410	64523	3589	906
1993	14371	20317	62716	4308	1132
1994	13450	24991	64052	4690	1537
1995	13813	31519	71940	5902	2238

    Here we seems to have stuck at 13+ Billion number, while every one else
    seems to have grown ( Some very significantly ).

Revenue Growth & for past 3 years
	Digital H.P.	IBM	SUN	SGI
1993	3.2	23.8	-2.8	20.1 	24.9
1994   -6.4     23.0     2.1     8.8    35.7
1995    2.7	26.1	12.3    25.8    44.9

3Yr 
CAGR%  -0.3	24.3	3.7	18	34.9


Total Assets
      
      Digital	H.P.	IBM	SUN	SGI
1993  10950	16736	81113	2767	1013
1994  10579	19567	81091	2897	1518
1995  9947	24427	80292	3544	2206


RATION ANALYSIS COMPARISON	

		Digital	H.P	IBM	SUN	SGI

Price		52.13	106.75	106.5	62.63	27.38
Price/Book	2.02	4.45	2.67	5.55	3.07
ROE		2.40	23.88	16.04	25.56	16.11
ROA		0.08	11.76	4.60	14.40	9.82

Profit Margin	0.09	7.72	5.81	6.03	10.11
Cost of Goods
Sold		67.9%	63.5%	57.8%	57.6%	41.8%

R&D		7.5%	7.3% 	8.4%	8.8%	11.1%

S&GA		23.7%	17.9%	23.3%	25.1%	27.8%

    (Now if we assume that our problem is S&GA, THINK AGAIN! we are in for 
    a BIG surprise! ) Next chart explains this clearly !
    

Ratio Analysis Comparison for Digital

			1993	1994	1995

Profit Margin 	
(NET INCOME/NET SALES)	-.1.7	-16%	0.09%

Cost of Goods Sold	60.6%	66.26%	67.99%
R&D			10.65%	 9.67%	 7.53%
Selling & Gen Admin	30.94%	29.64%	23.69%

    While S&GA and R&D has come down significantly in the past 3 years, 
    COGS has increased significantly. 
    
     

                                      

    Well, I'm not in graduate school, but still have to take Finance as
    part of the business core.  I used the 1994 SIC Disclosure database
    from the university library.  I formed industry averages using Compaq,
    Dell, HP, Silicon Graphics, Sun, Stratus, Unisys, IBM, Apple, and
    Tandem.  I'll do the same for the '96 report when it comes up because
    I'm curious to see what has changed over the last 2 years.
    
    Mike
    
    Ratio	             DEC    Industry
    Current                 1.17      2.08
    Quick                   1.14      1.47
    Inventory Turnover      6.73     10.13
    Days Sales Outstanding 83.90     78.06
    Fixed Asset Turnover    1.90      1.95
    Total Asset Turnover    1.39      1.35
    Debt-to-Equity          0.29      0.22
    Times-Interest-Earned   1.84     20.70
    Return on Assets        0.01      0.08
    Return on Equity        0.03      0.16
    Basic Earning Power     0.01      0.11
    Earnings Per Share      1.85      3.36
    Price/Earnings         28.70     17.73
    Book Value Per Share   22.32     28.15
    Market/Book             1.29      0.80

re: stuck at 13B

What about the fact that we have shed some revenue generators? Does that not
indicate that the remaining "core" businesses are increasing revenue?

r
[storage, terminals, ed services as examples]
 

    We have shed some revenue generators, and are moving (back) to hardware
    rather than s/w and (consulting) services, so I would have naively
    expected Cost of Goods Sold to go up as a proportion of revenue. E.g.
    Cost of Goods Sold for "Digital Consulting" when I was in it was zero
    (well, my wages still are nearly zero...).
    
    Alpha sales volumes are increasing (nice if they'd grow faster still,
    but...) Meanwhile, system prices *are* coming down. Would the volumes
    *actually* go up any faster if Alphas were any cheaper ? Would the
    extra volume more than outweigh the decreased unit price (the COST per
    box would be the same, the PRICE would come down) and increase the
    revenue and MARGIN ?
    
    I'm not a finance graduate. I'm not a VP. I haven't got a crystal
    ball. Your guess is as good as mine.
    
    regards
    john

RE: Growth.

We seem to have been stuck at around 13b in an industry growing 
20%+.

I believe that if Alpha products were more price competitive (particularly
at the low end) certainly volumes would increase.  Volumes would also 
have some economies of scale that would result in a lower cost per box.  
Particularly in Hudson chips.  Getting significantly more volume would 
(better) bring down the price of those chips which is a major portion of 
the cost of a base system.

One can question the pricing policy of Hudson on Alpha chips now.  They 
are extremely expensive, especially when compared to a Pentium Pro.  Not 
sure what the PA Risc costs or the Ultra Sparc, but would bet it is not 
as much as Alpha.

Whatever we choose to charge for a system, it is clear we are not growing 
at industry growth rates.  

What I would question is whether or not our Sales force even plans to hit 
industry growth rates.  In fact I would even suggest that any Sales manager 
that came in with a budget that was less than industry growth rates (should 
actually be more if we are aggressive) should be fired and told to go work 
somewhere else.

Are we aggressive enough in our expectations of ourselves?

    Growing the business always implies additional resources and
    investment, be they people, marketing budgets, capital expenditures,
    R&D, new products, etc.
    
    Everytime we start this effort, these additional line items become
    candidates for cutbacks.  So we eliminate them and then brag about how
    much additional "cost" we've removed from our structure.

    It seems to me that if you are going to be the "Hot Box" company with
    Alpha, you would think we would do whatever it would take (hmmmm......
    sounds like a marketing slogan we've heard somewhere) to get your
    products and NASA team into the testing labs and show them that we are
    the standard to which everyone else can look too.  I mean come on, this
    is as automatic as the Michael Jordan and the Bulls.
    
    So we have over 200 Veeps now and not one of them has the stones to
    step up and say, this ones mine and I'm going to make sure we don't
    screw this up.  Instead, we blow it big time and then from the sounds
    of it do a lot of Dilbert speak about motherboards (sounds like the
    perpetual liar on Saturday Night Live, "Yeah it's the motherboard, that
    it, that's the story!")
    
    Geesh people, it's not about benchmarks.  I made umpteem sales calls
    this week and you know what?  Not one, not 0, Zilch, Zippo, Nada of the
    people said to me, "I want to buy a SpecINT of xyz.  It's about the
    application stupid.  It's about solving the customers business problems
    and selling around and through the objections.  It's about pointing out
    the enemies cracks in their armor and convincing the customer we do it
    better.  Now, it would help that our illustrious line up of Veeps would
    somehow get their collective act together and deliver a NASA team with
    all the tools and blistering fast Alpha technology to the testing lab
    with such confidence that when we walk through the door we outa' be
    saying, "Wow, cool what a neat collection of ancient server technology"
    
    I just don't get it.  We sit lollygaging around instead of putting a plan
    together to go off and hit homeruns.  Man, I'm gettin' tired of trying
    to capture the high ground with a pair of unlaced Army boots and 45
    automatic with half a clip.
    
    This is getting exhausting....
    
    Mav

    Intel's P6 server variant has lots (512kb) of very fast (166 or 200Mhz)
    onboard L2 cache, or more than 5 times as much as Alpha.  Alpha tries
    to get away with a "one model fits all" architecture that works real
    well in the floating point intensive workstation role (lotsa Mhz =
    lotsa performance), but the smaller on board cache begins to show
    problems against the P6 optimized for the cache-sensitive server role.
    Alpha spins cycles waiting for external cache/memory while P6, albeit
    clocked slower, gets what it needs locally.
    
    A "hot box" for one role does not equate to a "hot box" for another;
    (so it's not as automatic as Michael Jordan and the Bulls... there's
    architecture/role issues).
    
    The question is: Does Digital push Limerock (PCBU's fast P6 server)
    over Alpha (where Limerock also mops up against the Mikasa and Sable
    low to midrange Alphas)... i.e. does Digital eat its children before
    COMPAQ, HP (and now Dell) have a chance?  (We could always lower the
    price on Alpha, but that doesn't seem very likely given that the
    semiconductor folks just point to SPECfp and close their eyes real
    tight when asked about other benchmarks).
    .02 Kratz 
    
    

RE: All

	We are in real trouble guys... We are just about to wipe out 
25% of MCS here in Oz, and I bet it wont be the managment that gets 
pushed. It will be people we talk to the customer and help them with 
there problem. If the customer has been burnt by bad support, do you 
think they will be more machines from us.. Not a chance.... 

	I know digital's policy ISN'T to bad mouth any other product, 
but very soon we won't have a product to bad mouth!, so we must 
change our marketing and sales QUICKLY or it won't matter. Many 
people around my area (Customer Support Centre) are already to 
annoyed and disillusioned with the way we are going so its going to 
need a BIG turn around to make us feel the way we used to feel about 
our gear. 

	I remember when we released the 2100 and most people loved 
them, now we are releasing the "RawHide" and Most people say, whats 
the point, we can release box's but we just don't sell enough of them 
to make any difference. Unless our NT program does something in the 
next 9 - 12 months, we can wave our good friends in digital good bye 
and just walk out. Its warmer and better paid out there..

Steve

    I thought the P6 onboard L2 cache was not on the same die, but rather
    another die in the same package.  Is this true?  If so, is this a
    feasible approach for Alphas?