Conference 7.286::digital

    That's too good an idea to waste by just suggesting it in a notesfile. 
    I don't know if anyone with enough clout to get it done reads this file
    but if so, please pass it up the line.  Or/And, if anyone has a contact
    perhaps the author of the base note could mail that person.  FWIW
    though, the key is to respond QUICKLY.  A response 3 months from now
    would have a much reduced/little/no impact.
    
    Jack 

As far as I know, Digital has not yet published any benchmark numbers
for the TPC-D benchmark (which is what the table reports).

Let us not act smug here.  While performance wise the 8400 still
outshines the UltraSPARC servers, price-performance wise Sun is
significantly ahead.  Combine this with the aggressive discounting I
am seeing from Sun in the field, we are COMPLETELY OUT OF THE
PRICE-PERFORMANCE ballpark.

The AlphaServer folks have their work cut out to get our AlphaServers
back on the price-performance curve DEFINED by Sun.

    -.1
    I was afraid of that. If it takes 3 months to do the benchmarking....oh
    well..
    
    /bu

    re: .0 This type of ad is typical Sun. I've noticed that almost every
    Sun ad I've seen which is promoting any network products is full of
    incorrect information (i.e. lies). They usually say something about how
    Sun had the "first xxxxx" network product, or Sun was the "first"
    vendor connected to the Internet or some such rubbish. It infuriates me
    that a company can get away with blatantly lying to the public. I would
    love to see Digital place ads which read something like:
    
    "Sun says they were the first to do xxxxxx"
    
    "Here's proof that Digital was actually first" (show proof)
    
    "They were wrong about this.....what else are they wrong about?"
    
    
    or some such wording.
    
    Debbie
    

>    "Here's proof that Digital was actually first" (show proof)
    
When looking through an old RFC that showed the order the earlier internet
domains were registered in, guess who I was pleasantly surprised to see was
the FIRST?

-mark

    re .5
    >It infuriates me that a company can get away with blatantly lying
    >to the public.
    
    Like that second source for Alpha (Mitsubishi) we announced in
    March 1993?  The Kobe earthquake excuse only lasted so long.
    Would you believe 17 year locusts now?  ;-)

    Sun's marketing is working.  I helped out a sales rep from Texas this
    week who was very close to closing a deal using Alphas and AltaVista
    search.  Sun came in with an UltraSparc and Ultraseek - and they will
    be doing a pilot and bakeoff - no sale for Q4.  Now it's a scramble to
    get resources that will work for free to help the customer have a
    successful pilot.  Want to take bets on who wins the business?
    
    bill

    Someone at a seminar indicated that Federal Government Organization
    that created/funded the Internet owns addresses 1.-15. and that Digital 
    was the FIRST commerical company to own an Internet address e.g., 16.
    
    Anyone know if this is true and this would be the type of Positive
    Marketing to indicate that Digital has been there the longest.
    
    	Regards,

	My favourite Sun-ism is their saying that went around about
	a year or two ago. "The Network is the System"

	Funny, we were saying that EXACT same thing about 10 years ago.

	Mark my words, Digital is in make or break mode and the key group
	to our success is not Engineering or Manufacturing, or even
	Sales. 

	It's Marketing. 

	Read what you want into that. This is the organization that needs
	some serious attention from upper management. I'm sure they are
	trying their best. I actually have no doubt. But why do I
	feel they are not being pushed (or helped) to their greatest
	potential? 

							mike

    jeez....
    
    1) Force a benchmark with your customers application (note first word -
       implies some effort, but usually worthwhile..).
    
    2) Read, inwardly digest, then *use* the stuff that Marc (Lando::)
       Warshawsky's team put together re competing against Sun 
    
    3) Try convincing your punter that *their* application is more
       typical of their application than TPC-A/B/C/D/Z. Then see 1) above.
    
    4) Dont ask Kratz for an Alpha endorsement. [How many more conferences
       will be hit with 'we lied' ? - give it a rest, then go help someone
       sell some systems, please !]. (ps - no high end results available
       from Intel server based systems... Hmm...)
    
    
    AW 
    
    
    

    ..and anyone seen the recent HP press saying 'our chip is faster than
    Alpha and Sun, and when we ship systems in the coming months, won't
    everthing be rosy, especially as we are also doing another chip with
    Intel.'
    
    Also includes nic graph showing H outperforming Alpha 400mhz.

    Actually, HP's 180MHz PA8000 based workstations do outperform our
    400MHz Alpha chip on SPECfp95.  And, they will be announced on Tuesday
    (June 4).  Sorry folks but we can't use the old "they're not shipping
    yet" for very much longer.
    
    As the person responsible for pulling together the positioning for the
    field, I will be posting a write-up to our web site sometime next week.
    
    Check out http://sbu.mro.dec.com/hq/comp for the latest competitive
    information on Sun, SGI, HP, IBM and Compaq.
    
    Julie

    Re .7
    
    Calling the Mitsubishi agreement a "lie" is a bit strong.  The
    agreement was real, it is still in force, and Mitsubishi did build
    Alpha architecture chips.  Can you buy Alpha chips now from Mitsubishi?
    Probably not.  The 21066A design was primarily a test bed for the 
    process and found little market, but Mitsubishi engineers continue to 
    work on Alpha designs using the 0.35-micron process.

    I sent this, the ad, up to Jack Novia V.P. Marketing for the America's
    asking him if we will be preparing a response to this B.S. from Sun.
    
    If he replies and gives me permission to post his reply into this note,
    I will do so.
    
    Enough of this, let's get to it.
    
    Mav

Does this explain why so many of us are getting frustrated at what appears 
to be the static nature and lack of response of some in the company? I 
found most of this article fascinating and helpful but the dichotomy 
displayed in the following really surprised me.


From the upside magazine article referenced elsewhere in this notesfile:

>Upside:
>Do you feel that Digital has done as good a job as it could in taking 
>advantage of Alta Vista?

>Palmer:
>Six million hits a day. You can't buy that kind of advertising. People 
>that use it like it. So that is a nonhype style of marketing and that's 
>what I prefer. Maybe it's an East Coast/West Coast thing. I'm not taking
>anything away from Scott McNealy, [CEO of Sun Microsystems Inc.]. He's 
>done an outstanding job of marketing his company and their products. 
>They've been a very innovative company; they're agile and a good 
>competitor. But we're more conservative. So I'll tell you to try my Alta 
>Vista search engine. I don't have to sell it; you go try it.

Maybe going broke when you shouldnt is an East Coast/ West Coast thing too.

Maybe having your field people spend all their time fighting crap 
benchmarks, convenient fictions and and unsubstantiated hype instead of 
closing deals is an East Coast/ West Coast thing too.

I'm about as West Coast as it gets, in fact I'm thousands of miles west of 
the West Coast ( so far West I'm actually on the East Coast!) and, quite 
frankly, I cant see how marketing your products and services effectively 
has anything to do with conservatism. 

What it has a lot to do with is the convenient manner our corporate origins 
can be used as an excuse to mask a lack of desire to go that extra inch to 
succeed. "Whatever it takes" becomes "Whatever it takes on the East Coast".

And, given that our Corporation is a diverse international undertaking, 
perhaps its about time we started to adopt a little of the winning 
characteristics and talents of this diverse culture (while of course 
maintaining some of the better aspects of our East Coast roots).

Committing to your products would be a kind of Armonk, New York, kind of 
thing..
Being proactive about benchmarking would be a good, East Coast kind of 
thing...
Advertising our software would be a good, Washington, kind of thing...
Marketing our hardware's capabilities would, indeed, be a good, West Coast 
thing to do...
Committing to a marketplace instead of a product would be a good, European, 
kind of thing...
Having Product managers who never respond to the field would be a bad kind 
of East Coast thing.
Investing in staff competence would be kind of a nice, Japanese coast kind 
of thing...
Using what we sell would be a good, West Coast kind of thing...
Continually improving our processes and practices would be a good Texan 
kind of thang...
Having a beer occasionally with the underlings is a good, West Coast kind 
of thing but a bad, East Coast kinda thing...
Saying, "we dont do that sort of thing on the East Coast", would be a kind 
of "Everwhere but the East Coast" kind of thing...
Taking your research and turning it into product quickly, then leveraging 
it to the hilt, is a West Coast thing.

In fact, I shudder to think of all the atrocities I've probably caused by 
not knowing how they would do it on the East Coast. 

This is not to throw the baby out with the bathwater. Just realise we need 
to change more than the headcount to be successful. We have to turn the 
company into a globally directed enterprise that is not afraid to play by 
rules set elsewhere other than the East Coast. 

Just think, we could come up with lots of good things that could improve 
our company because it obviously isnt perfect now. And all we have to do is 
recognise that we shouldnt bind ourselves to old models based on how our 
original culture came about.

BTW. I, like may others, have been in Digital for many years and not 
visited HQ on the East Coast (come to think of it, I havent been to any US 
coast, or the middle for that matter). I didnt really understand the deep 
sense of loss staff felt at not getting turkeys each year. Or the 
situational displacement that no longer going to Canobie Park must have 
been to people. Maybe its an East Coast kind of thing, something so 
ingrained into the psyche that to understand the difference...Nah, its just 
an institutional comfort zone that stops us from doing what is needed to 
really turn things around...Anyone for a HP style "beer-bust"?


- John
Canberra, Australia.

                      What a great note.  Kudos, John!!
    
                             Words to live by...
    

    >	My favourite Sun-ism is their saying that went around about
    >	a year or two ago. "The Network is the System"
    
    Sun's version was/is: "The Network is the Computer".
    
    ...petri

RE: .19

	They ran with "The Network is the System" a couple of years
	ago. Now they use "The Network is the Computer".

							mike

    
    re .5
    
    We justified our claim on DIGITAL 'having the most internet experience"
    for TV based on the registration on Arpanet - as a matter of fact Bill
    Hawe of Networks had a Arpanet map circa 1977 where all but 2 systems
    on that net were PDP machines.  
    
    Without Alpha having the same benchmark test, we are unable to run a
    competitive ad.  

|   Actually, HP's 180MHz PA8000 based workstations do outperform our
|   400MHz Alpha chip on SPECfp95.  And, they will be announced on Tuesday
|   (June 4).
    
    Today is June 4, they announced their systems.
    
|   Sorry folks but we can't use the old "they're not shipping yet" for
|   very much longer.
    
    Announced not equal to shipping.  According to the Wall Street Journal,
    PA8000 workstations are available in the fall.
    
    The AlphaStation 500/400 is available NOW.  Beating a shipping
    workstation by the end of the year is great for the people who
    want to wait.
    
    								-mr. bill

re .-1

>    Announced not equal to shipping.  According to the Wall Street Journal,
>    PA8000 workstations are available in the fall.
>    
>    The AlphaStation 500/400 is available NOW.  Beating a shipping
>    workstation by the end of the year is great for the people who
>    want to wait.

It is? Could you please give me a part number and estimated ship date? All
I can find information about are the 266 and 333 MHz workstation varients.
There is an upgrade that's ORDERABLE that will take the 500 to 400 MHz but
the announcement stated that the upgrade won't ship until October.

Sounds like we'll ship at about the same time H-P does. How does that make our
product superior?

tim m.

    Anyone see the latest Sun Observer Vol 10. No. 6 June 1996:
    Sun launches next-generation network servers.  The article is very
    positive in favor of Sun's new systems.
    
    ...
    Sun introduced four new Ultra Enterprise servers for department and
    data center computing environments that scale from six to 30
    processors. ...
    ...
    "You can pull a board out at the low end and put it in a high-end
    machine," Sakakeeny said.  "You obviously have to change the box as you
    move up, but the power supplies, the boards, the memory, everything is
    the same up and down the line.  You can't say that about the RS/6000,
    for example, or the HP9000s.
    ...
    "The scalability they've [Sun] announced is one of the best among the
    leading Unix suppliers in the industry," Bretzmann said.  "The main
    competitors, HP and IBM, are down around 12 and 14 processor range in
    HP's case and six to eight processors for IBM.  So, they're surpassing
    what those two supplierss are offering," he said.
    ...
    
    DEC not mentioned once is any article in the paper that I could see.
    
    	Regards,

    RE .17
    
    Wow.  What a note.  I wonder if the wonder boys upstairs are staring at
    that one.
    
    Naaaaa, that would speak to a reality check which is obviously not a
    realistic thought.
    
    A favorite Dilbert comes to mind:
    
    Dilbert and the pointy hair boss are standing outside of the V.P.'s
    office.
    
    V.P.:  "I'm running late.  But since I'm a Vice President you'll have to
    wait in the hallway."
    
    V.P.: "You'll be able to judge your relative worth by observing what
    things I do while you wait."
    
    Dilbert peering into the office:  "He's teaching himself the banjo!"
    
    Point hair boss is staring off into oblivion with wide eyed look.
    
    It's a west coast thing or an east cost thing, but someone please
    decide what thing it's going to be and get on with it.  You know when
    you are riding the tiger you can't dismount!
    
    Mav
    

    Posted with the author's permission.
    
    Greg
    
    
                  I N T E R O F F I C E   M E M O R A N D U M

                                        Date:     04-Jun-1996 02:15pm BST 
                                        From:     David Kerrell
                                                  KERRELLD@A1CHEFS@RDGMTS@REO
                                        Dept:     Marketing
                                        Tel No:   7-830-2279

TO:  Greg Hilton@LZO

Subject: Response to HP's PA-8000 ad.

                    Response to HP's PA-8000 advertisement.
    
    Prepared by David Kerrell, Server Marketing, Systems Business
    
    Last week HP placed an advertisement for their new 64-bit PA-8000 chip 
    on page 69 of Computer Weekly. In our opinion this advertisement is 
    very misleading for the following reasons:-
    
    HP quote the following figures:-
    
    	    	    		SPECint95	SPECfp95
    HP PA RISC 8000 		11.8		20.2
    DEC Alpha 21164, 400 MHZ	11.2		15.2
    
    We have never published the figures they've used for the 400MHZ Alpha 
    chip. Our published estimated figures are:-
    
    	    	    		SPECint95	SPECfp95
    DEC Alpha 21164, 400 MHZ	11.7		15.9
    
    Our actual figures for the AlphaServer 4100 5/400 are:-
    
    	    	    		SPECint95	SPECfp95
    DEC Alpha 21164, 400 MHZ	12		17
    
    Secondly their figures are not for the product that they will ship in 
    June as implied in their advertisement but for a 180MHZ chip not 
    expected to ship until November. Their June shipping product will 
    contain a chip at 160MHZ which has the following performance:-
    
    	    	    		SPECint95	SPECfp95
    HP PA RISC 8000 160MHZ	9		17
    
    As you can see we outperform them today.
    
    So they are quoting our figures incorrectly and they are comparing 
    currently shipping product from us with their future products. Lastly, 
    they are implying they have faster product now, when they do not.
    
    If they were to compare their future product with out future product, 
    we still win:-
    
    	    	    		SPECint95	SPECfp95
    HP PA RISC 8000 180MHZ	11.8		20.2
    DEC Alpha 21164, 500 MHZ	15		21
    
    If you have any questions call me on 01734 202279 DTN: 830 2279
    Email: dave.kerrell@reo.mts.dec.com

re:.26

I very much think this information should to be sent outside of Digital to the
press for publication. Distributing it inside Digital does no one any good.
Digital employees do NOT buy Digital Workstations or Servers. OR why not
advertise what we really have now and the futures just as HP is doing? Is it a
crime for Digital if we advertise our future products? 
Please DO SOMETHING FOR A CHNANGE

    Interesting how "today" and "future" can be defined to suit your case. 
    5/400 systems are still a future as far as I am concerned, unless I can
    order and ship it today, which we can't.

re.27:

This is being actioned in the UK. I can't speak for the rest of the world.

Re.28:

Good point. I could argue our 400MHZ chip is just as available today as HP's
PA-8000 160MHZ.

Dave. 

     .26>                           SPECint95       SPECfp95
     .26>   HP PA RISC 8000 180MHZ      11.8            20.2
     .26>   DEC Alpha 21164, 500 MHZ    15              21
    
    Does the fact that HP's 180 MHz system is just barely slower than our
    500 MHz system set off alarms for anyone other than me? It would appear
    that as HP gets better at their processes and is able to build chips
    with higher clock speeds, they're going to be eating our lunch, even
    with clock rates that are 50% of the Alphas. Is their processor that
    much more efficient at pipelining data or is there another explanation
    why they get such incredible performance from their system?
    
    .27> Is it a crime for Digital if we advertise our future products? 
    
    Perhaps not a crime, but a few years ago Digital took a lot of abuse
    for the delays in getting Alpha systems to market. We announced them so
    long before they were really available that many customers got very
    wary of *anything* we said. Only Microsoft seems capable of sucessfully
    marketing vaporware....
    
    Harry
    

    .30> Does the fact that HP's 180 MHz system is just barely slower than
         our 500 MHz system set off alarms for anyone other than me?
    
    Yo!
    

 rep .-2

 There's a note thread in DECHIPS which discussed one specific
 architectural issue between Alpha and HP. The issue deals
 with an in-order / out-of-order microarchitecture design. The
 assertion is that Alpha will achieve a signficant performance
 gain by 'converting' to an out-of-order execution without increasing
 the clock speed. The HP architecture has already converted to this.

 Anyway, it was interesting reading.

>    Does the fact that HP's 180 MHz system is just barely slower than our
>    500 MHz system set off alarms for anyone other than me? 
    
    Only if you assume that 500 Mhz is our limit for EV56. I do not. 

>    It would appear
>    that as HP gets better at their processes and is able to build chips
>    with higher clock speeds, they're going to be eating our lunch, even
>    with clock rates that are 50% of the Alphas. Is their processor that
>    much more efficient at pipelining data or is there another explanation
>    why they get such incredible performance from their system?

    EV6 will get it's performance increase over EV56 from the design, not 
    the process.  

    
-Bruce
    
    

    >Does the fact that HP's 180 MHz system is just barely slower than our
    >500 MHz system set off alarms for anyone other than me? 
        
     >>>  Only if you assume that 500 Mhz is our limit for EV56. I do not. 
    
    Bruce, I think you missed my point. HP can do at 180 MHz most of what
    our current Alphas need 500 MHz to perform. It looks to me like if they
    can just squeeze their processor up to 200 MHz, they'll be kicking our
    butts performance-wise, with 300 MHz to spare! 
    
    Now imagine their performance if they can improve their process to get
    to a 500 MHz version of their new chip. If the numbers extrapolate in a
    linear fashion, then it looks like their 500 MHz system would run at
    more than twice the speed of our current 500 MHz system. I hope that
    our EV6 design is capable of making a giant leap forward in throughput
    speeds so that we maintain an edge. Otherwise, I don't see Alpha
    systems having an advantage in the marketplace for much longer. :-(
    
    Harry
    

    I was waiting for someone to tell us about a new technology called CISC
    
    8^)
    
    r

    Watch the spin: when Ed caldwell boasted "We are going to make
    the fastest microprocessors forever" (see recent Boston Globe
    article), he was talking about Mhz clocking speed, not overall
    performance.
    

>    Bruce, I think you missed my point. HP can do at 180 MHz most of what
>    our current Alphas need 500 MHz to perform. It looks to me like if they
>    can just squeeze their processor up to 200 MHz, they'll be kicking our
>    butts performance-wise, with 300 MHz to spare! 
    
    Harry, read the whole reply.  I also said that EV6 will get it's
    increase in performance from the design not the process.
    

  This may have already been mentioned, but there's a *LOT*
  of detailed technical discussion in RICKS::DECHIPS regarding
  the different styles of architectures and implementations
  among at least:

    o H/P PA RISC
    o PowerPC
    o Alpha


  Rather than duplicate all that here, folks may want to refer
  to that notesfile.

  If you don't already have this conference in your notebook and
  would like to add it, press <KP7> or <Select> or type "SELECT"
  and the conference will be added to your notebook.

                                   Atlant

    Well, a 400 MHz unit may not be available to *you* today, but the
    system I'm using today was built by Manufacturing on 1-May.    But we
    got an early unit; official FRS is June.
    
    The exact SPEC numbers submitted for the AlphaStation 500/400 were:
    
    	SPECfp95   14.1
        SPECint95  12.3
    
    /John Henning
     CSD Performance Group
     Digital Equipment Corporation
     henning@zko.dec.com
     Speaking for myself, not Digital

    	Plans are in place and production is starting so
    	that the first revenue AlphaStation 500/400MHz
    	machines should ship in late June. Price file is 
    	slated to be up just before that.
    
    				Rich Dischler
    				Eng/Program Manager
    				AlphaStation 500

    From http://www.alphastation.digital.com/960311a.html
    
    "The AlphaStation 500/400 will be available in June."
    
    That's what the press release said back on March 11, 1996.  It's now
    June, 1996, and Rich Dischler tells you that this workstation will
    indeed ship in June 1996.  This isn't a surprise to me.
    
    This conference is full of replies explaining why something couldn't be
    done.  I've had the privilege of working with some people who just do.
    People who work over, under, around and through barriers (no matter
    what kind).  They take their commitments seriously.  
    
    -----
    
    But let's take a quick look at a snap shot of the end of the
    First Half Calendar Year 1996.
    
    Uniprocessor Workstations Shipping in H1CY96
    
    					SPECint95	SPECfp95
    Digital AlphaStation 500/400	12.3		14.1
    Sun Ultra 2 1200			7.72		11.4
    HP J210XC				4.64		8.30
    IBM	43P/133				4.72		3.76
    SGI Indigo2 10000			8.9		12.5
    Intel Alder Pentium PRO 200		8.09		6.75
    
    Adding in a couple of shipping dual processor workstations:
    Sun Ultra 2 2200			7.81		14.7
    HP J210XC				4.64		11.3
    
    Brief discussion:
    
    Sun can just beat out our SPECfp95 with a dual processor system, but
    lags significantly in SPECint95.
    
    HP's PA8000, first expected to be available Q1CY96, then
    expected to be available Q2CY96, now is expected to be available
    Q3CY96 or Q4CY96 (depending on your trust in HP or the Wall Street
    Journal).  They will *not* beat us at SPECint95 when they ship,
    though they might actually hold a brief lead in SPECfp95.
    
    IBM is finally sampling the PPC620.
    
    MIPS announced amazing chip estimates for the R10K/200 and R10K/275
    (and to a lesser extent, R5000).  SGI delivered workstations and
    severs with a fraction of those estimates.
    
    Intel began the year as the SPECint95 champ, but now has to be content
    with SPECint95 price/performance.  (A very nice thing to be content
    with, Digital's new PP200 systems look nice.)  Rumors of clock speeds
    of 233, 250, 266MHz Pentium PRO by the end of H1CY96 did not materialize.
    
    -----
    
    Finally, prospects for the future?
    
    Chip estimates from Digital Semiconductor have us beating HP
    at SPECfp95 this calendar year with EV56.  Everytime Digital
    Semiconductor has put out estimates I've wondered how we were going
    to do it, but everytime we've done it.
    
    If SGI can deliver a system that matches MIPS initial chip estimates
    for the R10K/200 SPECfp95, they'll be a contender for SPECfp95
    as well.  MIPS has also promised 275MHz by the end of the year.
    (Having lived through a couple of MIPS promises in the past, we
    shall see.)
    
    
    And Microprocessor Forum will probably be very interesting again.
    
    								-mr. bill

    
    >>             -< Their 4-banger can nearly outrun our V8! >-
    >>
    >>>Does the fact that HP's 180 MHz system is just barely slower than our
    >>>500 MHz system set off alarms for anyone other than me? 
    >>    
    >> >>>  Only if you assume that 500 Mhz is our limit for EV56. I do not. 
    >>
    >>Bruce, I think you missed my point. HP can do at 180 MHz most of what
    >>our current Alphas need 500 MHz to perform. It looks to me like if they
    >>can just squeeze their processor up to 200 MHz, they'll be kicking our
    >>butts performance-wise, with 300 MHz to spare! 
    
    You really should check out the DECHIPS conference mentioned a few
    replies back.  I attended a presentation put on by Hudson on EV6
    vs the competition.  This was a concern they had an answer for.
    The design chosen by the HP team places limits on how fast a system
    they can build.  It's very probable that for at least another few
    quarters it will not be possible for them to produce a system faster
    than around 180-200Mhz with that chip.  Anything above 200Mhz for
    that chip _may_ not be cost effective or practical.
    
    In fact, don't doubt our Hudson processor teams.  They have carefully
    looked at all possible competition, and designed EV6 accordingly.
    EV6 is a new design - which incorporates some design features that
    the newest HP, Intel, Sparc, and MIPS chips have now.  Think about
    this: if we can beat them with an 'in order execution' EV56 design 
    - think about an EV6 with a very agressive implementation of the Alpha 
    Architecture.
    
    Regards,
    
        Jim D.
    
    PS - If it were 'in vogue', I'm sure Hudson could produce an Alpha
         Microprocessor which could smoke the competition with a 33Mhz
         input clock.  In fact some of our chips can do just that (with
         internal frequency multipliers).  It just doesn't give the same 
         perception of speed....
    

    The main problem is that HP is MARKETING their new chip, printing
    numbers that many readers (customers) will believe. Meanwhile, we're
    debating the issue in notes. I doubt that very many Digital employees
    purchase Alpha systems. 
    
    In addition to the Wall Street Journal, I heard news about the HP
    announcement on both CNN Headline News and The Financial Network that
    day. That kind of exposure reaches millions of people. Notes reaches a
    few thousand at best, most of whom will never purchase a workstation.
    :-( 
    
    Harry
    

I also saw a COMPAQ ad in yesterday's WSJ with server performance numbers and again they show up 
as best in class but just happen to leave Digital out of the table.  We have got to get more of 
our ads with correct positioning out there.

        <<< Note 4629.44 by ALFSS1::nqsrv337.nqo.dec.com::Kevin Ryan >>>
                               -< COMPAQ WSJ Ad >-

I also saw a COMPAQ ad in yesterday's WSJ with server performance numbers
and again they show up  as best in class but just happen to leave Digital
out of the table.  We have got to get more of  our ads with correct
positioning out there.

    Take a look at www.tpc.org and you'll see that in that Digital is
    nowhere to be found in the top 10 $/tpm-C chart.  We certainly are
    found in the top 10 for actual tpm-C performance (all based on
    Turbolaser), but in the midrange, specifically with MS-SQL server
    Compaq beats us in actual performance and price/performance.  The 4100
    will make put us back into the hunt, but the Oracle/VLM numbers (just
    over 6000 tpm-C) recently posted for the 4100 are only 300 tpm-C
    greater than those that Compaq posted for a quad 166Mhz Pentium Pro
    system with MS-SQL.  If this is the level of advantage we get with our
    state of the art AlphaServer combined with 64-bit UNIX and VLM, then I
    expect to have a tough time convincing customers to pay the premium for
    this slight advantage.  I hope we have some other performance tricks up
    our sleeve.
    
    So don't be surprised if we're not in the chart of midrange systems
    that Compaq is competing against.  We ain't in the hunt.  Compaq is
    devouring the NT server marketplace.

re "Why aren't WE in the news?"

We ARE in the news. In fact, I just heard on the radio while driving to work 
that Digital laid off ~250 workers in Q3 and is on pace to match that number 
in Q4 :^(

    The comments about HP's and others clock speeds is very confusing.  I
    have had customers ask why we seem to be very fast, it requires very
    high clock speeds. As was stated here, if our competitors ran at our
    speeds they would kill us.
    
    The fact is, as I was told by a techie type, that clock speeds are
    really meaningless unless you stay with the same processor.  Also, and
    thi s I have no clue as it is technically over my head, there is an
    internal and external clock.  Apparently one is about half of the other
    in most sytems byt Digital.  We match clock sppeds and report
    accordingly.  Others report one of the clock speeds, shich is actually
    slower.  If this is true, double their reported clock spped an see what
    you get in terms of performance per clock speed.
    

     RE: 4629.48

    >>>if our competitors ran at our speeds they would kill us.

    Yea....if only cows could fly!

    The point is that HP designs are optimized in different ways and
    *cannot* simply be made to run at Alpha speeds.

    The clock speed in a CPU chip is like the number of cylinders in
    automobile engine.  Neither can be used as a single indicator of
    performance.

    Do V8 engines *always* run faster than V6 engines?  No!

    For the same reason, CPU chips cannot be compared solely on clock speed.


    							-Paul


    

    How do we make the comparison ?
    

    >How do we make the comparison ?
    
    By how much "work" each system can accomplish per unit of time. 
    eg: benchmarks (spec, tps, etc)...

    re .49
    
    > *cannot* run at
    
    Nothing personal, but every time I hear this kind of statement I am
    reminded of a conversation some years ago;  where someone was very much
    convinced that the laws of physics would prevent hard drive data
    densities much greater than what was in an RD50 [the mention of a multi
    GB drive that could sit on your hand without crushing it would have
    been ridiculed].
    
    r
    
    (OK. it was a BUNCH of years ago 8^)  ).
    

    Ah yes, just the like the fellow at Hudson that told me back around
    1983 that it was impossible to make a semiconductor chip with
    geometries smaller than 1 micron. I told him that he was wrong and
    would be proven so in the years to come. Smallest that I've heard of
    being produced these days is .35 micron (That's point three-five) and
    I'd bet that someone, somewhere is working fast and furious to get even
    smaller than that.
    
    Harry
    

    re: .53 (smaller than .35 micron)
    
    At a Large-AeroSpace-Company I frequent, they are down around .1 (If I
    remember correctly), but these are specialized devices in the 90+GHZ
    range for satellites and jet fighters.
    
    They also told me that Digital has a setup "just like theirs".
    
    Cool stuff.
    
    .mike.

    Re: .53 and .54 - although not really "common", .2 /.25 is 
    produced in commercial quantities right now, and as mentioned in
    note .54, .1 is produced for R&D, military, etc., etc.  Makes for 
    some "interesting" lithography equipment.
    
    	Bill
      

    I remember sitting in a Hudson presentation some years ago. .1micron is
    the size where electrons are no longer feasible? Where you must go to
    photons?  Memory dims.....
    
    

    You have to get a lot smaller than .1 micron before you start sqeezing
    electrons.
    
    Below .1 micron some interesting things can happen, including some
    quantum effects, but you don't have to go to photons.
    
    Besides, .1 micron is very short wave ultra-violet, so the photons
    would be somewhat hard to handle with today's optical technology.