Conference 7.286::digital

Article: 2800
Path: nntpd.lkg.dec.com!sousa.amt.ako.dec.com!pa.dec.com!decuac.dec.com!haven.umd.edu!news.umbc.edu!europa.eng.gtefsd.com!howland.reston.ans.net!vixen.cso.uiuc.edu!sdd.hp.com!hpscit.sc.hp.com!alech
From: [email protected] (Alec Henderson)
Newsgroups: comp.sys.hp.hardware
Subject: HP-Intel Announcement
Date: 8 Jun 1994 19:15:24 GMT
Organization: Hewlett Packard
Lines: 77
Message-ID: <[email protected]>
NNTP-Posting-Host: ppg01.sc.hp.com
X-Newsreader: TIN [version 1.2 PL1.4]
 
Here is the press release issued today by Hewlett-Packard and Intel
regarding our joint future development plans.  Please
do NOT send me mail with questions about this announcement; I cannot
provide any more information, nor can I give you any directions on
any place or any one who can.
 
HEWLETT-PACKARD AND INTEL COOPERATE ON
MICROPROCESSOR-BASED ENTERPRISE COMPUTING TECHNOLOGIES
 
FOR RELEASE Wednesday, June 8, 1994 at 5:15 a.m. PST
 
   SANTA CLARA and PALO ALTO, Calif.,  June 8, 1994 --
Hewlett-Packard Company and Intel Corporation today announced a
joint research-and-development project aimed at providing advanced
technologies for end-of-the-decade workstation, server and
enterprise-computing products.  The companies' efforts will
encompass 64-bit microprocessor designs, advanced semiconductor
processes and software optimization.
   HP is the second-largest computer company in the United States
and is the leading supplier of open-computing solutions for small to
large companies.  Intel, the world's largest chip maker, is a leader
in providing high-volume microprocessors for cost-effective desktop
computing.
   The companies said that by pooling their respective strengths,
they expect to create powerful new solutions that will deliver
unprecedented performance to meet the needs of users well into the
next century.  The planned architecture will maintain binary
compatibility* with both companies' software bases.
   The companies said they will work toward optimizing their
fundamental technologies to enhance their future product lines. The
work will be conducted jointly to take full advantage of
complementary capabilities in the two companies.
   "This effort with Intel is aimed at providing a unified computing
infrastructure that accomplishes three fundamental goals:  preserves
current customer investments, readies corporate customers for the
next century and offers high-volume cost models," said Richard W.
Sevcik, HP general manager, Systems and Server Group. "Intel brings
unparalleled high-volume semiconductor-process technology,
leading-edge design tools and an installed base estimated at 150
million personal computers."
   Albert Yu, Intel senior vice president, general manager,
Microprocessor  Products Group, said, "This is the linkage of our
expertise as a merchant market supplier of  tens of millions of
microprocessors annually, with HP's strength as a provider of
high-performance, Precision Architecture*-based workstations,
servers and enterprise-computing systems.  As we plan for the
future, we think this is a very complementary mix."
   Both companies have long histories of providing their customers
with ever-increasing performance and software-investment protection
through binary compatibility and are committing significant
resources to the cooperation.
   The companies expect to submit the proposed project for review
under the Hart-Scott-Rodino  Act.
    Intel, the world's largest chip maker, is also a leading
supplier of personal computer networking and communications
products.
   Hewlett-Packard Company is an international manufacturer of
measurement and computation products and systems recognized for
excellence in quality and support.  The company's products and
services are used in industry, business, engineering, science,
medicine and education in approximately 110 countries. HP has 97,000
employees and had revenue of $20.3 billion in its 1993 fiscal year.
   # # #
*Binary compatibility means customers' currently installed software
will run without modification and without performance penalty.
*Precision Architecture -- or PA-RISC -- stands for HP's Precision
Architecture reduced-instruction-set computing.
 
--
Regards,
Alexander Henderson
Strategic Relations Manager
 
Telephone: 408-553-4297                 Hewlett-Packard Company
Fax:       408-553-7656                 5301 Stevens Creek Boulevard
Internet:  [email protected]        P.O. Box 58059, MS 53U-FJ
HPDesk:    Alec Henderson/HP4200        Santa Clara, CA  95052-8059

    Oh I think we've got both of them scared...  real scared.
    
    Everyone is attacking Intel:  from Digital's Alpha AXP and the
    IBM/Motorola/Apple PowerPC to AMD and Cyrix.
    
    HP has peaked.  HP-UX is nearly out of steam and the PA-RISC needs to
    desparately be revamped to stay competitive for much longer.  Their
    quickly realizing that their computer success is in great part due to a
    remarkably popular laser printer line.
    
    This clinging to each other reminds me of Apple clinging to IBM.
    
    ..Larry

re Note 3157.2 by DPDMAI::ROSE:

>     HP has peaked.  HP-UX is nearly out of steam and the PA-RISC needs to
>     desparately be revamped to stay competitive for much longer.  Their
>     quickly realizing that their computer success is in great part due to a
>     remarkably popular laser printer line.
  
        Well, then, this must be about the best time to get top
        dollar for the AXP business and technology.  :-{

        Bob

    
    Isn't PA-RISC big-endian? And x86 little-endian? If so, how is it
    possible to build a chip that is binary-compatible with both?
    
    JP

It also seems that our lead over HP is a bit more than the two years that
was expected, if the new chips won't come out until the end of the decade.

a chip can be desinged to be both endian, the PA7100LC is a bi-endian and I 
think the PowerPC is also bi-endian

ed

To take advantage of our supposed lead, don't we need to make people believe
that 64 bits is needed NOW (for some type of application that they either are
using now or will want to use real soon)?

This announcement looks like more of a reassurance to the installed base that
an upgrade path will be available.
								- paul

    Hmmmmm -- Great piece of FUD here folks.
    
    If I read it correctly, the H/P and Intel folks are going to develop NEW
    COMPILERS -- That's the same knot hole we had to go through with ALPHA
    and the 64 bittiness.
    
    Soooo, seems to me we have something we can exploit -- We could use
    H/P's tactics against them -- Better than that -- If we had something
    from them in hard copy that we could show customers and then contrast
    that to this new announcement, we could discredit the company, rather
    than the product line -- SOrt of a killing shot, rather than a wounding
    one.
    
    Now, H/P has also announced (In DWT conference) that the are getting
    heavinly into SMP in the near term.  That suggests the whole story --
    
    Near term -- SMP
    Long term -- 64 bit
    
    There's FUD material in both areas.
    
    Also, I have an H/P internal use only Product/Company comparison sheets
    that compare their product line and Company to the competition. 
    Verrrry interesting!
    
    Here's an example of their FUD:
    
    DEC Weaknesses:				SGI Weaknesse
    
    Applications			Weak commitment to Stds		
    Weak O/S strategy			Weak System architecture
    Weak Compilers			Little expertise in Commercial Mkts
    Limited Scalability with OSF
    
    H/P Strengths
    
    Strong UNIX and Leader in Open Stds 	Strong Support of Open Systems
    ROBUST, Proven 64-BIT architecture  	
     w/excellent everything
    H/P performance better than DEC		Top technical and FP perf
    
    Knock-Offs
    
    This is a good one:
    
    "Now that OSF is no longer developing	Active OSF/cose participant
    SW, DEC is left as the ONLy company
    using OSF/1 and they can no longer 
    depend on OSF to help enhance it."
    
    Note the differences in the knock-offs.  I just put in some of the
    stuff they had.
    
    Anyway -- Time to kick some a##.
    
    IMHO
    Dennis
    

    
    I saw a great example of how to do this in ComputerWorld.
    
    Oracle had a full page ad that offered an Oracle migration 
    for Ingres users.  ASK/Ingres was recently bought up by ?
    (escapes me), and Oracle is seizing the opportunity to
    create FUD about Ingres' long term prospects.
    
    Great ad, great tactic.  In less than 3 months we could torpedo
    *everyone's* 32-bit UNIX's as dead-end products.  And position
    Alpha as a 64-bit-ready WNT platform to boot.

From what I read in the paper this morning, this Intel-HP chip will be VLIW,
not RISC.

    From what I read in the paper<C/S Gazette Telegraph> yesterday, Alpha
    isn't the fastest chip anymore! MIPS claims to have created the fastest
    chip in the world. The article stated that the new chip was faster than
    chips made by competitors Digital and IBM. The article didn't contain
    any specifications at all!
    
    2 year lead? Looks to be a little less than that.
    
    ....jc

    Jeez,
    
    One more person says that HP's performance has peaked and I'm
    going to have a coronary. We have been saying that for 1.5 years now
    and the June competitive guide shows them with the best price
    performance for the 735/125 and they discount us to death too.
    Not only that but their VARs and resellers do a great job
    of kicking our butts. 

    well, whatever the case may be it would be terrific if we could exploit 
    this little bit of news, but somehow I don't think so. Instead I will 
    probably receive another memo regarding "Telephone Etiquette"
    standards! sigh.....

    There's an article in Livewire on this topic.  I don't know how to post
    it here, but wanted to let everyone know that Digital has responded...
    
    Appropriately?  Who knows!!!
    
    Toodles.....JP
    

Worldwide News    Date: 09-Jun-1994

          Digital welcomes HP/Intel venture into 64-bit technolog
 
         Hewlett-Packard and Intel cast a vote of confidence this week 
   in the growth potential of the market for 64-bit computing by announcing 
   "a joint research and development project aimed at providing advanced 
   technologies for end-of-the-decade ... products," including 64-bit 
   microprocessor designs.
         "It's a strong message to HP and Intel customers that 32-bit 
   computing won't be enough to meet their needs," said Willy Shih, vice 
   president, UNIX and Windows NT systems.  "We welcome the news that 
   they're following Digital's six-year lead into 64-bit computing.
         "HP and Intel are now at the stage where Alpha AXP technology 
   was back in 1988," Shih continued.  "In six years, we've advanced from 
   research to commercially viable chips, migration tools, compilers, 
   64-bit UNIX, more than 5,000 third-party applications and more than 
   55 OEM partners.  
         "We've already gone through the transition from 32-bit to 64-bit 
   architecture," he went on.  "Don't confuse dreams with practical reality. 
   These transitions are much harder than people realize.  We've put a lot
   of that work behind us. 
         "We've lived through the difficult.  The chip architecture is 
   only the beginning of the hard work required to make this transition.  
   Migrating the operating system and recruiting solutions partners to 
   port their applications make the task more difficult."
         Shih said that the HP-Intel announcement "should make Digital 
   look more attractive right now, especially to HP customers.  
         "With nearly $1 billion in Alpha AXP technology sales, we've 
   proven the credibility of market, as well as the viability of solutions 
   that we can sell today.
         "We look forward to competing with HP and Intel at the end of 
   the century, when they are ready to begin selling 64-bit solutions," 
   he said.
         Digital is seeing a linear growth in the number of customers 
   requiring 64-bit computing capability today, according to Shih.
         "The market is here now, and we're ready.  We don't have to wait 
   two years for chips, systems, and applications," he said.


                       FOR DIGITAL INTERNAL USE ONLY
 

    Re .17>                  FOR DIGITAL INTERNAL USE ONLY
    
    That's the ticket! Let's keep it a secret! :-(
    
    Harry
    

    I think the HP/Intel announcement can be a positive thing for Alpha.
    One of the key elements keeping customers from buying Alpha was that
    we it was the only show in town. Who would want to risk porting  
    their business over to Alpha with the threat that DEC would not continue 
    to support it in the future, especially with DEC's poor performance
    of the past two years ? Now with HP/Intel's announcement, customers can
    feel safe in knowing that there is an alternative company should Alpha go
    down the tubes.

    Jim

re Note 3157.19 by PCCAD::RICHARDJ:

Jim,

Is there an assumption or missing step in your logic -- or were you being very
tongue-in-cheek?

HP and Intel haven't announced adoption of Alpha, as far as I can tell!

Bob

    rep:20
    Bob, 
         HP and Intel's announcement confirms to customers that 64 bit RISC
    is here to stay. Buying Alpha isn't so risky having this understanding.

    I guess that's my point, but it was late in the day yesterday.

    Jim

    Willy Shih is one of the best things to happen to Digital.

Note that HP/Intel aren't calling this thing RISC (or CISC for that matter).
Analysts suspect they're planning to do something with VLIW.

    	This looks to me like an opportunity to dust off the old "Digital
    has it NOW!" campaign.  Can't you see it in the Wall Street Journal?

    Right-hand page:

    		HP/Intel to deliver 64-bit processor in 1998

    		MIPS sampling 200MHz processor

    		Sun still trying to build SPARC that performs

    		IBM/Apple touting 100MHz 32-bit PowerPC

    Turn the page:

    		Digital has it NOW!

    	This feels like an opening we could drive a truck through.  If the
    Alpha people would come up with aggressive marketing and ad campaigns
    like the PC group has done recently, we could actually see this.  Does
    anybody from Alpha-land have a comment?

    			David

    re: Opportunity knocking.
    
    If we get Digital advertising behind this thing right now we'll have an
    effective ad campaign by Q4FY97, just in time for Intel and HP to
    launch their 10,000th box!!!
    
    Sorry, just a little cynical today.

    re: Willy Shih's response
    
    I'm very happy to see us responding to this FUD.  But what I really
    want to see is analysts pointing out the fact that Digital Has It Now. 
    Let's hope they pick up on this response and do some "convincing of the
    masses from an impartial party".  MHO.
    
    	Steve

    I agree with Steve. We need to hear this from the analysts. I would
    hope our PR folks are feeding this to them in a steady stream so that
    the most brain-dead of the analysts will see that Digital Has It Now.
    
    Gil

    	I'd just like to see that little twirp in the Globe write something
    with a positive spin for Alpha, and he's not going to do that unless
    someone in Digital goes and yanks him by the ear.

    	Am I wrong, or is all this too obvious?

    		David

    
    

Re: .9
    
>>    Great ad, great tactic.  In less than 3 months we could torpedo
>>    *everyone's* 32-bit UNIX's as dead-end products.  And position
>>    Alpha as a 64-bit-ready WNT platform to boot.


This is VERY FRUSTRATING, in that we obviously LACK the marketing 
"pit-bull-types" to really exploit this in the media, and therefore give us 
sales types an "edge" over the competition. 

We seem to practice an "almost-apologetic", cryptic and unintuitve marketing
and ad campaing as a whole. (Please refer to those dismal PC ads on CNBC --in
which we don't even bother to show a PC or PC environment.)

I will put together my own marketing brochures with the aid of PowerPoint. Too
bad we can't afford color printers/copiers  in the office....   :-(

Oh well, another wasted opportunity... I am sure ...!!   :-(

    The analysts are largely ignored by the buying public.  Digital needs
    to take advantage of the opportunity to stress that We have it Now! 
    Our first ad should have been in the SF, LA, Denver, Chicago, NY,
    Dallas, Houston, Atlanta, St. Louis, Philly, Washington DC, and
    Minneapolis papers this morning.  We are already late.
    

    The following excerpt is from today's Vogon News...
    
    
 Hewlett Packard, Intel - Enter deal to create 64-bit chip by end of decade
	{The Boston Globe, 9-Jun-94, p. 41}
   Intel and HP said yesterday they would collaborate on a new generation of
 microprocessor, an alliance that baffled as many players in the industry as it
 surprised.  If successful, the deal could further consolidate Intel's grip on
 the microprocessor market, where it dominates in the PC segment but faces
 stiff competition from those used to more powerful workstations and network
 servers.  The new partners could also present a potent threat to Digital,
 which is relying on its own advanced chip as its life raft to the future, as
 well as to other rivals.  But Intel and HP provided few technical of financial
 details, leaving many analysts, investors and competitors skeptical.
 "Theoretically, this deal is important.  In actuality, we can't say how
 behavior in the industry is really going to change," said Susan Frankle, chief
 HP analyst with International Data Corp. in Framingham, Mass.  The companies
 said engineers had been meeting for "many months" on plans to make chips and
 other products.  Specifically, they said they would create a 64-bit chip.

NOTE THE FOLLOWING: (<- my insert... t.)

 Digital already has a 64- bit microprocessor, the much-vaunted Alpha AXP chip.
 But while many analysts said an HP/Intel chip and the PowerPC would likely be
 market leaders of the future, less clear are the prospects for smaller vendors
 like Sun Microsystems and Digital.  Digital has made a huge investment in
 Alpha, including $25 million in a new chip fabrication plant in Hudson, and
 millions more to develop Alpha itself.  By all accounts, Alpha is a technical
 success.  It was the first 64-bit chip, and it uses RISC technology.  However,
 as Digital has found, speed alone does not ensure success in the marketplace.
 "I can now name two of the three surviving microprocessors in the year 2000,"
 said David Wu, a stock analyst with SG Warburg & Co. in New York.  "There is
 PowerPC and there is this Intel/Hewlett Packard product, whatever they call
 it."  "Alpha's chances look worse," Wu said.  "I would say it puts the future
 of Alpha under even more of a question mark than it is today.  Not that today
 is any great shakes."  

(Interesting observations!!!)

			Not so, said Willy Shih, a Digital VP.  He said Digital
 does not need a single high-volume partner like Intel to boost Alpha.  The
 chip, he said, will attract niche vendors who make such high-performance
 products as high-end workstations, or who will use the chip as an embedded
 processor, controlling the internal functions of equipment.  On Wall Street,
 investors were not impressed by the Intel/Hewlett Packard alliance.  Intel
 lost 1 3/4 to close at 59 3/4; HP dropped 1 1/2 to  75 3/4.  Digital was off
 1/8 at 20 7/8.  Frankle, the IDC analyst, said an Intel and Hewlett Packard
 partnership would not likely cause other hardware marketers to steer clear of
 Alpha.  "Most vendors have to sell products not in the year 2000 but in the
 year 1994 and 1995," she said.  "If Digital can demonstrate that its
 capabilities are real, then that's what people will base their decisions on."
 But Linley Gwennap, the editor of the Sebastopol, Calif.-based Microprocessor
 Report, said Digital is in an awkward position.  "If Hewlett Packard is
 saying, 'Gee, our RISC architecture is running out of gas,' there's nothing I
 can see that Alpha or Sun or those other guys will have any better luck with,"
 he said.  Intel and HP officials yesterday said their new chip would be
 neither CISC nor RISC, but something new or different.  While the companies
 provided few details, Gwennap and other analysts said the new products would
 likely take advantage of a technology called VLIW, for very long instruction
 word.  To become faster, chips now execute multiple instructions at the same
 time, Gwennap said.  However, there appears to be a limit to how much a chip
 can do at once.  If it works as promised, VLIW will make chips even speedier
 while reducing the demand that they handle multiple instructions
 simultaneously, he said.

> Digital has made a huge investment in
> Alpha, including $25 million in a new chip fabrication plant in Hudson

That's $425 million, not $25 million.

/Mitch

    Oops.... Happens when you're typing these things in at 1 a.m. ;-)
    Correction in Monday's VNS.
    
    						Trace

    
    IBM is ready to orphan AIX.  Another truck-sized opportunity for
    us if we'll seize it:
    
    
    "After months of soul-searching, IBM is set to embark on a fresh bid for
    the desktop software market that Microsoft now dominates.  The
    cornerstone of the PC software effort - and for shoring up IBM's entire
    software business - is a high-stakes gamble on what the company is
    calling Workplace technology."...
    
    "The fate of AIX, IBM's version of Unix, is now up in the air.  No AIX
    "personality" is planned for the Workplace microkernal."

    
    	Ohboy.....
     
    		IBM Shooting AIX
    		Intel Dumping x86
    		Hewlett-Packard Trashing PA-RISC
    
    		
    	Hey, Bob - people in here say somebody reads this stuff from time
    to time for you. Where we be on this? Isn't it time to focus our market
    on us as technology providers as opposed to the current technology soap
    opera d'jour. We're not talking press releases here, not print
    advertising, we need something more dramatic. How 'bout we do an
    Infomercial for PBS - or CBS for that matter (know for a fact they need
    the money). We have very articulate folks here, Bob. Customers, too.
    Lets get some attention. The Rip Van Winkle stuff isn't cutting it
    anymore. Hello....
    
    		the Greyhawk

Unconfirmed, but...

Heard from my customer today that they heard that SUN may be dumping Solaris!

 

Combating HP FUD

Comments by Dick Sites, Alpha Co-architect

Digital Internal Use Only



Time to take off the gloves. This is FUD. 

HP has announced that they are moving to a VLIW architecture near the
end of the decade. 

I'M DELIGHTED TO SEE HP ANNOUNCE THAT PA-PISC IS
APPROACHING END-OF-LIFE! 

We know it, they know it, and now they have publically announced it. the
reason they are moving to VLIW is that they have hit the wall in trying to
build a true multiple-issue implementation of PA to compete with the
upcoming 21164 4-way-issue Alpha. Their current "multiple issue" 7100
implementation can issue an integer instruction and a floating instruction,
but no other combinations. You can see this in the integer Specmarks,
which get no faster than the clock scaling when going from older
single-issue snakes chips to "dual-issue". 

The last time HP put their customer base through a transation (from the
HP3000 to PA), they did it very badly. Their customers should be asking
"what will be different this time?" 

64-bitness: 64-bitness is oversold. However, if HP wants to make an issue
out if it, consider these points: 

A computer is a 64-bit computer if it can do a 64-bit wide integer add. This
is the necessary primitive operation to do true flat 64-bit addressing. A
number of companies are now cloaking themselves in 64-bit claims, but
they aren't in fact building 64-bit computers. 

Claims include "we have a 64-bit floating-point datapath, so we have a
64-bit computer (intel)", "we have a 64-bit pin bus, so we have a 64-bit
computer", "we have old-style inflexible segmented addresses that consist
of a 32-bit integer and a separate 32-bit segment number, so we have a
64-bit computer (HP)", and "we have a 64-bit datapath but no software
that uses it, so we have a 64-bit computer (MIPS)." 

True 64-bit integer add (and therefore addressing) TODAY: 

Alpha AXP       YES (since 1992)
MIPS R4000      YES (since 1992)
Sun Sparc       no
HP-PA           no
IBM PowerPC     no
Intel x86       no

True 64-bit software shipping today:
Alpha AXP OSF/1 YES OSF/1
Alpha AXP VMS   no  (futures announced)
Alpha AXP NT    no  (have to ask Bill Gates)
MIPS            no
Sun             no
HP-PA           no
IBM PowerPC     no
Intel x86       no

FUTURES claimed for 64-bits:
Sun             YES, V.9 architecture. But their V.9 implementatin
                partner, HaL computer recently got rid of their
                founder and folded in to Fujitsu. They are a year or
                more behind schedule.
HP              YES, 1995 or later
PowerPC         YES, in about a year, in the 620 chip
Intel           YES, in P7

(If 64 bits is so terrible, why has EVERY SINGLE major player announced
64-bit futures? Why do those who don't actually have it claim to?) 

When the VAX architecture was announced in 1978, Ed deCastro from
Data General said "No one needs more than 16 bits." As we well know from
the book Soul of a New Machine, Ed was lying. When Ed said that, DG was
feverishly workingon their own 32-bit architecture, but were 2-3 years
behind Digital. 

HP appears to be doing the same thing -- claiming "You don't need 64
bits," but announcing their own 64-bit futures a few years from now. Which
is it HP? If you really DO need 64 bits in a few years, then buy Alpha now
-- there will be no 32- to 64-bit transition in a few years, or even in the
next two decades. If you buy HP now, there will be a transition to 64 bits in a
few years, and then ANOTHER transition to an entirely different, VLIW,
architecture at the end of the decade! Why sign up for TWO transitions
from a company that fumbled the last one for their customers? If you
DON'T need 64 bits, then their 64-bit futures are a lie. Which is is HP? 

These questions and comments are listed below: 

 1. Virtual Memory Addressing: As you know, most applications
   take less than 100 MB of Virtual Memory to run. With 32 Bits
   you get 4 GB of virtual memory, you will never go over this, why
   would you need more from 64 bits? 

   With PA's weird segmented addressing under HP/UX, you only get
   30 bits of address from each pointer (the other 2 are a segment
   number). So today's 100MB application needs 27 address bits, and
   HP has 30. For the past 25 years, demand for address bits has grown
   at the rate of 0.6 bits per year, so HP's 3 remaining address bits will
   last their customers about 5 more years. Of course, this is assuming
   that their customers continue doing 1980's standalone workstation
   computing and do NOT: 
      - Use large databases 
      - Use full text indexing of massive amounts of online data 
      - Use images 
      - Use video 
      - Use speech 
      - Use 3D 
      - Build servers 
      - Use additional cheap memory to do larger commercial
      sorts 
      - Use additional scientific processing power from RISC chips
      to work on larger matrices 
      - Use disks larger than 2GB. 2GB is the built-in file-system
      limit for 32-bit UNIX systems. Today's disks are 3.5 GB and
      growing about 50% per year. HP's customers were out of file
      system address bits YESTERDAY. Only Digital's 64-bit
      OSF/1 breaks this limit TODAY. large database companies
      understand this well and are moving to Alpha. HP either
      doesn't understand this or doesn't admit to it. 

   The last time HP ran out of address bits, with their HP3000
   machines at 16 bits, they spent a year doing a massive operating
   system rework to increase the address range -- from 16 to 18 bits!
   By the time the year was up (circa 1974), their customers were
   already out of address bits again. There is no evidence that HP
   actually understands the addressing needs of their customers. 

 2. All current software applications are written for 32 bits, 32 bit
   software performs best on 32 bit systems not 64 and we continue
   to prove this in on going benchmarks against Alpha. 

   All well-written software applications recompile and run on 64-bit
   machines. Badly-written C (and only c, not Fortran, Ada, Pascal,
   etc.) programs, often written by gradute students, do not recompile
   and run without source-code changes that will have to be faced in
   the near future anyway. Digital's customers with such programs can
   use the mx translator to convert them to Alpha OSF/1 programs
   that run just fine. Digital's example in moving our customer bases
   from VAX and MIPS to Alpha is well-recognized as being the best
   in the industry -- to the extent that other companies have expressed
   substantial interest. 

   All current software applications are not written for 32 bits.
   Applications from software companies that expect to make a profit
   across multiple platforms have no such machine dependencies. 

   If they did, such programs would also likely have big-endian
   dependencies and not run on HP's upcoming little-endian 7100LC
   chip, which HP will assure you is a trivial change. 

   Which is it HP? Are machine dependencies in C program a
   problem? If so, what is your customer migration strategy to 64 bits
   and VLIW and the 7100LC? If not, shut up. You can't have it both
   ways. 

   Software that only uses 32 bit data performs EXACTLY as fast on
   32 bit and 64 bit systems -- there is no FUD overhead of "unused"
   bits. There is no memory wastage for data items declared 32 bits. 

   HP wins some customer benchmarks against us, and loses some.
   They come to the table with an old (near end-of-life) architecture
   with well-tuned implementations and well-tuned compilers. We are
   in the first two years of an exciting new growth period with systems
   and compilers that are still undergoing tuning. In hundreds of
   specific contested customer cases over the past 2 years, we have
   done that tuning and won customer orders. All the improvements
   are rolled into more recent systems and more recent compilers. We
   will continue improving over the next several years. HP on the other
   hand has done all the tuning they are able to do. They have no levers
   left except to build faster chips, and their upcoming switch to VLIW
   is an admission that they fundamentally cannot build faster chips.
   Their losing customer benchmarks to Alpha or vice versa has
   nothing to do with 64 bits. 

 3. Also, as you know, all RISC CPU's have specially written
   Virtual Address Managers. Digitals Alpha uses 43 bits, HP
   PA-Risc uses 48 bits now. Who then provides the most from the
   RISC architecture HP or DEC? Conclusion is if memory
   addressing is not the issue then what is it? 

   This is pure FUD. The first Alpha chip implementations, the 21064,
   21064A, 21066, and 21068 have an implemented 43-bit virtual
   address, but they check the full 64 bits. We chose 43 bits because it is
   enough to last our customers through at least the end of the decade
   without making the initial chip too big, by which time the above chip
   implementations will have been superceeded. Future chips (as soon
   as 1994) will have more implemented address bits. In other words,
   we sized the initial implementations to meet real customer needs.
   Note also the inconsistency of HP claiming 64 bit addressing is too
   large, but 43 bits is too small, but their own 32-bit architecture is
   fine. 

   Which is it HP? Do your customers TODAY need more than 43 bits
   of virtual address? If so, your paging file must be spread across at
   least 4096 different 2GB file systems (see above discussion of the
   limitations of 32-bit addressing on file-system size). Show me a
   reference account. If they don't need 43 bits today, then shut up. 

   If they need more than 43 bits tomorrow, how are they going to
   manage the more than 8192 different required segments in the
   old-style HP segmented addressing scheme? HP only has 3 segment
   registers accissible to user code, and no HP/UX compiler support for
   managing them at all. Alpha is at 43 bits now with a flat address
   space, NO segment registers, and full compiler support. All of
   today's images will run on tomorrow's processors. We will be there
   for customers with more than 43 bits when they need them -- we
   are simply executing successfully and on schedule to the roadmap we
   announced in February of 1992. 

 4. HP and Alpha use the same 64 bit data paths and 30% of
   processor time is in handling the data. 

   This is a lie. HP does NOT have a 64-bit integer data path. In fact,
   Alpha has won character-string movement airline customer
   benchmarks against HP because of Alpha's wider datapath and full
   compiler and C library support. 

 5. Performance 32 bit vs. 64: If you run 32 bit software on 64 and
   don't take advantage of the 64, the 64 bit system actually runs
   slower than a 32 bit system due to cache fragmentation.
   Because, as you know, if you store a 32 bit data word in a 64 bit
   cache you also store with it 32 bits of zero unused. This
   fragments the cache, cuts the cach in half and the result is the
   system runs slower. These results are being proven in
   benchmarks every day using our customers program suites. 

   This is a lie. If you have software with 32 bit data, it takes up
   EXACTLY 32 bits in memory in both machines. Alpha compilers
   do not do something as stupid as storing 32 bits of data in 64 bits of
   memory. If the customer program declares 64-bit data, it takes up
   EXACTLY 64 bits on both machines. 

   It is true that to allow customers to take advantage of the large
   addressing, pointers under OSF/1 are allocated 64 bits and this can
   make some pointer-intensive programs take up more memory than
   using 32-bit pointers. Programs translated with mx use 32-bit
   pointers under OSF/1, as do VMS and windows NT programs. Only
   Alpha offers the choice, so custoemrs can use whatever is best for
   them. HP customers simply hit the wall at 30 bits of segment offset.
   If there is true custoemr need, future versions of OSF/1 could
   support 32-bit pointers. The SAME Alpha hardware runs all styles
   of software. 

 6. This was proven in your benchmark where a 99 MHz 735
   performed better than twice as fast as a 132 MHz DEC Alpha. 

   There are a few customer programs where this is true; there will be
   fewer over the coming months. There are also many customer
   programs where the opposite is true; and there will be more of those
   over the coming months. This doesn't "prove" anything. Also note
   the "132 MHz Alpha. The only system we ship like this (at 133
   MHz) is the 3000-300L, which is a low-priced,
   performance-crippled 3000-300. Don't ever let HP get away with
   comparing anything except their stripped system against our
   stripped system; force the comparison back to whatever the
   customer actually anticipates buying. 

 7. Does DEC supply a 64 bit compiler? Remember 1st there's the
   64 bit hardware, then the 64 bit operating system, 64 bit
   compilers and 64 bit libraries, which are all industry standard of
   coarse, before you can write your 64 bit applications. 

   HP messed up the lawyer's maxim of "never ask a question you don't
   know the answer to." 

      Yes, Digital's compilers are 64 bits (even under VMS and
      NT, you can declare 64-bit data). 

      Yes, Digital's hardware is 64 bits. 

      Yes, OSF/1 is 64 bits, and VMS has announced future
      support (again ask Bill Gates for NT). 

      Yes, Digital's libraries are 64 bits (the VMS and NT libraries
      use 64-bit data movement extensively when it buys
      performance). 

      Yes, we support the 64-bit standards that exist, and are
      leading the industry in establishing more; HP is following. 

   HP also left out our 64-bit calling standard across all languages and
   designed for the future, upward compatible for the next 2 decades or
   more, rather than THEIR plan to completely change the calling
   standard in a few years and make all customers recompile. 

 8. Why pay a premium now for 64 bits when you don't even need it
   now? 

   This is FUD. Could you be more specific about the price premium
   HP? Are you referring to our published chip prices against your
   complete lack of chip prices and sales, or are you referring to our
   workstation prices against yours of similar performance, or are your
   referring to our high-end system prices that meet customer needs
   that your workstations cannot touch, or are you referring the the
   DECsystem 300L as the ONLY low price-performance point on our
   curves? Which is it HP? 

Questions you should ask DEC:

 1. Is the C language interger 32 or 64 bit? 

   The default is 32 bits for int and 64 bits for long. This matches the C
   language requirements, and is what our customers told us they want.
   Since Digital is ahead of the rest of the industry in delivering 64-bit
   C implemetations, we had no industry standard to fall back on, so we
   asked our customers extensively. If this default doesn't fit your
   needs, a simple compiler switch chooses a different size. Other than
   ignoring the issue entirely, what did you do HP? (Note that if we had
   made the default integer 64 bits, they would have trotted out
   extensive integer programs that require TWICE AS MUCH data
   memory space and flogged us for that instead.) 

 2. Is the FORTRAN double 64 or 128 bit? 

   It is 64 bits, the same as the PA's 64-bit "floating-point
   doubleword" (their notation, PA-RISC 1.1 Architecture and
   Instruction Set manual 09740-90039 November 1990, page 6-37).
   What is the point here HP? 

 3. Is OSF1 the only 64 bit that DEC supports? 

   Today, yes, one more than HP offers. VMS has announced future
   support. However, the SAME Alpha systems that run OSF/1 also
   run OpenVMS, Windows NT, Netware, and more to come. HP only
   offers HP/UX as an open operating system at the moment. 

 4. Who else supports OSF1 in either 32 or 64 bits? 

   HP was one of the original founders of OSF. they have bid and
   contributed a number of software pieces to the OSF code base.
   Nonetheless, they chose to build their own proprietary UNIX system,
   used by NO other vendor. 

 5. How many real applications run under DEC OSF1? How many
   of these are the performance versions to run 64 bits? How many
   were written for 64 bit? 

   They missed the lawyer advise again: thousands; ALL of them; ALL
   of them. 

 6. How much does it cost to buy the full GEM compiler suite and
   pre-processors needed to insure maximum performance on
   Alpha OSF? 

   I don't know. What is HPs price-book entry for the KAP
   preprocessor, which they made infamous by their specmanship of
   using it to "crack" the Matrix300 program in the Spec89 suite
   (subsequently deleted from the Spec92 suite because the KAP
   results on Matrix 300 bear no relationship to customer-observable
   processor performance). 

 7. Will DEC supply a 64 bit version of Windows NT.? 

   The correct question is "Will Microsoft supply a 64 bit version of
   Windows NT?" Ask Bill Gates. If Microsoft does anytime soon, it
   will be developed on Alpha systems. 

 8. Why should we go through the Alpha migration program now
   only to go through another 64 bit migration program in the
   future? Since Digital hasn't really implemented 64 bits now as
   per points above and below. 

   This is self-directed FUD. There is no 64-bit migration in the
   future for Alpha customers. 64-bit hardware, data, compilers,
   calling standards, etc. are what ships today. When operating systems
   add support for bigger pointers, all old programs will continue to run.
   New hardware implementatins will run all old programs. 

   It is HP's customers who in fact face TWO migrations in the next
   few years (if HP stays on announced schedule), first to 64-bit PA,
   then to an entirely different VLIW architecture. 

 9. Why has Digital sold you down the garden path on 64 bits when
   you don't get the performance of 64 bits now? 

   This blanket statement is false. HP wins a few performance contests,
   Alpha wins some. We have a lot of happy customers who DO get the
   performance of 64 bits now. 

 10. Do I have to convert all of my software programs to be ANCI C
   compliant? 

   No. 

 11. Future Costs: If I go to DEC Alpha with 64 bits do I have
   increased RAM costs over 32 bit? 

   No, unless you need, and therefore declare, larger data items. But in
   this case, you need something that HP cannot deliver in the first
   place. This is a typical HP canard, attempting to draw attention away
   from their own shortcomings. 

 12. Do I have increased disk costs to be used as backup storage for all
   that virtural memory, half of which is not even used. For
   example if I have a 5 GB application will I need 10 GB of disk for
   it? 

   You will need as much disk space as your business work needs. If you
   can do more work with larger applications on a Digital platform than
   you can on an HP platform, you may well need more disk space for
   that additional work. "Half of which is not even used" is simply false.

 13. All of my software applications will double in size and take up
   more disk space and disk drive costs as you know are still $2.00
   per MB without the additional controllers needed. 

   False. Ask HP to point to specific full programs that take up twice as
   much disk space. Be sure to do apple-to-apples on shared libraries
   or not and on compiler optimizatin levels (low optimization tends to
   give large code), and on equal datatype sizes. Disk are under $1/MB
   in the Digital open market; maybe HP is still at $2. 

 14. Early adopters to the "Next Bit Technology" ALWAYS pay top
   dollar for off the shelf applications. 

   Eh? 

Non-disclosure questions again:

 1. The customer has read that William R. Demmer DEC V.P. of
   Alpha and VAX Systems said that "64 bits will enable the full
   potential at affordable prices of entire application arenas, THAT
   ARE ONLY NOW BEING EXPERIMENTED WITH TODAY;
   such as multimedia, imaging and virtual reality." A DEC V.P.
   has just told you that the Alpha Technology is unproven and is
   only intended for emerging, non-commercial applications like
   virtual reality entertainment games. 

   Don't be an idiot, HP. "Experimented with" modifies "application
   areas," not "Alpha systems" in the quoted sentence. Do you want to
   buy from a vendor that has no vision at all of upcoming application
   areas? 

 2. In the customers investigation of 64 bit technologies he
   discovered that the PA-Risc was architected in 1982 to be
   scalable to 64 bit, and has had 64 bit capability since 1986. Thus,
   DEC is just catching up, which they are, so I want the facility for
   64 bits when I need it. Why should I pay upfront for facilities I
   don't need today. 

   False on all counts. PA-RISC was never architected for 64 bit
   integers, only old-style segmented addresses with a 2-bit segment
   register number that can contain a 32-bit segment number, plus a
   30-bit (1GB) offset within a segment.HP does not have true 64-bit
   integers and address arithmetic today. You do not pay extra for the
   64 bit widths in Alpha, even if you don't use them. 

 3. HP has already implemented more 64 bit features than DEC as
   per all above. DEC has to run at more than 50% higher
   frequencies to produce the same performance as PA-Risc. 

   Digital has chosen to build "short-tick" implementations that do a
   huge number of short clock cycles, rather than "long-tick"
   implementations that must do a lot of work per longer clock cycle.
   As Linley Gwennap concluded in October xx issue of
   Microprocessor Report, the short-tick design scales better in the
   long run. Since HP is unable to build fast-cyle-time processors,
   their only design choice is the inferior long-tick style, which they
   attempt to make into a virtue. All of this is implementation style
   and has nothing to do with the instruction-set architecture. 

 4. All of this leads to the conclusion that buying Alpha is like
   buying a new Mercedes that has only first gear and no steering
   wheel, it is comfortable, and has terrific acceleration but you
   don't feel safe driving it. 

   Even HP's customers recognize that this statement is absurd. 


A postscript: The leader of HP's VLIW effort was one of the founders of
Multiflow Computers, which brought VLIW machines to market and went
bankrupt. So HP is in the curious position of asking people to buy dead-end
PA-RISC machines today based on the idea that HP will supercede them in
the future with a different, bankrupt technology. We should very
deliberately and publically have a field day with this. /dick 

   Back to the NYO Home Page. 

    I took a fast browse thru the NYO Home Page (referenced at the very
    bottom of that absolutely DYNAMITE document), but didn't find the link
    to it.  I need it to be on my hotlist!!!  Pointer please...
    
    (This is, btw, "Nature's Way" of saying that a WebServer should be
    WAIS- or CBRS-searchable from its front page... imho)
    
    Tnx for posting Sites' article here.  What a tonic!!
    
    PS -- Please PLEASE, Alpha Marketeers, take out an absolute MINIMUM of
    the combativeness of that note and recycle it into OFFICIAL, NON-
    "INTERNAL-USE-ONLY" sales combat materials!!!
    

    Why remove _anything_? It's a fantastic document! But I don't think
    it's new. I seem to remember seeing it last time HP announced VLIW. 
    Yes they have announced VLIW before, no partner that time.

    /Kjell

    re: .37
    
    Whew...
    
    Dick, you are the best!
    
    Everyone should extract .37 and distribute it widely throughout your
    organizations...  sales and marketing.
    
    BTW, has Dick done any similar comparisons to FUD by Sun?  Pentium? 
    IBM?  and especially SGI/MIPS?
    
    I'd be anxious to see them.
    
    ..Larry

    URL: http://www.nyo.dec.com/info/hp-fud.html
    
    I found it off of the "Sales Support" page,
    http://www.nyo.dec.com/info/sales-support.html
    
    					Steve

From:	US2RMC::"[email protected]" "Jon Callas" 11-JUN-1994 16:02:27.74
To:	[email protected]
CC:	[email protected], [email protected], [email protected]
Subj:	more on HP/Intel (fwd)

[forwards deleted]

Research Note
Copyright (c) 1994 Illuminata

Topic:  HP/Intel
Date:   June 8, 1994  (Preliminary)
Author: Jonathan Eunice, Terry Shannon

HP and Intel today announced that they will merge their HP-PA and
x86 architectures by the end of the decade. In addition to "full
binary compatibility" with both of its predecessors, the
resulting microprocessor is to implement a 64-bit architecture,
"super-parallelism" (described as "an extension of VLIW"), and
other unnamed technologies, including some that remain to be
invented. With this pre-announcement the companies claimed to be
setting a direction for not only themselves but also, "we
believe, the industry."

Our overall impression is decidedly negative. No products will
come from this arrangement for at least four to six years.
Product plans--even in outline form--are simply unavailable. And
despite numerous promises to dramatically leapfrog all other
competitors in performance and capability while at the same time
maintaining absolute compatibility with both source
architectures, no hint was forthcoming on how this colossal feat
would be accomplished. Remember that old saying, "if something
seems to be too good to be true, it probably is"? It seems to
apply. Rather than setting real directions, HP and Intel gave a
world-class demonstration of "smoke and mirrors." Indeed, in one
short morning they have exceeded even IBM's historically
prodigious capability to spread Fear, Uncertainty, and Doubt
(FUD).

Though the agreement may yield important and valuable results, we
are deeply skeptical. We have the following concerns:

*  No details are available. We automatically get itchy and
   suspicious when people start talking about new "paradigms"
   that will bring revolutionary advances, mysterious synergies,
   and forthcoming inventions. There have been too many
   instances of: "Yes, with a few adjustments to your
   carburetor, your car can run on water. Of course, we can't
   divulge any details, because the big oil companies are always
   watching."

   Though the teams have been working for several months now,
   there is no roadmap. They can tell us that the product will
   be great, and that we should wait for it, but not anything
   about what it will look like or how it will work.

*  Lots of baggage. HP and Intel have promised full, native
   compatibility with both PA and x86 architectures. Despite
   their protests that complete backwards compatibility has
   never been a problem and won't be a problem, we're not
   convinced. While Intel is adroitly using its investment
   capability and semiconductor expertise to keep current x86
   performance close enough to RISC performance to neutralize
   much of its competitors' hoped-for advantages, even Intel
   cannot do so without extensive effort and considerable
   delays. The problems caused by excess baggage will only
   increase. How do you, after all, have two instruction sets,
   two memory management models, two device interaction schemes,
   etc., and still manage to maintain compatibility with each
   half? Is it even possible? And if you do accomplish it, what
   price have you paid for this merger? Wherein lies the value?

*  Intense complexity. Modern chip designs are already extremely
   complex. The merger of two architectures (with two full
   instruction sets--a necessity given the promised native-mode
   binary compatibility) and new performance technologies (some
   not even yet invented) will make the design far more complex.
   Complexity increases the risk of failure, reduces
   performance, and slows development cycles. It's easy to
   imagine the HP/Intel processor a Rube Goldberg device, and
   hard to imagine it as anything else. You'd have thought Intel
   learned its lesson with the 432 project; if so, they've
   forgotten it again.

*  It's awfully far in the future! Despite one statement that
   they are "going to have their product to market very
   quickly," they generally set the expectation that joint work
   would not yield results for a number of years--after Intel's
   P7 and PA's separate 64-bit migration, for example. Five-year
   development plans didn't work for the Soviets, and they're
   even more ludicrous for today's computer industry.
   Technologies and their uses change so dramatically on a year-
   to-year basis that multi-year planning, and promises for a
   half-decade hence, are simply not credible.

Cheap Shots

In positioning both themselves and their competitors, both HP and
Intel were disingenuous and showed appalling arrogance. It's
something one expects from a quasi-monopolistic supplier such as
Intel. HP's standards of deportment, however, seem to be falling
to new lows. For example:

*  They talked about putting the customer first. That is pure
   malarkey--they're putting HP and Intel first. The trick for
   them is figuring out how to get customers to go along for the
   ride.

*  "If I were a competitor, I'd be really worried. If you think
   you have a future, you don't." Talking about those who use
   the chips, "if you happen to be with someone else, you'll be
   in trouble."

*  When asked if they weren't scared into this deal by the
   prospects of PowerPC or an inability to press forward on
   either PA or x86, they claimed they were not concerned with
   PowerPC, and in contrast, the PowerPC folks should be scared
   of them.

*  Said PowerPC doesn't have real binary compatibility because
   there are multiple chipset implementations. As though x86 and
   PA don't have multiple, slightly incompatible
   implementations!

*  When asked if Digital's AXP didn't have an advantage (as it
   already delivers a 64-bit architecture), they replied that
   "AXP is an early RISC implementation." PA86-64, on the other
   hand, would be a "leap forward in performance and
   capability."

*  "PA continues to be the best performer for transaction
   processing and real applications." Sometimes yes and
   sometimes no. Doesn't that depend on the application?
   Apparently not. Intel later called HP "the leader in the
   overall enterprise space." We'd let this standard kind
   marketing bravado go without notice were it not so much
   indicative of the hubris that now seems to afflict HP.

*  "We're not emulating the IBM/Motorola/Apple model--we can do
   better."

*  To "prove" that these companies know how to do generational
   migrations that protect customers, Intel claimed that HP's
   historical transition from the HP3000 to the PA RISC
   architecture was one of the smoothest transitions in the
   industry. How quickly they forget!

Other Tidbits

*  For HP, the arrangement represents an enormous image win. Of
   the many suitors for the desirable Intel bride, HP snagged
   the engagement.

*  If the agreement doesn't backfire, HP will be able to use it
   to extremely good advantage, given how well they sell
   strategically and sell themselves as partners for the
   customer. We "have the best products today. Now we have the
   best plan for the future."

*  Intel brings strong marketing, strong design, and its volume
   silicon processes to the table. HP brings an enterprise
   focus, RISC/UNIX expertise; microprocessor design skills,
   high-end silicon technology, and software (OS and compiler)
   technologies.

*  Both HP and Intel would continue existing designs for the
   next few years. So Intel's P6 and P7 and future HP-PA chipset
   would still be designed.

*  Given the parallel and redundant investments being made, the
   companies do not expect to reduce investment any time soon.

*  Developing Pentium took ~$400M-$500M over several years.

*  Positioning "super-parallel" technologies as an extension of
   VLIW (Very Long Instruction Word), RISC, and super-scalar.
   This is interesting, but not necessarily meaningful. Super-
   scalar RISC designs can already be thought of as post-VLIW
   architectures. Many vendors have already discussed plans to
   increase the number of parallel functional units they use.




% ====== Internet headers and postmarks (see DECWRL::GATEWAY.DOC) ======
% Received: from inet-gw-2.pa.dec.com by us2rmc.bb.dec.com (5.65/rmc-22feb94) id AA28321; Sat, 11 Jun 94 15:58:16 -040
% Received: from mv.MV.COM by inet-gw-2.pa.dec.com (5.65/27May94) id AA19257; Sat, 11 Jun 94 12:57:38 -070
% Received: by mv.mv.com (8.6.9/mem-931109) id PAA26684 for eristocracy-ll; Sat, 11 Jun 1994 15:55:45 -040
% Received: by mv.mv.com (8.6.9/mem-931109) id PAA26676; Sat, 11 Jun 1994 15:55:42 -040
% Date: Sat, 11 Jun 1994 15:54:24 -0400 (EDT)
% From: Jon Callas <[email protected]>
% Subject: more on HP/Intel (fwd)
% To: [email protected]
% Cc: [email protected], [email protected], [email protected]
% Message-Id: <[email protected]>
% Mime-Version: 1.0
% Content-Type: TEXT/PLAIN; charset=US-ASCII
% Sender: [email protected]
% Precedence: list

Who says Digital is not taking the offensive against our competitors.  Keep
hitting 'em hard, Willy! 

Jim

 Digital welcomes HP/Intel venture into 64-bit technology
	{Livewire, Wordlwide News, 9-Jun-94}
   Hewlett-Packard and Intel cast a vote of confidence this week in the growth
 potential of the market for 64-bit computing by announcing "a joint research
 and development project aimed at providing advanced technologies for
 end-of-the-decade ... products," including 64-bit microprocessor designs.
   "It's a strong message to HP and Intel customers that 32-bit computing won't
 be enough to meet their needs," said Willy Shih, vice president, UNIX and
 Windows NT systems.  "We welcome the news that they're following Digital's
 six-year lead into 64-bit computing.
   "HP and Intel are now at the stage where Alpha AXP technology was back in
 1988," Shih continued.  "In six years, we've advanced from research to
 commercially viable chips, migration tools, compilers, 64-bit UNIX, more than
 5,000 third-party applications and more than 55 OEM partners.  
   "We've already gone through the transition from 32-bit to 64-bit
 architecture," he went on.  "Don't confuse dreams with practical reality. 
  These transitions are much harder than people realize.  We've put a lot of
 that work behind us. 
   "We've lived through the difficult.  The chip architecture is only the
 beginning of the hard work required to make this transition.  Migrating the
 operating system and recruiting solutions partners to port their applications
 make the task more difficult."
   Shih said that the HP-Intel announcement "should make Digital look more
 attractive right now, especially to HP customers.  
   "With nearly $1 billion in Alpha AXP technology sales, we've proven the
 credibility of market, as well as the viability of solutions that we can sell
 today.
   "We look forward to competing with HP and Intel at the end of the century,
 when they are ready to begin selling 64-bit solutions," he said.
   Digital is seeing a linear growth in the number of customers requiring
 64-bit computing capability today, according to Shih.
   "The market is here now, and we're ready.  We don't have to wait two years
 for chips, systems, and applications," he said.

<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
  For information on how to subscribe to VNS, ordering backissues, contacting
  VNS staff members, etc, send a mail to EXPAT::EXPAT with a subject of HELP.

    Permission to copy material from this VNS is granted (per DIGITAL PP&P)
    provided that the message header for the issue and credit lines for the
    VNS correspondent and original source are retained in the copy.

<><><><><><><><>   VNS Edition : 3096      Monday 13-Jun-1994   <><><><><><><><>

<<many forwards removed>>

From:	ISLNDS::MALTZMAN "ALAN 229-7742 PD&M TECHY  10-Jun-1994 0803" 10-JUN-1994 08:15:02.08
To:	@CA.
CC:	MALTZMAN
Subj:	DIGITAL REACTION TO INTEL-HP ALLIANCE FOR 64-BIT MICROPROCESSOR

     "Digital welcomes the news that Intel and Hewlett-Packard are following 
      Digital's six-year lead into 64-bit computing, and looks forward to com-
      peting with them at the end of the century when they're ready."  




	FUNC: Public Relations Mgmt.          
	TEL: 223-4930                         <FREDRICKSON.MARK AT A1 at EMASA2 at MLO>
Date:	09-Jun-1994
Posted-date: 09-Jun-1994
Precedence: 1
Subject: PUBLIC RELATIONS ADVISORY: INTEL/H-P ALLIANCE                          


On Wednesday, Intel and Hewlett-Packard announced a joint effort to 
design a 64-bit microprocessor for "end-of-the-decade workstation, 
server and enterprise-computing products."

The attention created by this announcement -- as well as the absence 
of compelling information put forth so far by H-P and Intel -- will 
provide numerous new opportunities for Digital to remind the world 
that 64-bit computing is here now with Alpha AXP. 

The attached position statement and talking points may be used in 
taking advantage of these opportunities. 

In the U.S., news media inquiries should be directed to: 

Alpha AXP systems PR:		Alpha AXP semiconductor PR: 
Karen Quatromoni 		Linda Sanders
508-264-5358			508-568-6501
or				or
David Farmer			Lisa Lipson
508-493-0179			508-568-4352


Regards, 

Mark 




                 Digital Equipment Corporation's reaction
                      to Intel/HP chip announcement
    
    
    	 Digital welcomes the news that Intel and Hewlett-Packard are 
    following Digital's six-year lead into 64-bit computing, and looks 
    forward to competing with them at the end of the century when 
    they're ready.  
    
    	 With nearly $1 billion in annual Alpha AXP sales, Digital is 
    already demonstrating that a significant market exists for this 
    high-performance RISC technology.  Only 18 months after Digital 
    introduced its first Alpha AXP-based systems, their popularity is 
    growing at a staggering rate -- 66 percent from quarter to 
    quarter. 
    
    	 Digital began developing Alpha AXP in 1988 because the 
    company saw a market emerging for a high-performance, open 
    architecture.  As a result, Digital has the chips, the systems, 
    and the third-party applications that customers need, and has them 
    today.  
    
    	 Alpha AXP offers performance, openness, scalability and 
    longevity, on the operating systems customers want today and 
    tomorrow: DEC OFS-1 (Digital's UNIX), OpenVMS and Window NT.  
    Digital has the migration tools, the compilers, and in DEC OSF-1, 
    the 64-bit UNIX operating system that is the most open and 
    compliant with industry standards.  There are more than 5,000 
    applications shipping today on Alpha AXP from third-party software 
    developers.  More than 55 OEM partners are building Alpha AXP into 
    their products.  And constant improvement is being achieved on an 
    Alpha AXP technology that was designed to last 25 years.
    
    	 Digital led the industry transition to 32-bit technology in 
    the 1970s with its VAX architecture, and has the experience to 
    lead the way on 64-bit computing as well.  Based on the growth 
    being experienced today, Digital believes the market for 64-bit 
    computing will grow exponentially into the 21st century.  Alpha 
    AXP can help businesses make the transition now.  With Digital, 
    the future of computing is available today.
    


Talking Points

o   Digital welcomes the news that Intel and Hewlett-Packard are 
    following Digital's six-year lead into the 64-bit computing 
    marketplace.  We've proven the credibility of the market -- almost 
    $1 billion in annual Alpha AXP sales.  We look forward to 
    competing with H-P and Intel at the end of the century when they 
    are ready.  Digital is the pioneer in 64-bit computing. 

    Background: 

    1988	   Special engineering group looked into the need for 
                   commercial RISC and extrapolated that by 1992 
                   customers would need close to 100 MHz, and address 
                   space beyond 32-bit; Designed to be an open 
                   architecture, operating system independent, 
                   longevity into the 21st century; 2-1/2 years in 
                   development 

    October 1990   CMOS 3 

    December 1990  First silicon 

    January 1991   Booted ULTRIX on an Alpha prototype system 

    February 1992  First AXP chip announced (150-MHz DECchip 21064)

    June 1992	   Plans announced for $425 million semiconductor 
                   fabrication facility in Hudson, Mass., for future 
                   Alpha AXP microprocessors
    		   
    November 1992  First Alpha AXP systems (CPU clock speeds 133-200 
                   MHz), software, applications, services, and 
                   business practices announced
    		   
    March 1993	   Mitsubishi Electric Corp. signs agreement to become 
                   second source for manufacturing Alpha AXP 
                   microprocessors

    1993 - 1994	   Refreshed Alpha AXP line every 10-12 months, 
                   consistently and predictably, critical mass of 
                   applications reached


o   Alpha AXP engineering milestones: 
    
    . World's first full 64-bit microprocessor
    . World's fastest chip
    . World's fastest and best price/performance workstations at all 
      price bands
    . World's fastest and best price/performance servers
    . World's fastest performance on a PC 

o   Digital is seeing a linear growth in the number of customers 
    requiring 64-bit computing capability TODAY.  We are in that 
    market; we don't have to wait years for products.  The market is 
    here now, and we're ready; we have the chips, systems and 
    applications that customers need TODAY, and the technology with 
    longevity -- 25 years.  
 
o   Digital has gone through the 32-bit to 64-bit transition. We've 
    got the migration tools, the compilers, the only 64-bit UNIX (the 
    most open industry compliant standards), more than 5,000 
    applications, and more than 55 OEM partners.

o   Examples of Digital Alpha AXP chip customers:

    . Kubota, Mitsubishi, Carerra, Aspen, Cray, Vobis, Elite Group, 
      and various PCI periphal chip business in the future 

o   Examples of Digital Alpha AXP systems customers:

    . Multimedia customer service - Barclays Bank	
    . Video on demand - NYNEX
    . Database servers - Oracle
    . Virtual reality/video games - Visions of Reality
    . Object oriented software/BLOBS - Object Design
    . Statistical analysis - SAS 
    . MRP scheduling - ASK 
    . Image processing - Core Software Tech. 
    . Financial modeling - KaPRE Software 
    . Complex simulation - Gaussian Inc. 


o   Digital believes that this market will grow exponentially -- over 
    next few years and into the 21st century. 

    Examples of future requirements: 

   . Virtual reality			. Global information 
   . Language translation		. Multimedia 
   . Voice recognition			. Advanced simulation 
   . Artificial intelligence		. Imaging 
    
 
o   Digital's Alpha AXP systems business has taken off over the last 
    two quarters -- 15% of all Digital product revenue and 27% of 
    systems revenue, approaching $1 billion in annual Alpha AXP sales. 
    Alpha AXP revenues grew 66 percent from Q2 to Q3. 

o   The Intel/H-P announcement creates further uncertainty within the 
    H-P customer base over the future of PA-RISC. As H-P wrestles with 
    the challenge of bridging the gap for their customers between now 
    and the end of the decade, H-P customers can discover the benefits 
    of 64-bit RISC computing now by migrating to Alpha AXP. 

o   We welcome this latest validation of the importance of 64-bit  
    computing. We've passed the learning curve and are now on a 
    straight course with applications, systems, chips and the 
    experience we can offer our customers. 


                              #   #   #

================== RFC 822 Headers ==================
Received: by xirtlu.zk3.dec.com; id AA02530; Thu, 9 Jun 1994 09:57:40 -0400
Received: by blkang.zk3.dec.com; id AA12268; Thu, 9 Jun 1994 09:57:39 -0400
Message-Id: <[email protected]>
Date: Thu, 09 Jun 94 09:57:39 -0400
X-Mts: smtp

    
    A public relations response is not a campaign.
    
    While the info is good and valid, when's the last time
    you read Word Perfect's response to a Microsoft announcement,
    or AMD's response to the Pentium announcement?
    
    The announcement gets everyone's attention, the responses
    get filed.
    
    I want to see aggressive advertising.  I want to see editorials
    commenting on our advertising.  I want to see Fear, Uncertainty
    and Doubt interfering with every sales call that HP, SUN and IBM
    make when selling their UNIX's!
    
    I'd also like to see AMD endorse the Alpha architecture in a
    partnership similar to that of HP/Intel and IBM/Apple/Motorola.
    The endorsement would add further credibility to the architecture
    and would get us a lot of press attention.  As a contest between
    alliances, we would be compared favorably and possibly frequently
    with the PowerPC and ? (VaporPC ?).

    
    We have to be careful that we don't start heaving stones at the
    cometition.  Redirect the focus from what they are going to do, to what
    we have already done.  HP and the others have some ammo against us and
    the perception (valid or not) regarding our viability.  We have a
    golden opportunity here, will we take smart advantage of it or not?
    
    
    
    Mike

    We can *use* HP and Intel.  Let them sell folks on going to 64-bit. 
    Instead of investing in marketing 64-bit architectures, leverage our
    marketing on their efforts.  Perhaps an ad along the lines of:
    
    [Picture of two pole-vaulters, about to make the big jump.  They grab
    their poles.  One is marked "Alpha."]
    
    So, now that you're ready to make the jump to 64-bit ...
    
    [Cheering crowds, slow motion as the first vaulter approaches an
    impossibly high goal.]
    
    Just make sure you don't come up a little ...  short.
    
    [First vaulter "swishes" about 6 feet under.]
    
    Digital's 64-bit Alpha ...
    
    [Second vaulter easily goes over, "Alpha" and "Digital" marks
     all over the place.]
    
    Why take chances with anything less than the best?  
    Digital.  64-bits?  We have it now.  In fact, we've been setting
    records with Alpha for some time.  
    
    Let us help you set your own records ...
    
    [Cheering crowds.]
    
    
    Hey, so it's not great, but it's a shot.  You get the general idea.
    
    Steve
    
    Steve

    <-- YES! Excellent!

    >>[First vaulter "swishes" about 6 feet under.]
    
    How about first vaulter smashes head into crossbar.
               ;)
    
    ..Larry

    
    	Better yet - how 'bout first vaulter impales himself on broken
    pole.
    
    	

    
    Re: .47,
    
    Made a few changes, Steve. Hope you don't mind. Great concept!

    -----------------------------------------------------------------------
    
    [Picture of two pole-vaulters, about to make the big jump.  They grab
    their poles.  One is marked "Alpha."]
    
    So, now that you're ready to make the jump to 64-bit computing...
    
    [Cheering crowds, slow motion as the first vaulter approaches an
    impossibly high goal.]
    
    Just make sure you don't come up a little bit... short.
    
    [First vaulter falls short, knocking the pole off of the beams...]
    
    Digital's 64-bit Alpha AXP systems...
    
    [Second vaulter easily goes over, "Alpha" and "Digital" marks
     all over the place.]

    ...have been setting records for quite some time now.

    [scrolling graphic: Fastest CPU, fastest database sort, etc, etc - 
    maybe reference Guiness Book of Records]
    
    Why take chances with anything less than the best?  

        Open Systems.
	TRUE 64-bit computing.
    
	We have it now.  [Pole vaulter waving to cheering crowds. Finish 
			  with big Digital logo on screen]

    Let us help you set your own records blah blah blah. Call 1-800-DIGITAL.
    
>    Hey, so it's not great, but it's a shot.  You get the general idea.

	Actually, I think it's superb, or could be with a little development
	by persons who do ads for a living. (Too bad we don't know any such
	persons &^).
    
    

    Or how about George Burns as the second pole vaulter.
    
    ;) (reference to 3153.x noters)
    
    ..Larry

    re: .51  I like it!  But mostly, I would like us to cash in on the
    marketing efforts of others -- especially when the "goof" of others is
    to use their marketing to sell something they don't have that that we 
    clearly already offer.
    
    Steve

    If I see it on T.V. I can say I read it here first! 

         Actually I think the analogy is a bit off.

         Show a pole vault with the pole set at the "64" level.  At
         regular intervals the pole is cleared by men and women in
         various dress, suits, casual, etc.  The poles they are using
         have Digital printed on them.

         The view scans the line of people waiting their turn.  You
         can see people joining the line, all with Digital poles. 
         They are coming from a line of concession stands.  Each stand
         has a logo, Digital, HP, Intel, IBM, Apple, etc.  These
         stands are selling the poles.  The Digital poles are
         obviously longer ("64") versus the others which are "32" and
         "16".  The people buying at the Digital stand get in line for
         the "64" level vault.  May be there should be "16" and "32"
         vaults?

         The camera zooms in on the HP and Intel booths which are next
         to each other.  The HP and Intel sales person are checking
         out a Digital "64" pole and agreeing that they need to build
         one like it.

         thanks,
         Fran

>         The camera zooms in on the HP and Intel booths which are next
>         to each other.  The HP and Intel sales person are checking
>         out a Digital "64" pole and agreeing that they need to build
>         one like it.

	The HP and Intel reps attempt to glue one of each of their "32" 
	poles together end-to-end to create a new, better "64". But of
	course they can't agree on which ends to glue together...

    
    
    
    How about a commercial involving horses.  Have a scene labelled them
    with a foal struggling to get up (actually in this scenario it would be
    more like the horses mating, but I don't think that would do very well 
    in a commercial ;')).  For us, we can have the Kentucky Derby winner.
    
    Mike

    re:56
    
    
    brilliant
    
    
    rick

    Today's Boston Globe has an editorial cartoon that pictures two
    fishermen.  One, rather rotund, is labelled "HP/Intel" and he is
    shown with a "thought balloon" picturing a large fish labelled
    "64 BITS".  The other, thinner, looking more "seasoned", has
    a balloon picturing a tiny fish that looks sick.
    
    					Steve

    Shouldn't it be dubbed VLIV (Very Large Integrated Vapour)??
    
    -dermot

    We ought to place an add in all of the trade rags, TV and radio stating
    that Digital offers HP and Intel an ALPHA to help them develop their 64 bit
    chip... We could say that it will take too many years to develop such a
    a chip on Intel or HP technology, so build the best with the best... 
    Then end it with ALPHA, the fastest RISC technology available today in 
    the world. ;)
    
               

    re: .63
    
    This idea is so good that it won't be taken seriously!  Maybe you could
    throw in some frivolous notion also that we'll supply the engineering
    expertise as well and your idea will get more attention!
    
    Great idea in my book!
    
    cw

This announcement is one of the best things to happen to us in a long
time. An opportunity like this doesn't come along very often.

    			How about this then
    
    We display the news about HP and Intel and then say
    
    			Been there, Done that
    
    
    	Then we give them the stats on how many applications can be run
    under each platform that we support with our 64 bit ALPHA.
    
    ;)

    From PC Week -- the latest guess as to when this thing might possibly
    ship...  That'll mean at least five years from announcement to
    first ship.
    
    Intel's 64-bit 'Merced' chip won't ship until 1999 By Lisa DiCarlo and
     Rob O'Regan
    
    SANTA CLARA, Calif.--Intel Corp.'s first 64-bit processor, code-named
    Merced, is now scheduled to hit the market sometime in 1999.  Contrary
    to widely published reports that late 1998 was the target date for the
    processor, Intel spokeswoman Marion Koehler said the company has never
    publicly committed to a time frame except to say the chip would be
    released by the year 2000.  The shift means that Microsoft Corp.'s
    64-bit version of Windows NT won't appear until 1999.  Microsoft and
    Intel executives said at last September's NetWorld+Interop show that
    they would synchronize the release of their respective 64-bit
    products.  Merced is a high-end server processor with both CISC and
    RISC properties. Intel worked with Hewlett-Packard Co. on the initial
    design; however, Intel will be responsible for manufacturing,
    marketing and licensing.

    
    What an opportunity for "Digital has it now"!!! :>
    
    
    I'm sure this has been said before. But maybe it will start sinking it.
    

    >>Intel's 64-bit 'Merced' chip won't ship until 1999
    
    Um, gambling is not allowed in here, otherwise I would be offering odds
    about this mirage chip not being there till the next millenium.
    
    ..Kevin..

RE: .66

>				The shift means that Microsoft Corp.'s
>    64-bit version of Windows NT won't appear until 1999.  Microsoft and
>    Intel executives said at last September's NetWorld+Interop show that
>    they would synchronize the release of their respective 64-bit products.  

Does anyone have any definitive answers for this?  It goes against what we
were told earlier, about 64-bit NT being on Alpha first.  Has Microsoft
changed direction, or are they continuing to tell people what they want
to hear?  Can we have some official statement about this?

-- Ken Moreau

    Look at it like Win32s...
    the initial VLM support will be an add-on to NT to enable Alphas
    to have 64 bit apps that can address >2Gb (much like Win32s was to
    WinV3.11 to have 32bit apps).  The "real" 64 bit NT comes later.
    K

    It would be nice to hear some news on this - .66 does make it sound
    like Microsoft is saying, "If Intel won't be ready until 1999, we
    will actively hold up release of 64-bit NT until then." Is this the
    case, or is the reporter simply using an earlier statement (that
    Microsoft and Intel said in September that they would synchronize
    their 64-bit products) and assuming that the statement still holds
    even with this delay?
    
    With UNIX beginning to push "64 bit computing," it seems hard to
    believe that M$ would actively hold up the 64-bit release. I know
    Microsoft probably has a strong relationship with Intel, but I
    doubt they'd hesitate to snub Intel if it were in their financial
    interest...
    
    Awaiting the "official" word.
    

    
    
    
    
    
    None of the following based on real knowledge just supposition.
     
    Keeping in mind Microsoft's business model rather than their confusing
    "announcements".  Although confusing announcements seem to be part of
    their business model!
    
    Microsoft will probably release a marketing beta of NT 5.0 this year
    which has some 64-bit support.  The only hardware commonly available
    will be Alpha.  That also meets some promises they made to the market
    and to customers without actually  shipping a product.
    
    They will hold their official release of commercial code until Merced
    is available.  Why release a new O/S before there is a volume platform?
    That's the way they've done it in the past.
    
    

    
    http://www.microsoft.com/corpinfo/press/1996/jun96/nt64btpr.htm

RE: .73 -< If you want "official," try... >-

Perfect: exactly what I needed.  Thanks!

-- Ken Moreau

    
    I got
    
    ERROR
    
    The requested URL could not be retrieved
    
    While trying to retrieve the URL:
    http://www.microsoft.com/corpinfo/press/1996/jun96/nt64btpr.htm 
    
    The following error was encountered: 
    
         Connection Failed 
    
    The system returned: 
    
        (61) Connection refused
    
    ;>)

From http://www.microsoft.com/corpinfo/press/1996/jun96/nt64btpr.htm

"This functionality (64-bits) is targeted for availability in the Windows 
NT "Cairo" time frame".

Note that Microsoft has subsequently re-defined "Cairo" from a specific 
NT release to a set of enabling technologies.  Don't assume that 
NT5.0 = Cairo = 64-bit!  Perhap somebody can also post a pointer to the 
Cairo re-definition.



						-Paul