Conference 7.286::digital

    Is this PC going to use standard PC parts (ie standard PCI
    motherboards, standard scsi disk and controllers etc) or are we going
    to create another RAINBOW?? 
    

    Re. .1,
    
    Well, the Alpha XL Personal Workstation *is* a PC with standard PC
    parts (a Celebris XL) with an Alpha daughtercard.
    
    Regards,
    David.
    
    
    
    

It'll be a standard ATX form-factor board, with standard PCI & ISA slots, standard 
IO cards, fitting in a standard case.

Look at chipbz::production$:[ds_info.alphapc164] for some PS files on the PC164 
board that'll be shipping this month in volume.

Jim
DS FAE, Europe

    Not that I'm a 'chip' guru, but... I can't see
    Compaq/IBM/HP/Dell/AST/etc *ever* using an Alpha and marketing it as a
    PC.  A Mitsubishi (being in my mind a fringe electronics company -
    atleast in N.America) may have limited success; it's up to Digital to
    push this into the mainstream, and us alone.
    
    Doug

re .4

Mitsubishi may be a 2nd tier electronics company, but Samsung is NOT!! If Samsung
gets good yields, their cost to build 21264 processor (not the wimpy
barely-beats-pentium-pro 21164PC) will be well under $400, and it will be
interesting to see if they can build a market for it. The wild card is "demand
creation", a process Digital simply is not familiar with. It's not really
Samsung's strong suit, either.

To make this work, we need to sign on a first tier PC player, e.g.
HP or Dell or Compaq or IBM. IMO, IBM's the one we should shoot for. They no
longer have a viable RISC product of their own, so they might be willing to
look at it at this point.

--RS

    re: .5
    
    >IBM's the one we should shoot for. They no longer have a viable RISC 
    >product of their own...
    
    Did I miss the complete demise of the PPC?

    But if Samsung has this great low cost product that blows away the
    competition, who needs to buy a computer from DEC?
          

    Well, we know there's not enough margin in making low-end PC's (or
    terminals or low-end printers or...) for a slow-moving company like
    Digital with its outdated business practices and systems. Nor can we
    apparently make money in now-volatile markets like memory, for the same
    reason, I guess: we're corporately unable to move at the same pace as
    our competitors. (Why?)
    
    Maybe the likes of Samsung can move at a quicker pace, given their
    background in volume consumer and industrial electronics. But can they
    design a 4100/Rawhide or an 8000/Turbolaser ? If Digital can and they
    can't, AND Digital can make money doing it, maybe that's where The
    Management wants to be. I'm sure we'll find out in due course.
    
    regards
    john

Re 4689.6: "Did I miss the complete demise of the PPC?"

What constitutes commercial success?  Do you believe PPC meets those
criteria?  Other than the relatively small Apple niche, where are the
commercially successful PPC *systems*?  Not in IBM workstations and
servers, that's certain.

If Alpha is "off the radar", then PPC is a UFO: apparently on the radar,
but not actually there.

___Pete

> criteria?  Other than the relatively small Apple niche, where are the
> commercially successful PPC *systems*?  Not in IBM workstations and
> servers, that's certain.

Well, last time I heard that "relatively small Apple niche" constituted 
somewhere around 1.2 MILLION PPC systems. Do you work in our competitive 
analysis group Pete?  ;^)

tim m.

    re .9: I keep stumbling on these PPC AIX servers all the time when
    visiting our (potential?) customers.
    
    I don't know the numberts though, neither do I know how commertcially
    succesful they are...
    

    I think it's over 5 million PPC's sold now (actually that was at the
    end of '95)... or outselling Alpha @25:1.

    Isn't the power PC used as the "brains" of all Ford automobiles?
    
    It's powerful  (measured in horsepower) and fast (measured in mph or
    kph) and portable (on road and off road).  It also drives volume.

    re: Ford: not yet.  Essentially, PPC volume = Apple volume at
    this point.

And volume for software apps to make attractive is arguably the most
important item.  (And neither Apple nor embedded systems do squat
if you believe that Windows is going to be dominant for some time -
like it or not.)

(i.e. PowerPC for embedded apps doesn't particularly help Apple.  Likewise
Alpha/VMS vs. Alpha/NT doesn't help except to share system designs and
overhead - something we seem to be ****-bent on squandering anyway with
separate consoles, boxes, etc.)

Alpha/NT with FX!32 will certainly improve the situation in terms
of making Alpha "more like" a fast, standard Windows box, but it
will still have more than its share of limitations/quirks to offset
faster performance.)

    Prior to Ford using PPC, they used Intel. Intel reportedly lost a few
    $$ on each one they sold. Multiply that up and that's a lot of $$. Not
    all of Intel makes money (but those bits that do, currently, make
    lots). 
    
    And the Ford box won't be using any OS or apps that the average
    consumer or MIS manager will recognise. It's a small part in the big
    picture, and dubious profitability long-term. 
    
    regards
    john

Suppose Apple counts 5 million PPC systems (which seems extravagant).  How
does that compare with Intel-based PC sales since volume shipment of PPC?
Obvious answer: it's peanuts.

Are your customers discussing PPC systems to run Windows NT, or are they
discussing Pentium and Pentium Pro?

Re 4689.11, "stumbling on these PPC AIX servers...": are those *PPC*
servers or *POWER* (multichip, non-PPC) servers?  Check it out.  So far as
I know, the PPC team in IBM has yet to deliver a single chip general and
powerful enough to satisfy the server-builders in IBM.  IBM workstations
and servers (like the SP2) don't use PPC.

Is Alpha a niche?  More like a scratch.

___Pete

    re .17: PPC (yes, I've opened some of them up to install H/W).
    
    I guess the numbers are low though (i.e. peanuts).
    

    re: .13
    
    PPC is [supposedly] the future in engine control systems at Ford. 
    However, it's still a ways off, and even if it does happen, it's not
    likely to the the standard off-the-shelf PPC chip, but rather some
    custom part with lotsa I/O on-chip.

I'm apparently not alone in my jaundiced view of PPC.  Take a look at this
article from "PowerPC News", 7 July 1996.  ("PowerPC News" is going out of
business -- read to the end of this -- and this is their final article.
You can get the entire article by sending a message with subject "1523" to
[email protected].)

___Pete

[...]

In March 1994, the PowerPC ecosystem was a particularly rich one.  It
promised to be an aggressively expanding one too.  At it's heart was the
very simple proposition that a new RISC architecture, would provide a
price-performance advantage so compelling that PC manufacturers would be
willing to abandon the huge Wintel application space for an entirely new
class of applications that the technology would enable.  The sheer power of
the processors would be such that the need for expensive add-on cards to
support multimedia applications would melt away.  Real-time dictation ,
intelligent agents, voice output and exciting new user interfaces would all
become the norm, all courtesy of a series of paradigm-shifting software
strategies.

IBM's leaders fell upon the PowerPC and associated software as a way to
rationalise the company's hardware and OS chowder.  All hardware divisions,
bar the System 390 guys would base future models on PowerPC technology.  A
grand operating strategy, called Workplace OS would similarly sort out the
incompatible operating systems.  A single, unified microkernel would host
the various OS personalities.  The first implementation would be OS/2.  For
those who wanted something a little more radical, IBM and Apple proposed
'Taligent' a next generation operating system, based on the same
microkernel, but built top-to-bottom from C++ objects and offering a world
where the distinction between operating system and application became
blurred.  The duo also came up with OpenDOC, a object-oriented framework
that would ease application development across the PowerPC OS's and also
extend into the Windows heartland, going up against Microsoft's Object
Linking and Embedding.  Meanwhile, there was the PowerOpen Association,
charged with giving the PowerPC community an open Unix environment based on
IBM's AIX.  It was also tasked with developing Macintosh Application
Services; that would let Apple applications run on top of this Unix.  IBM
set up a CompuServe forum to serve the users clamouring to buy the
forthcoming products from its Power Personal Systems division.
comp.sys.powerpc was quickly formed it was full steam head.

Two years on and the community has failed to flourish.  Workplace OS: dead.
Taligent as an independent force: dead.  PowerOpen dead.  OS/2 for PowerPC,
if not dead then at least in a permanent vegetative state.  Solaris on the
RS/6000: dead.  IBM's Power Personal System's Division: Dead.  IBM's
attempt to get PowerPC adopted across all its lines: died with the death of
Phil Hester's System Technology and Architecture Division.  Oh, and the
CompuServe PowerPC forum?  Just killed off by IBM.

So far, only Apple has made a real, commercial go of using the PowerPC on
the desktop.  None of those from the Wintel camp have defected.  And why
should they, given the almighty thumbs-down that IBM has given to its own
technology, its own standards?  Talk to IBMer's today about the company's
plans for PowerPC processors and the whole thing is accompanied by a whiff
of embarrassment.  The PC company is openly hostile to the processor.  The
RS/6000 division isn't touching the PowerPC 620 with a barge poll and the
AS/400 division, well true, its using the architecture, but only after the
Rochester labs built its own implementation so far removed from Somerset's
offering that any savings in terms of research of manufacturing costs
disappeared down the plug-hole.

The galling fact for the engineers at the Somerset design lab is that
(apart from the horror of the 620) they have met and indeed surpassed their
design goals, but it didn't prove sufficient.  The most important
contribution that the PowerPC has made to the IT industry to-date has been
the Intel camp's increased speed of innovation.  Sure the Pentium Pro is a
big old, power-hungry chip with a massive cache, Sure, it works poorly with
16-bit software.  The fact remains that it performs well enough that
neither system-makers nor software developers feel boxed in.  They aren't
defecting, there is no need to - yet.

That's not to say that it is time to put up the Game-Over sign and turn out
the lights.  If Apple gets its OS licensing policy right.  If Copland turns
out to be much better than Windows 95 and if CHRP/PPCP machines take off
then we could still see an interesting fight for the desktop.  But success
on the desktop essentially depends on Apple, which has enough problems of
its own.

The truth is that what promised to be a vibrant jungle of a PowerPC market
has so far turned out to be a few rather dusty plants in a parched-looking
field.  From a journalist's point of view, reporting on the latest chip
release (invariably another clock-speed hike), or profiling a new PReP
clone running Windows NT is not much fun.  Neither is it much fun having to
continually write honest analysis pieces which conclude that (for the time
being at least) PowerPC is Just Another RISC chip.  As for the
still-vibrant Macintosh market, well our compatriots in various excellent
Mac titles can follow Apple's PowerPC-related doings just as well as we
can.

So there you have; whether PowerPC News is closing due to a fundamental
weakness in the Internet publishing model or the PowerPC market is left as
an exercise for the readers.  And while you are pondering that, it is time
to move onto something new.  New, in this case, means Online Reporter, a
weekly newsletter for Internet insiders with particular pension for things
Java-esque, and related to electronic commerce.  You can get a taster of
the first issues of Online Reporter for free at

	www.computerwire.com/online/

However this is a subscription newsletter.  OR is part of the hot new
ComputerWire profile service which is being run by APT.  Computerwire lets
you build your own personal e-mail newsletter.

Finally, we's like to thank everyone who have expressed their support for
PowerPC News over the years; we hope you enjoy the new publication just as
much.