EnigmA Amiga Run 1997 July/EnigmA AMIGA RUN 20 (1997)(G.R. Edizioni)(IT)[!][issue 1997-07 & 08][EAR-CD IV].iso/lightwave/lwmlist/95.lightwave-10/000122_owner-lightwav…mail.webcom.com_Tue Oct 3 09:06:31 1995.msg

Internet Message Format | 1995-11-07 | 7KB

Received: by mail.webcom.com
    (1.37.109.15/16.2) id AA163446391; Tue, 3 Oct 1995 09:06:31 -0700
Return-Path: <[e-mail]>
Received: from mrpitt.dti.com by mail.webcom.com with SMTP
    (1.37.109.15/16.2) id AA163386365; Tue, 3 Oct 1995 09:06:06 -0700
Received: from assman (192.84.116.12) by mrpitt.dti.com id <[e-mail]>; Tue, 03 Oct 1995 10:27:53 -0500
Comments: Authenticated sender is <[e-mail]>
From: "Blaise Fanning" <Blaise [e-mail]>
Organization:  Deskstation Technology
To: [e-mail]
Date:          Tue, 3 Oct 1995 11:00:10 +0000
Subject:       RE:News About Raptor
Reply-To: [e-mail]
X-Confirm-Reading-To: [e-mail]
X-Pmrqc:       1
Priority: normal
X-Mailer: Pegasus Mail for Windows (v2.01)
Sender: [e-mail]
Precedence: bulk

> Date:          Tue, 3 Oct 1995 08:08:41 -0700
> From:          Thad Kanko <[e-mail]>
> To:            [e-mail]
> Subject:       RE:News About Raptor
> Cc:            [e-mail]

I'd like to say in advance that I'm sorry that my reply here is as
emotionally charged as it is.  I liken the situation to how some of
you animators would feel if I (who have almost no knowledge of
animation) were to come out with the position that the Blue M&Ms were
only half as good as they could have been because all the polygons
should have had 5 sides and lens flare should have been used more.
Your response would probably have been that I'm full of crap.  You
would have been dead right.  The assertions made by Thad have about
the same amount of correctness to them and his credibility as a
hardware critic is about 0 with me.  It looks from the CC as though
he's acting as a mouthpiece for Alan Kahn at Aspen Systems, since
Alan probably didn't want to appear to show up on this list only to
bash our technology.

> Well I said the 21164 300Mhz chip on the Raptor was crippled, So I will
> explain what I mean.
>

Thank you.

> was designed to have a 256 bit data path to memory. The Raptors data path is
> only 128 bits. HALF what the chip was designed for. Why did this happen?
>

This is an absolute load of crap.  There are 128 data pins going to
the chip.  I'm weelll, fairly sure of this, since it's all I've been
doing for the past nine months.  If you have a data book, you'd know
that.  If anyone you've talked to had a data book, they'd know that
as well.

> Well Dec Also needed to finish the Logic Chip Code Set, or Alcor. This chip
> handedled that path in part. Deskstation wanted to be first to the market
> with the 21164 and decided to make their own.
>

Exactly.  I know this.  I have the chipset databook.  I also have the
schematics and PC Board layout files for the designs that our
competition will be shipping, so I know *exactly* what their systems
are capable of.

> So, now you have a highspeed chip getting information from memory at half
> the speed intended. I also hear that that 128 bit path can be shaky at
> times.

This is nonsense, so I'll ignore it.

> Deskstations design only allows for 2 megs of cache, while the
> REAL chip can handle like 16 megs of cache.
>

Our design doesn't *allow* for 2 megs of cache.  It *includes* 2
megs of cache.  If you wanted to step up to buy 50 units, I'd have
them for you in about 4 weeks.  Also, any animations you're doing
will almost certainly fit inside of 2 megs of cache anyway.  While
we're at it, did you know that the 21164 designs that everyone else
will be shipping "real soon now" run their cache paths more slowly
than ours because of their modular design.  This has something like
a 5% effect on rendering times.  Oh well, I guess you can try to
make up for that by buying more cache, after all, it's only money.
Yours.

> From what I understand and have discussed with other people on the inside
> of the development game even faster speeds on the 21164 are even yet to come.
> Yet this 128 Ebit path on the Raptor is not something that can be changed
> by swaping out a card. Its on the mother board and is permanent.
>

Who are these other people in the development game?  The only people
developing technology for Alpha chips are Deskstation, Digital, and
perhaps Cray.  If you're going to tell me that you're talking about
Aspen Systems, then I'd probably be compelled to say that they really
haven't designed anything original yet, and that I'm not quite sure
why they'd start now.  The only reason that their designs don't look
*exactly* like the Digital reference designs is that they change the
placement of the components and the wires on the printed circuit
boards.  This is called *layout*, not *design*.  We've done 3 layouts
in the past two weeks.

However, you are right.  The 128 bit path is on the board and cannot
be changed.  Nonetheless, the design should be adequate for about the
next 2 generations of Alphas, which is all that I ever claimed.

By the way, Raptor users will have access to the faster grades of
Alpha up to six months before the people that are waiting for DEC to
do their work for them.  That's because we'll do *designs* based on
them, not just change the layout.

> From what I under stand the Raptor will remain the Fastest Workstation
> for about a month or 2 Max. Thats when the companies that waited for
> Dec to finish Alcor, will release their 300Mhz machines that will
> be faster and cheaper than the Raptor.
>

As I said in another message.  The other Alpha vendors will catch up
to us soon in the same sense that the back end of a caterpillar
occasionally catches up to the front half.  To say that the others
will catch up to us in a month or two minimizes the fact that we've
been shipping these machines in volume since May, which means by
your own calculations that our users have had a six to seven month
technology lead.  Also, who says that our competitors' designs will
be cheaper?  The processor clearly dominates the price of the system
and our processor prices are at least as good as the others', if not
better.  May I also add that it's far easier to price a system
cheaply if you don't actually have to ship any.  Our competitors were
probably basing their prices on forward processor pricing from DEC,
knowing that no one could hold their feet to the fire on the price if
they didn't even hint at shipping anything.

> Would you consider a animal with only half its legs crippled? I would.
>

I have no idea what this means, but I'll assume it's based on your
ignorant and ridiculous idea that the chip has 256 data pins.

Blaise Fanning
Chief Technology Officer
Deskstation Technology
[phone]
[e-mail]
--
"Blaise Fanning" <Blaise [e-mail]> sent this message.
To Post a Message           : [e-mail]
Un/Subscription Requests To : [e-mail]
(DIGEST)                 or : [e-mail]
Administrative Items To     : [e-mail]