more on Apple to ditch IBM, switch to Intel chips | CNET News.com

[David Farber <dave () farber net>]

Begin forwarded message:

From: "Jonathan S. Shapiro" <shap () eros-os org>
Date: June 5, 2005 1:28:56 PM EDT
To: dave () farber net
Cc: Ip ip <ip () v2 listbox com>
Subject: Re: [IP] more on Apple to ditch IBM, switch to Intel chips |  
CNET News.com


On Sun, 2005-06-05 at 09:28 -0400, David Farber wrote:

> Just a few details that seem to have been missed in this discussion
> so far --
>
> 1) Sun is certainly not abandoning SPARC.  The 90nm US IV+ will be in
> the market later this year and "Niagra" shows up in early 2006 to be
> followed by new Fujitsu designs, etc...

Just to clear the air, I did not mean to suggest that Sun will cease
production of SPARC chips this year or next. What I do expect is that
Sun will formally discontinue SPARC development within the next 4-5
years, and will begin transitioning customers to AMD64 comfortably
before that timeframe. When the announcement will come is a bit up in
the air. Sun certainly will *not* abandon existing customers, and will
do everything possible to continue to support them.

One reason for this is that Sun's attempts to develop high performance
SPARC chips have failed.

For obvious reasons, I won't disclose my sources, but it should be
suggestive to Sun watchers that 100% of Andy Bechtolsheim's attention
has remained on AMD64-based machines after his company was acquired by
Sun. I would cite the name of the emulator company that supplied the
dynamic translation technology, but I'm not entirely sure that this has
been announced publicly yet.

Aside from the issue of application compatibility (which is very
important), the SPARC architecture has become a significant drain on
Sun's financial resources at a time when they cannot afford to continue
to invest in a dead end design. The benchmark numbers for SPARC-based
designs are quite poor. In any given implementation technology, if you
evaluate processors on MIPS per transistor OR on the performance of the
fastest available part, the SPARC simply isn't competitive. Lately it
has trailed the pack. So turn Mark Stahlman's statement around: if Sun
continues to invest in a part that exhibits a factor of 2 to 4 loss in
relative performance against competitors, why on earth would anybody
continue to buy their products? Why would analysts support this drain on
the company's resources when a transitionable alternative exists? Why
would shareholders?

Sun's sustainable advantages at this time lie in a very good I/O
subsystem architecture and (arguably) the most robust and mature UNIX
variant available. SPARC is a very expensive distraction, and the
company can no longer afford to indulge it.

Scott McNealy is fond of telling Sun employees that Sun's cash on hand
exceeds debt. This simply wasn't true the last time he said it -- I
looked at the SEC filings and checked. The accurate statement was that
Sun's cash on hand exceeded their **short term** debt. When a company's
revenue is falling and its credit rating comes into question, the
difference becomes significant.

When you have a part that is both financially and technically
unsustainable, it's a safe bet that the part is doomed. The only
question is when.


Jonathan Shapiro
Department of Computer Science
Johns Hopkins University



-------------------------------------
You are subscribed as lists-ip () insecure org
To manage your subscription, go to
 http://v2.listbox.com/member/?listname=ip

Archives at: http://www.interesting-people.org/archives/interesting-people/