From more@power.globalnews.com Fri Jul 22 16:02:53 1994
Received: from power.globalnews.com by zabriskie.EECS.Berkeley.EDU (4.1/1.42)
	id AA05361; Fri, 22 Jul 94 16:02:36 PDT
Received: from localhost (daemon@localhost) by power.globalnews.com (8.6.4/8.6.4) id XAA22027; Fri, 22 Jul 1994 23:02:04 GMT
Date: Fri, 22 Jul 1994 23:02:04 GMT
From: more@power.globalnews.com
Message-Id: <199407222302.XAA22027@power.globalnews.com>
To: burd@zabriskie.EECS.Berkeley.EDU
Subject: 1123 MICROPROCESSOR REPORT ARGUES FOR NEW INSTRUCTION SET
Content-Length: 3319
X-Lines: 57
Status: RO

MICROPROCESSOR REPORT ARGUES FOR NEW INSTRUCTION SET
In the face of the new Intel-HP alliance, the Microprocessor Report
has delivered itself of the argument that maybe somebody out there
should come up with another chip architecture. It reasons that with
the exception of the PowerPC, none of the current RISC platforms has
the software support or the money to compete against the on-coming
post-RISC Intel-HP chip -  which for lack of name supplied by its
putative parents the newsletter has taken to calling the P86. It says
it's likely most PC vendors will follow Intel. 

Sun, it says, "will need to abandon Sparc at some point, possibly for
PowerPC" and other RISC vendors could go either P86 or PowerPC. Lest
the industry be left with only one or two choices, editor Linley
Gwennap says a "third architecture could be created by joining
Compaq, which clearly dislikes having a single CPU source, with the
MIPS team, bringing in NEC, IDT and Toshiba as multiple chip sources.
Silicon Graphics would provide CPU design expertise as well as system
products complementary to Compaq's. NEC is the leading Japanese
system vendor. The new architecture could offer compatibility with
both MIPS and x86. To assist with the latter, AMD's design skills
would be useful; AMD is also a CPU vendor that Compaq is comfortable
with. 

Although this scenario is just one possible outcome," Gwennap claims
that "pressures are building to cause such a radical industry
realignment." Gwennap's argument assumes that Intel is really
dedicated to replacing x86 with the P86 and that five or 10 years
from now there'll be real volume demand for chips that deliver
thousands of MIPS. Those two assumptions in mind, he reckons x86
vendors have only a limited window left. Once the P86 appears, the
x86 market, including Pentium and still-to-be-seen P6, will
inexorably decline and vanish by the year 2002, eight years from now.


Intel will probably make the alternative of building P86-compatible
processor nigh onto impossible both by keeping portions of the new
architecture a secret and by surrounding it with patents. As for the
RISC camp, well, in Gwennap's view only PowerPC has a chance. The
others he consigns to the purgatory of low-volume markets unable to
attract ISVs to support them or major systems vendors to sell them.
It is also unclear to him whether the RISC folk will be able to keep
up with the post-RISC P86 design though they may be able to keep
abreast by putting multiple processors on a chip. To compete with
post-RISC designs of their own would require heavy investment and
probably the support of major systems vendors who in turn would
attract the ISVs. Gwennap isolates four companies that can make a big
difference to a fledgling architecture: Compaq, DEC, Sun and NEC.
"These companies," he says, "can stay with their current processors
and risk becoming uncompetitive in the long run. They can adopt P86
and lock themselves into a single chip vendor. They can move to
PowerPC, but that architecture may not provide compatibility with
these vendors' current software base." 

DEC, he reckons won't do it. Sun and Compaq could form a powerful
third axis but Sun, he thinks, is headed towards PowerPC leaving
Compaq to team with SGI and the MIPS vendors.
(c) Unigram.x | Select story 5001 for more details