1996-01-10 - RISC not everything

Header Data

From: tcmay@got.net (Timothy C. May)
To: cypherpunks@toad.com
Message Hash: 7d9f8ed7c863fdecb6264ce26fb6796ecb0d045ed111176a3fcc079849767ee7
Message ID: <ad185be407021004bb85@[205.199.118.202]>
Reply To: N/A
UTC Datetime: 1996-01-10 02:13:34 UTC
Raw Date: Wed, 10 Jan 1996 10:13:34 +0800

Raw message

From: tcmay@got.net (Timothy C. May)
Date: Wed, 10 Jan 1996 10:13:34 +0800
To: cypherpunks@toad.com
Subject: RISC not everything
Message-ID: <ad185be407021004bb85@[205.199.118.202]>
MIME-Version: 1.0
Content-Type: text/plain


At 8:24 PM 1/9/96, Kent Dahlgren wrote:

>better way to tell them?  Maybe I'm just paranoid.  Its just that I kind
>of feel sorry for DEC; its not easy being burdened with the worst
>marketing staff in the world, having the world's fastest RISC processor,
>and having the media go wild over the P6.

I'm also a fan of Alta Vista, and use it daily. And I'd love to have a DEC
Alpha workstation.

However, there's more to success than being "the world's fastest RISC
processor," as history has shown for the past decade or so. (Amongst other
things, the SPECInts for the Alpha are actually lower than for the P6,
though SPECfps are higher. And some of the MIPS/SGI processors are about as
powerful as either.)

The various high-end Alphas have a high per-chip cost. Very high. (Low-end
Alphas are cheaper, but mainly for good reasons...a low-end Alpha is not
compelling.) The high per-chip cost is associated with the large die size,
DEC's lack of volume in making chips (which largely determines chip
yields), the "mostly clock" layout (the 300 MHz clock is hard to distribute
across the entire die area, and DEC uses a considerable fraction of the
chip area and metallization in distributing the clock without significant
skew), and the architecture.

I'm not a P6 expert (nor do I even own or use any Intel processor machines,
save for an old laptop, a first-generatino IBM PC, and a 1978-era Sol), but
my friend John Wharton has written glowingly of the P6 architectural
innovations in the P6.

In any case, Intel has the manufacturing machine able to make Pentiums in
large enough volume for low enough cost to be a major market force. DEC
does not have the same advantages.

There are of course lots of issues to consider. If NT is as successful as I
think it will be (see, I'm not _only_ a basher of Microsoft!), and if the
versions of software for NT will not require extensive tuning for various
platforms, then I think Intel's dominance will be slightly weakened.
However, Intel is not standing still--it's busy building several new fabs
that each cost more than a billion dollars (including one that will cost $2
B). Its "P7" processor is far along in development, and reportedly will
merge today's features with "very long instruction word" (VLIW) techniques.

DEC is back to making profits, but it sure wasn't for several years while
it coasted on the work done earlier on the VAX.

--Tim May

We got computers, we're tapping phone lines, we know that that ain't allowed.
---------:---------:---------:---------:---------:---------:---------:----
Timothy C. May              | Crypto Anarchy: encryption, digital money,
tcmay@got.net  408-728-0152 | anonymous networks, digital pseudonyms, zero
W.A.S.T.E.: Corralitos, CA  | knowledge, reputations, information markets,
Higher Power: 2^756839 - 1  | black markets, collapse of governments.
"National borders aren't even speed bumps on the information superhighway."









Thread