Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site brl-tgr.ARPA
Path: utzoo!linus!philabs!cmcl2!seismo!brl-tgr!tgr!GUBBINS@RADC-TOPS20.ARPA
From: GUBBINS@RADC-TOPS20.ARPA (Gern)
Newsgroups: net.micro
Subject: Re: NEC V20 ---> 8088 (actually CMOS)
Message-ID: <1626@brl-tgr.ARPA>
Date: Fri, 20-Sep-85 10:09:35 EDT
Article-I.D.: brl-tgr.1626
Posted: Fri Sep 20 10:09:35 1985
Date-Received: Sun, 22-Sep-85 16:16:56 EDT
Sender: news@brl-tgr.ARPA
Lines: 25


Bipolar is the only way to go for speed.   The industry is spending too
much effort trying to make CMOS better/faster in the quest for better
thruput.  CMOS is the way to go for low power applications and is an
advantage only in circuits where most of the silicon is idle much of
the time.

In wafer-scale integration, heat disssipation becomes the limiting factor.
A 4" wafer can dissipate about 100 Watts free air, with full use of the
silicon, it will have a capacitance to ground of 3uF.  Under these conditions
CMOS tops out around 6MHz, NMOS at 17MHz and Bipolar's high transconductance
and high speed-power product tops out at 340MHz.

If industry insists on staying with CMOS technology and not working
on bipolar, we are not going to have the fantastic computing power (better
than Cray II) that we need and are capable of.

Cheers,
Gern

P.S. - Anyhow, I don't like any IC technology that you can't pick up
with your bare hands while standing on a soft carpet.  (-:


-------