Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!gem.mps.ohio-state.edu!tut.cis.ohio-state.edu!pt.cs.cmu.edu!MATHOM.GANDALF.CS.CMU.EDU!lindsay
From: lindsay@MATHOM.GANDALF.CS.CMU.EDU (Donald Lindsay)
Newsgroups: comp.arch
Subject: Re: Asynchronous cpu
Keywords: references, self-timed circuits
Message-ID: <6424@pt.cs.cmu.edu>
Date: 5 Oct 89 15:12:42 GMT
References: <5241@dime.cs.umass.edu> <9320@saturn.ucsc.edu>
Organization: Carnegie-Mellon University, CS/RI
Lines: 42

In article <9320@saturn.ucsc.edu> haynes@ucscc.UCSC.EDU.UUCP 
	(Jim Haynes) writes:
>see Ivan Sutherland's Turing Award Lecture
>printed in a recent issue of CACM.

Specifically, it's

"Micropipelines"
Turing Award Lecture
Ivan E. Sutherland
Communications of the ACM, June 1989 (32, 6, 720-738)

Easy to find, it's just before the killer Literate Programming 
column.

Also, I saw a book by Dave Dill in the latest blurb from MIT Press.

My main reactions to "Micropipelines" were:
1) Sounds promising.
2) Isn't transition signalling noise-susceptible?

The Sigarch article is

"The First Asynchronous Microprocessor: The Test Results"
Martin, Burns, Lee, Borkovic, Hazewindus
Computer Architecture News, June 1989 (17, 4, 95-110)

3) Fast! Smaller than I expected, too.
4) Neato design method.
5) One reason that asynchronous systems fell into disfavor, was that
   they were often slightly wrong. They would have low-probability
   failure modes. Supposedly, the new methods give circuits which are
   guaranteed correct. So, I'm a bit worried by the line on P.96:
   "We cannot explain the phenomena yet."
6) A famous reason for synchronous systems, is that one can make all
   of the electrical noise happen at once, and arrange that no one is
   listening then. Self-timed circuits can't do this: what do they do
   instead?
7) Is it really necessary to independently self-time each and every
   line in a bus? 
-- 
Don		D.C.Lindsay 	Carnegie Mellon Computer Science