Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!cbatt!ihnp4!inuxc!pur-ee!uiucdcs!uxc.cso.uiuc.edu!uicsrd!mcdaniel
From: mcdaniel@uicsrd.UUCP
Newsgroups: comp.arch
Subject: Re: Tightly-coupled, multi-processor UN
Message-ID: <43700016@uicsrd>
Date: Sat, 14-Mar-87 01:59:00 EST
Article-I.D.: uicsrd.43700016
Posted: Sat Mar 14 01:59:00 1987
Date-Received: Sun, 15-Mar-87 23:48:04 EST
References: <280@cg-d.UUCP>
Lines: 46
Nf-ID: #R:cg-d.UUCP:280:uicsrd:43700016:000:1918
Nf-From: uicsrd.CSRD.UIUC.EDU!mcdaniel    Mar 14 00:59:00 1987


> Illinois is the place you want! :-) Have you heard of the 
> Cedar multiprocessor? See e.g., the 1985 & 1986 Proceedings of ICPP

[ICPP := the International Conference on Parallel Processing, held in
scenic St. Charles, Illinois.]

"Senator, that happens not to be the case."

First: from the base note,
>> For the last three or four years, I've heard rumors of a very
>> successful multi-processor UNIX system implemented at some (sub)set of
Cedar does not exist yet.  Of course, when it IS built, it WILL be
wildly successful.  [The dolts on the net force me to add a :-), the
electronic laugh track.]

Second:
>> to me, one processor ran the kernel, having total access to the
>> internal data structures.  All system calls, from all processes on all
>> processors, were directed to the designated kernel processor.
Cedar will not have a master processor or cluster.  The OS
will be distributed over all the Cedar clusters, and all the parts
will communicate as peers.

Third:
>> independent caches but operating from a common physical memory address
>> space and a common set of peripherals.
Third: Cedar will have a global memory, but each cluster will have its
own cluster memory.  Furthermore, the memory map in various clusters
(various processors, perhaps?) may differ.  Thus, Cedar will not have
a common address space, in that a given address x in one cluster has
no necessary relationship with address x in another cluster.

Presently, Purdue people put progress papers in ICPP proceedings,
though.  Check the last few years's proceedings.  (I thinks the
machine is called PASM.)

--
Tim, the Bizarre and Oddly-Dressed Enchanter
Center for Supercomputing Research and Development
at the University of Illinois at Urbana-Champaign

UUCP:	 {ihnp4,seismo,pur-ee,convex}!uiucdcs!uicsrd!mcdaniel
ARPANET: mcdaniel%uicsrd@a.cs.uiuc.edu
CSNET:	 mcdaniel%uicsrd@uiuc.csnet
BITNET:	 mcdaniel@uicsrd.csrd.uiuc.edu