Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!emory!hubcap!moskowit
From: moskowit@paul.rutgers.edu (Len Moskowitz)
Newsgroups: comp.parallel
Subject: Re: Research for RETE-based systems on Intel hypercube
Message-ID: <12464@hubcap.clemson.edu>
Date: 3 Jan 91 17:56:12 GMT
References: <12437@hubcap.clemson.edu>
Sender: fpst@hubcap.clemson.edu
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 20
Approved: parallel@hubcap.clemson.edu

Though they're not working specifically on the Hypercube, Sal Stolfo
at Columbia University and Anurag Acharya at CMU have been doing much
of the work on parallelizing OPS5-type languages.  

We at Bendix have CMU's CParaOPS5 running here on our Suns and with a
little luck it'll soon be running on our new 8 node Tadpole 88000 RISC
multi-processor.  CParaOPS5 uses a truly parallel Rete.

We've spoken recently with Stolfo and his group is doing very
interesting work on changing OPS5's semantics to allow truly parallel
programming.  Might pay to contact him.

I'd like to hear what you're doing.  Have you looked into other
approaches than parallel Rete, such as rule level parallelism or
multiple independent rule systems?

========================================================================
Len Moskowitz Allied-Signal Aerospace Bendix Test Systems Division
Teterboro, NJ 07608
moskowitz@bendix.com moskowitz%TSD@atc.bendix.com moskowit@paul.rutgers.edu