Path: utzoo!attcan!uunet!lll-winken!lll-lcc!ames!xanth!ukma!psuvm.bitnet!cunyvm!nyser!cmx!goedel!hamid
From: hamid@goedel.uucp (Hamid Bacha)
Newsgroups: comp.arch
Subject: Re: Content Addressible Memories
Message-ID: <1036@cmx.npac.syr.edu>
Date: 26 Jan 89 03:22:51 GMT
References: <12371@srcsip.UUCP> <367@enint.Wichita.NCR.COM> <6694@pucc.Princeton.EDU> <13308@srcsip.UUCP>
Sender: usenet@cmx.npac.syr.edu
Reply-To: hamid@logiclab.cis.syr.edu (Hamid Bacha)
Organization: Logic Lab, CIS Dept., Syracuse University
Lines: 57

During the month of December, there was an interesting discussion
concerning Content-Addressable Memories. With regard to some possible
uses of the CAM, there was this exchange:

In article <367@enint.Wichita.NCR.COM>, lpelleti@enint.Wichita.NCR.COM (Larry
Pelletier) writes:
>There was never a "popular" programming paradigm that could really take
>advantage of content addressable memories.  Although some things could be
>done, it really didn't help the concepts of data and code in languages such
>as Fortran, Cobol, Pascal, or C.

and

In article <13308@srcsip.UUCP>, shankar@src.honeywell.COM (Son of Knuth) writes:
>That's probably correct with procedural languages, but how about languages
>like Prolog or other logic languages?  I believe that the use of CAMs would
>lead to significantly greater performances then the use of the more
>traditional indexing schemes for rule unification and retrieval, despite
>problems with inherently sequential portions of the problem.  CAMs can also
>be used for maintenance of the binding environments.

I recently developed a Prolog Abstract Machine for content-addressable
memory (along the lines of the Warren Abstract Machine).  This new abstract
machine is shown to be superior to the WAM in many respects. It is discussed
in detail in a paper that was submitted to the next International Conference
on Logic Programming (Lisbon '89). It has also been published as a technical
report by the CASE Center (New York State Center for Advanced Technology in
Computer Applications and Software Engineering) at Syracuse University.
Anybody interested in obtaining a free copy of this report should contact
Donna McCammon at (315)-443-1066 or write to her at:

        CASE Center
        120 Hinds Hall
        Syracuse University
        Syracuse, NY 13210

The technical report number is: 8816
and the title is: "Beyond the WAM: A PAM for the CAM" (a Prolog Abstract
Machine for Content-Addressable Memory).

Following is a brief list of the advantages that this abstract machine
model enjoys over the Warren Abstract Machine:

 *When binding a variable, there is no need to check whether it should
        be trailed.
 *No trailing means no resetting of variables on failure, and no trail to
        manage.
 *No need to check for unsafe variables because there are none.
 *No read or write mode when dealing with lists and structures.
        This eliminates the S register.
 *No need to initialize unbound variables when a structure is built.
        Arguments of structures are accessed by position and their absence
        from the CAM indicates that they are unbound.
 *Fewer number of instructions means a less complex compiler and a smaller
        emulator. The 26 GET, PUT and UNIFY instructions from the WAM are
        replaced by 15 new instructions that are no more complex than
        their WAM counterparts.