Browse Prior Art Database

Crystalline Parallel Processing Structural Architecture

IP.com Disclosure Number: IPCOM000110887D
Original Publication Date: 1994-Jan-01
Included in the Prior Art Database: 2005-Mar-26
Document File: 2 page(s) / 36K

Publishing Venue

IBM

Related People

Rodite, RR: AUTHOR

Abstract

As shown in the Figure, a structural architecture for a parallel processor system is patterned after crystal structures that exist in nature using two distinct classes of processors. Processor 10 of class A is logically located at a crystalline "atom" site, and processor 12 of class B is logically located at an "interatomic bond" site of a particular structure patterned after a crystal or molecule. The figure illustrates the relative position of the processors 10, 12 of classes A and B respectively in a sample structure. The class B processors only communicate directly with two A processors (see lines 14,16), and the A processors serve and switch as many B processors as there are bonds in the analogous crystal structure.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 100% of the total text.

Crystalline Parallel Processing Structural Architecture

      As shown in the Figure, a structural architecture for a
parallel processor system is patterned after crystal structures that
exist in nature using two distinct classes of processors.  Processor
10 of class A is logically located at a crystalline "atom" site, and
processor 12 of class B is logically located at an "interatomic bond"
site of a particular structure patterned after a crystal or molecule.
The figure illustrates the relative position of the processors 10, 12
of classes A and B respectively in a sample structure.  The class B
processors only communicate directly with two A processors (see lines
14,16), and the A processors serve and switch as many B processors as
there are bonds in the analogous crystal structure.  For example, the
diamond cubic structure would have four B processors communicating
with each A processor, and a simple cubic structure would have six.
This structure has the capability of operating in a multidimensional
environment with vast numbers of total processors while still
maintaining a simple nearest neighbor association and a simple
addressing scheme.  Neural network computing and learning are also
naturally facilitated by the B processor functioning as a highly
sophisticated synapstic element.  This architecture for processing
communication can be extended to organic molecular structures
including DNA.