An optical content-addressable parallel processor for fast searching and retrieving

Associative processing based on content-addressable memories has been argued to be the natural solution for non-numerical information processing applications. Unfortunately, the implementation requirements of these architectures using conventional electronic technology have been very cost prohibitive, and therefore associative processors have not been realized. Instead, software methods that emulate the behavior of associative processing have been promoted and mapped onto conventional location-addressable systems. This however, does not bring about the natural parallelism of associative processing, namely the ability to access many data words simultaneously. The inherently parallel nature and high speed of optics, combined with the recent technological advancements in optical logic, storage and interconnect devices are raising hopes for practical realization of highly parallel optical computing systems. This paper presents the principles of designing an optical content-addressable parallel processor, called OCAPP, for the efficient support of high speed symbolic computing. The architecture is designed to fully exploit the parallelism an high speed of optics. Several parallel algorithms are mapped onto OCAPP in bit-parallel as well as word-parallel fashion, resulting in efficient symbolic algorithms with execution times dependent only on the precision of the operands and not on the problem size. This makes OCAPP very suitable for applications where the number of data sets to be operated on is high e.g., massively parallel processing. A preliminary optical implementation of the architecture using currently available optical components is also presented.