High Performance Associative Coprocessor Architecture for Advanced Database Searching

C. Layer and H.-J. Pfleiderer (Germany)

Keywords

Bit Attribute Matrix, Associative Access Database

Abstract

This paper presents a hardware solution for the realization of an intelligent information retrieval system: the ACE (As sociative Computing Engine). The technology behind this is based upon the computation of text characteristic signa tures, which are used to recover, in a stated quantity, the most similar documents in a coded database. Derived from a textual database paradigm known as approximate match ing, the ACE resorts to a very flexible and fault tolerant search algorithm using associative computing to improve memory access capacity. Its architecture has been imple mented in an FPGA (Field Programmable Gate Array) as a search coprocessor communicating with the host computer through a PCI (Peripheral Component Interconnect) inter face. To achieve speeds required by today's applications, many kinds of media have been tested for the storage of a high throughput database and are discussed in this paper.

Important Links:



Go Back