A coprocessor for fast searching in large databases: Associative Computing Engine
FacultiesFakultät für Ingenieurwissenschaften und Informatik
LicenseStandard (Fassung vom 03.05.2003)
As a matter of fact, information retrieval has changed considerably in recent years with the expansion of the Internet and the advent of always more modern and inexpensive mass storage devices. Due to the enormous increase in stored digital contents, multi-media devices must provide effective and intelligent search functionalities. However, in order to retrieve a few relevant kilobytes from a large digital store, one moves up to hundreds of gigabytes of data between memory and processor over a bandwidth-restricted bus. The only long-term solution is to delegate this task to the storage medium itself. For that purpose, this thesis describes the development and prototyping of an associative search coprocessor called ACE: an Associative Computing Engine dedicated to the retrieval of digital information by means of approximate matching procedures. In this work, the two most important features are the overall scalability of the design allowing the support of different bit widths and module depths along the data path, as well as the obtainment of the minimum hardware area while ensuring the maximum throughput of the global architecture.
Subject HeadingsDigitalschaltung [GND]
Field programmable gate array [GND]
Text Mining [GND]