Parallel Processing Algorithm Design for Optically Interconnected Systems
The introduction of tremendous bandwidth supplied by free-space optical interconnections to computing systems is expected to solve the lack of communication capability in computing systems. Optoelectronic LSIs having optical interface with processing capability and free-space optical interconnects make it possible to fabricate optoelectronic parallel computing systems. However, although the properties of optoelectronic systems are much different from those of conventional computing systems using electronic interconnection, algorithms and applications for optoelectronic systems haven’t been discussed considering the difference of the physical layer properties. The algorithm design should be re-considered so that the merit of free-space optical interconnection is fully applied. We show a method to implement given algorithms onto optoelectronic systems that yields the minimum processing time. As an optoelectronic computing system, OCULAR-II is also shown where the proposed algorithm design method is successfully applied.
- Makoto Naruse and Masatoshi Ishikawa: Parallel Processing Algorithm Design for Optically Interconnected Systems, Transactions of Information Processing Society of Japan ,Vol.41, No.5, pp.1509-1516 (2000)