Abstract

Modern, high-performance computing requires an optimum implementation of switching, communication, control, and storage functions. In the past decades research in optics has concentrated on optical means of performing these functions using, for example, optical bistability for switching, guided and free-space interconnections for communication, holographic optical clock distribution for control, and optical disk and holographic memories for storage. Thus, in principles photons can be useful in all aspects of computing. However, during the late Eighties it has been shown based on known engineering principles and existing materials, that electronics is more efficient and more flexible than optics in performing the switching function while optical interconnects are more area and power efficient than their electrical counterparts for communication over longer distances. As a result, a new interdisciplinary field, that of optoelectronic computing has emerged. Optoelectronic computing aims to optimally combine electronics and photonics to perform computing functions. Recently, engineering design studies on optoelectronic computing system architectures have shown that such a hybrid approach to computing will lead to competitive high-performance systems such as multistate interconnection networks. Similar studies are also underway to show that the parallelism and the three-dimensional (3-D) nature of some optical, storage systems can be used advantageously in optoelectronic database systems. However, for optoelectronic computing to become a viable and accepted means of computing, it is essential to make key enabling devices such as smart pixels, diffractive optical elements and 3-D parallel-accessed memories operate within specified tolerances while maintaining a reasonable manufacturing cost. Significant progress on these devices has recently been made. This talk will review some of these developments and explore their usefulness for certain computing applications.

© 1993 Optical Society of America