The promise of digital optical computing is based on massively parallel interconnections between logic gates, which allow for novel architectures, and the possibility of ultrafast switching devices. This paper spells out the computational requirements and limitations for non-linear optical devices and optical interconnects. Relationships between the optical properties of devices (transmission and contrast) and their potential computational properties (fanin and fanout) are derived. The accuracy of the intensity levels required in the system are estimated. The requirements for a minimal device useful for digital optical computing are stated. The ‘volume’ of a device in phase-space limits fanin, switching energy and the degree of space variance in the interconnections. Space-invariant and space-variant interconnections are compared. Limits of random interconnects by volume holograms are discussed.
[1]
Nasser N Peyghambarian,et al.
Optical Bistability III
,
1986
.
[2]
Karl-Heinz Brenner,et al.
Digital optical computing with symbolic substitution.
,
1986
.
[3]
M J Murdocca.
Digital optical computing with one-rule cellular automata.
,
1987,
Applied optics.
[4]
T Yatagai,et al.
Cellular logic architectures for optical computers.
,
1986,
Applied optics.
[5]
Y Ichioka,et al.
OPALS: optical parallel array logic system.
,
1986,
Applied optics.
[6]
A. Huang,et al.
Architectural considerations involved in the design of an optical digital computer
,
1984,
Proceedings of the IEEE.