PARALLEL ALGORITHMS AND ARCHITECTURES BASED ON PIPELINED OPTICAL BUSES

Optical signals have some unique properties, such as unidirectional pr opagation and precisely predictable path delays in waveguides, which a re not shared with their electronic counterparts. By taking advantage of these unique properties, we can use optical interconnections to ach ieve speed improvements in single-instruction stream, multiple-data st reams (SIMD) computations. We first show how optical buses can be util ized advantageously in SIMD architectures to obtain fast solutions to several computational problems, including integer addition, counting a nd logical XOR, sorting, and fast Fourier transforms. We then present a new implementation of the optical buses to meet the unique requireme nts in highperformance optical-electronic computing systems. Such an i mplementation allows the transmission of messages at speeds ideal for optics and, in the meantime, the processing of data at speeds ideal fo r electronics, dealing successfully with the speed limitation by elect ronics in optical-electronic computers. The primary effects of this bi modal optical bus are twofold: reduction of fiber lengths and reductio n of system latency. Reduced latency is a unique advantage to an optic al bimodal bus. Together, these observations make optical-bus-based ar chitectures appear to be a promising approach to SIMD processing. (C) 1995 Optical Society of America