High throughput optical algorithms for the FFT and sorting via data packing

This paper demonstrates that the potential for very high speed computation provided by optical technology can be achieved for important computational problems including the fast Fourier transform and sorting. First a programming model that captures the 'time-of-flight' characteristics of optical fibers and switching elements is presented. Then it is shown that the FFT and the radix sort algorithms can both be reduced to the computational kernel of 'packing' intermediate data results that are continuously moving through an optical fiber. New algorithms are developed for the general packing problem, and the details of how to implement them in optics are presented. The resulting systems have the potential for operating at clock rates that are two or more orders of magnitude higher than conventional computers.