Parallel Computation of Data Summation for Multiple Problem Spaces on Partitioned Optical Passive Stars Network

In Partitioned Optical Passive Stars POPS network, nodes and couplers become free after slot to slot in some computation. It is necessary to efficiently utilize free couplers and nodes to be cost effective. Improving parallelism, we present the fast data summation algorithm for multiple problem spaces on POPS(g, g) with smaller number of nodes for the case of d = root n = g. For the case of d > root n > g, we simulate the calculation of large number of data items dedicated to larger system with many nodes on smaller system with smaller number of nodes. The algorithm is faster than the best know algorithm and using smaller number of nodes and groups make the system low cost and practical.