Abstract

Folded optical implementations of interconnection networks are desired if, in the wire grid model of computation for VLSI, the networks require more than O(N) area or time. The strategy behind folded interconnects is both to reformat a long 1-D sequence and to transform the required 1-D operations into 2-D, so that the third spatial dimension available in free-space optical systems can be used to perform the interconnection operations.

© 1988 Optical Society of America