Bending light for multi-chip virtual PRAMs?

A new paradigm for an all-to-alloptical interconnection network is presented.
An interesting modeling aspect is that (limited)
bending of optical communication channels is
allowed. In a computer system, the paradigm
could provide part of an interconnection fabric
between several tens (e.g., 64) of chips
comprising parallel processing elements and the
first level of the cache. An optical
interconnection network raises an intriguing
possibility: obtain both improved performance
and significant cost reduction with respect to
standard serial computer system models.