Analogic computing: system aspects of analogic CNN sensor computers

A new principle of computing and computers is emerging: the analogic cellular computer. Its architecture, the Cellular Neural Net (CNN) Universal Machine, is now implemented in several different physical forms and the first practical experiments exhibit very good frame rate and computing power. Several "Kilo real-time video" frame rate (more than 10000 frames per second) and TeraOPS computing power on a 1 cm/sup 2/ CMOS (0.5 micron) chip were measured. In this review article, the systems aspects and the new directions in this field are considered. A new world of software and a new notion of computing are taking ground. The possibility of software in optical computing becomes feasible. Likewise, programming on atomic and molecular scale implementations may be possible. Hence, photons and molecules may be used for signal representation in these analogic computers.