A high-speed analog neural processor

Targeted at high-energy physics research applications, our special-purpose analog neural processor can classify up to 70 dimensional vectors within 50 nanoseconds. The decision-making process of the implemented feedforward neural network enables this type of computation to tolerate weight discretization, synapse nonlinearity, noise, and other non-ideal effects. Although our prototype does not take advantage of advanced CMOS technology, and was fabricated using a 2.5-pm CMOS process, it performs 6 billion multiplications per second, with only 2W dissipation, and has as high as 1.5 Gbyte/s equivalent bandwidth. lthough neural networks offer excep-tionally powerful parallel computation performance, most current applica-tions focus on exploiting their leam-ing capabilities. The ability of neurai networks to learn from examples has given rise to several quite successful experiments. Those involving handwritten character recognition, speech recog-nition, and similar challenges in which biologi-cal systems overshadow artificial intelligence come to mind. Still, the benefits of unique parallel process-ing that neural networks afford warrant our atten-tion as well. With fully parallel neural hardware, processing time is independent of the data-size the network must process. Only a few comput-ing steps require serial processing, making com-putation time extremely short. The inner product computation involved does present one major challenge for realizing the hardware of neural nets. Therefore, if an application does not demand high precision, the compact, high-speed analog approach provides great advantages. Analog techniques let us create single-chip architectures of complex neural networks, fea-turing low cost and low power dissipation. Such hardware, offering processing times as low as several microseconds for as large as 128 dimen-sional input vectors, is already commercially available. ' Still, solving for application domains that demand tens of nanoseconds of processing delayzt3 for similarly large input vectors is almost impossible. The new architecture this article describes provides precisely this sort of high-computing performance, offering one solution for a number of demanding applications