Abstract
A variety of computer architectures has been developed which efficiently exploit the parallelism inherent in a problem. But all problems have a practical limit to the degree of parallelism available while maintaining computer efficiency, with the available parallelism often being fairly small. Also, no matter how much parallelism can be exploited in a given problem, the computational throughput is increased by increasing the rate at which a given process flows through logic gates. Thus increasing the serial speed of computers will improve the performance on all problems, at all levels of parallelism. It is this fast serial processor technology we seek to create.1
© 1989 Optical Society of America
PDF ArticleMore Like This
Harry F. Jordan
WA1 Optical Computing (IP) 1989
JOHN D. CROW
WJ3 Optical Fiber Communication Conference (OFC) 1989
M. Mary Eshaghian, D. K. Panda, and V. K. Prasanna Kumar
TuI28 Optical Computing (IP) 1989