Abstract
This article describes some of our recent work in the development of computer architectures for efficient execution of artificial neural network algorithms. Our earlier system, the Ring Array Processor (RAP), was a multiprocessor based on commercial DSPs with a low-latency ring interconnection scheme. We have used the RAP to simulate variable precision arithmetic to guide us in the design of arithmetic units for high performance neurocomputers to be implemented with custom VLSI. The RAP system played a critical role in this study, enabling us to experiment with much larger networks than would otherwise be possible. Our study shows that back-propagation training algorithms only require moderate precision. Specifically, 16b weight values and 8b output values are sufficient to achieve training and classification results comparable to 32b floating point. Although these results were gathered for frame classification in continuous speech, we expect that they will extend to many other connectionist calculations. We have used these results as part of the design of a programmable single chip microprocessor, SPERT. The reduced precision arithmetic permits the use of multiple arithmetic units per processor. Also, reduced precision operands make more efficient use of valuable processor-memory bandwidth. For our moderate-precision fixed-point arithmetic applications, SPERT represents more than an order of magnitude reduction in cost over systems with equivalent performance that use commercial DSP chips.
Similar content being viewed by others
References
N. Morgan, J. Beck, E. Allman, and J. Beer, “RAP: A ring array processor for multilayer perceptron applications,” Proc. IEEE Intl. Conf. on Acoustics, Speech, & Signal Processing, 1990, pp. 1005–1008.
N. Morgan, J. Beck, P. Kohn, and J. Bilmes, “Neurocomputing on the RAP,” In K.W. Przytula and V.K. Prasanna, eds,Digital Parallel Implementations of Neural Networks, Englewood Cliffs, NJ: Prentice Hall, 1993, In Press.
N. Morgan and H. Bourlard, “Continuous speech recognition using multilayer perceptrons with hidden Markov models,”Proc. IEEE Intl. Conf. on Acoustics, Speech, & Processing, 1990, pp. 413–416.
H. Hermansky, “Perceptual linear predictive (PLP) analysis of speech,”J. Acoust. Soc. Am., vol. 87, April 1990.
P. Kohn, “CLONES: Connectionist Layered Object-oriented Network Simulator,” Technical Report TR-81-073, International Computer Science Institute, 1991.
T. Baker and D. Hammerstrom, “Modifications to artificial neural network models for digital hardware implementation,” Technical Report CS/E 88-035, Department of Computer Science and Engineering, Oregon Graduate Center, 1988.
M. Duranton and J.A. Sirat, “Learning on VLSI: A general purpose digital neurochip,”Proc. IJCNN-89, pp. II-613, Washington D.C., June 1989.
J.L. Holt and J.N. Hwang, “Finite precision error analysis of neural network hardware implementations,”Proc. IJCNN-91, 1991, pp. I-591-525.
D.E. Rumelhart, G.E. Hinton, and R.J. Williams, “Learning internal representations by error propagation,”Parallel Distributed Processing. Exploration of the Microstructure of Cognition, Vol 1, Cambridge: MIT Press, 1986.
P.J. Werbos,Beyond Regression: New Tools for Prediction and Analysis in the Behavioral Sciences, Ph.D. Dissertation, Dept. of Applied Mathematics, Harvard University, 1974.
K. Asanović, J. Beck, B.E.D. Kingsbury, P. Kohn, N. Morgan, and J. Wawrzynek, “SPERT: A VLIW/SIMD microprocessor for artificial neural network computations,” Technical Report TR-81-0972, International Computer Science Institute, 1991.
B.E.D. Kingsbury, K. Asanović, J. Wawrzynek, B. Irissou, and N. Morgan, “Recent work in VLSI elements for digital implementations of artificial neural networks,” Technical Report TR-91-074, International Computer Science Institute, 1991.
D. Hammerstrom, “A VLSI architecture for high-performance, low-clost, on-chip learning,”Proc. International Joint Conference on Neural Networks, 1990, pp. II-537–543.
U. Ramacher, J. Beichter, W. Raab, J. Anlauf, N. Bruls, M. Hachmann, and M. Wesseling, “design of a 1st generation neurocomputer,” inVLSI Design of Neural Networks, Boston: Kluwer Academic, 1991.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Asanović, K., Morgan, N. & Wawrzynek, J. Using simulations of reduced precision arithmetic to design a neuro-microprocessor. J VLSI Sign Process Syst Sign Image Video Technol 6, 33–44 (1993). https://doi.org/10.1007/BF01581957
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/BF01581957