DocumentCode :
3431182
Title :
Arithmetic for digital neural networks
Author :
Zhang, D. ; Jullien, G.A. ; Miller, W.C. ; Swartzlander, Earl, Jr.
Author_Institution :
Dept. of Electr. Eng., Windsor Univ., Ont., Canada
fYear :
1991
fDate :
26-28 Jun 1991
Firstpage :
58
Lastpage :
63
Abstract :
The implementation of large input digital neurons using designs based on parallel counters is described. The implementation of the design uses a two-cell library, in which each cell is implemented using switching trees which are pipelined binary trees of n-channel transistors. Results obtained from initial switching trees realized with a 3-μm CMOS process indicate that the design is capable of being pipelined at 40 MHz sample rates, with better performance expected for more advanced technologies. It appears feasible to develop a wafer-scale implementation with 2000 neurons (each with 1000 inputs) that would perform 3×1012 additions/s
Keywords :
CMOS integrated circuits; digital arithmetic; neural nets; CMOS process; arithmetic; digital neural networks; n-channel transistors; parallel counters; pipelined binary trees; sample rates; switching trees; two-cell library; wafer-scale implementation; Binary trees; Computer architecture; Counting circuits; Digital arithmetic; Libraries; Neural network hardware; Neural networks; Neurons; Transistors; Very large scale integration;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computer Arithmetic, 1991. Proceedings., 10th IEEE Symposium on
Conference_Location :
Grenoble
Print_ISBN :
0-8186-9151-4
Type :
conf
DOI :
10.1109/ARITH.1991.145534
Filename :
145534
Link To Document :
بازگشت