DocumentCode :
1997600
Title :
Evaluating on-chip interconnects for low operating frequency silicon neuron arrays
Author :
Cassidy, Andrew ; Murray, Thomas ; Andreou, Andreas G. ; Georgiou, Julius
Author_Institution :
Dept. of Electr. & Comput. Eng., Johns Hopkins Univ., Baltimore, MD, USA
fYear :
2011
fDate :
15-18 May 2011
Firstpage :
2437
Lastpage :
2440
Abstract :
We present a quantitative analysis of the limits of the time-multiplexed Address Event Representation (AER) bus for on-chip connectivity of silicon neuron arrays. In particular, we evaluate its potential to support high density and low power neural arrays operating in the subthreshold regime. Our analysis shows that due to low clock frequencies when operating in the subthreshold regime, the traditional single AER bus does not scale to large neural arrays. We find that a switched mesh network improves scalability, however, a crosspoint architecture overcomes the bandwidth limitations altogether. By trading off area for improved performance, it increases the number of neurons that can be supported in a single chip neural array.
Keywords :
elemental semiconductors; integrated circuit interconnections; low-power electronics; neural chips; silicon; Si; bandwidth limitation; crosspoint architecture; low clock frequency; low operating frequency silicon neuron array; low power single chip neural array; on-chip interconnect evaluation; quantitative analysis; switched mesh network; time-multiplexed AER bus; time-multiplexed address event representation bus; Bandwidth; Computer architecture; Integrated circuit interconnections; Neurons; Routing; Silicon; Wires;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Circuits and Systems (ISCAS), 2011 IEEE International Symposium on
Conference_Location :
Rio de Janeiro
ISSN :
0271-4302
Print_ISBN :
978-1-4244-9473-6
Electronic_ISBN :
0271-4302
Type :
conf
DOI :
10.1109/ISCAS.2011.5938096
Filename :
5938096
Link To Document :
بازگشت