Summary

International Symposium on Nonlinear Theory and Its Applications

2016

Session Number:A3L-G

Session:

Number:A3L-G-4

Spiking Neural Network Simulation on FPGAs with Automatic and Intensive Pipelining

Taro Kawao,  Masato Neishi,  Tomohiro Okamoto,  Amir Masoud Gharehbaghi,  Takashi Kohno,  Masahiro Fujita,  

pp.-

Publication Date:2016/11/27

Online ISSN:2188-5079

DOI:10.34385/proc.48.A3L-G-4

PDF download (232.7KB)

Summary:
There are lots of scientific interests in the behaviors of large spiking neural networks. One way to understand the behaviors is to simulate them as fast as possible with customized hardware, such as FPGA. This paper shows highly pipelined implementations of spiking neural networks on FPGA using a high level synthesis tool. Our accelerator allows 256 neurons to operate at 280 times faster than real time brain operations, which is around 8 times faster than the previous reports on similar directions. It is also designed for a multi-FPGA