Open AccessOptimized Simulation Framework for Spiking Neural Networks using GPU's
Author(s) -
Radu Mîrșu,
Sebastian Micut,
Cătălin Daniel Căleanu,
D. B. Mirsu
Publication year - 2012
Publication title -
advances in electrical and computer engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.254
H-Index - 23
eISSN - 1844-7600
pISSN - 1582-7445
DOI - 10.4316/aece.2012.02011
Subject(s) - computer science , general purpose computing on graphics processing units , spiking neural network , artificial neural network , cuda , artificial intelligence , computer graphics (images) , computational science , parallel computing , graphics
This paper presents a hardware accelerated model of a spiking neural network implemented in CUDA C. It does a short description of the mathematical model for the neural network and presents the implementation on the GPU. Additionally, it presents three methods of further accelerating the model by eliminating excess kernel launch overhead time, efficiently using shared memory and overlapping computation with data transfer. Finally, the implementation is benchmarked against an existing C++ equivalent model