Analog Encoded Neural Network for Power Management in MPSoC - Archive ouverte HAL Accéder directement au contenu
Article Dans Une Revue Analog Integrated Circuits and Signal Processing Année : 2014

Analog Encoded Neural Network for Power Management in MPSoC

Résumé

Encoded neural networks combine the principles of associative memories and errorcorrecting decoders so that they are good candidates to solve problems where decisions have to be made based on partial input information. This paper introduces an analog implementation of this new type of network to manage the power distribution in a Multiprocessor System-on-Chip(MPSoC). The proposed circuit has been designed for the 1V supply ST CMOS 65nm process, with a low complexity and low power consumption (less than 1% of the MPSoC power). Compared to a digital counterpart based on game theory, this analog solution consumes 6800 times less energy and reacts 4500 times faster. Thus, this analog circuit allows to fully exploiting DVFS circuits switching capabilities to continuously adapt the power distribution of an MPSoC. From a given energy budget, GT saves 38% while the analog ENN saves 60%.
Fichier non déposé

Dates et versions

hal-01170221 , version 1 (01-07-2015)

Identifiants

Citer

Benoit Larras, Bartosz Boguslawski, Cyril Lahuec, Matthieu Arzel, Fabrice Seguin, et al.. Analog Encoded Neural Network for Power Management in MPSoC. Analog Integrated Circuits and Signal Processing, 2014, 81 (3), pp.595 - 605. ⟨10.1007/s10470-014-0420-z⟩. ⟨hal-01170221⟩
189 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More