Neuromorphic computing with multi-memristive synapses.
Irem Boybat, Manuel Le Gallo, S R Nandakumar, Timoleon Moraitis, Thomas Parnell, Tomas Tuma, Bipin Rajendran, Yusuf Leblebici, Abu Sebastian, Evangelos Eleftheriou
Author Information
Irem Boybat: IBM Research - Zurich, Säumerstrasse 4, 8803, Rüschlikon, Switzerland. ibo@zurich.ibm.com. ORCID
Manuel Le Gallo: IBM Research - Zurich, Säumerstrasse 4, 8803, Rüschlikon, Switzerland. ORCID
S R Nandakumar: IBM Research - Zurich, Säumerstrasse 4, 8803, Rüschlikon, Switzerland. ORCID
Timoleon Moraitis: IBM Research - Zurich, Säumerstrasse 4, 8803, Rüschlikon, Switzerland.
Thomas Parnell: IBM Research - Zurich, Säumerstrasse 4, 8803, Rüschlikon, Switzerland.
Tomas Tuma: IBM Research - Zurich, Säumerstrasse 4, 8803, Rüschlikon, Switzerland.
Bipin Rajendran: Department of Electrical and Computer Engineering, New Jersey Institute of Technology, Newark, NJ, 07102, USA. ORCID
Yusuf Leblebici: Microelectronic Systems Laboratory, EPFL, Bldg ELD, Station 11, CH-1015, Lausanne, Switzerland.
Abu Sebastian: IBM Research - Zurich, Säumerstrasse 4, 8803, Rüschlikon, Switzerland. ase@zurich.ibm.com. ORCID
Evangelos Eleftheriou: IBM Research - Zurich, Säumerstrasse 4, 8803, Rüschlikon, Switzerland.
Neuromorphic computing has emerged as a promising avenue towards building the next generation of intelligent computing systems. It has been proposed that memristive devices, which exhibit history-dependent conductivity modulation, could efficiently represent the synaptic weights in artificial neural networks. However, precise modulation of the device conductance over a wide dynamic range, necessary to maintain high network accuracy, is proving to be challenging. To address this, we present a multi-memristive synaptic architecture with an efficient global counter-based arbitration scheme. We focus on phase change memory devices, develop a comprehensive model and demonstrate via simulations the effectiveness of the concept for both spiking and non-spiking neural networks. Moreover, we present experimental results involving over a million phase change memory devices for unsupervised learning of temporal correlations using a spiking neural network. The work presents a significant step towards the realization of large-scale and energy-efficient neuromorphic computing systems.
