A scalable custom simulation machine for the Bayesian Confidence Propagation Neural Network model of the brain
2014 (English)In: Design Automation Conference (ASP-DAC), 2014 19th Asia and South Pacific, IEEE , 2014, 578-585 p.Conference paper (Refereed)
A multi-chip custom digital super-computer called eBrain for simulating Bayesian Confidence Propagation Neural Network (BCPNN) model of the human brain has been proposed. It uses Hybrid Memory Cube (HMC), the 3D stacked DRAM memories for storing synaptic weights that are integrated with a custom designed logic chip that implements the BCPNN model. In 22nm node, eBrain executes BCPNN in real time with 740 TFlops/s while accessing 30 TBs synaptic weights with a bandwidth of 112 TBs/s while consuming less than 6 kWs power for the typical case. This efficiency is three orders better than general purpose supercomputers in the same technology node.
Place, publisher, year, edition, pages
IEEE , 2014. 578-585 p.
, Asia and South Pacific Design Automation Conference Proceedings, ISSN 2153-6961
DRAM chips;belief networks;biomedical electronics;brain;mainframes;medical computing;neural chips;neurophysiology;parallel machines;3D stacked DRAM memories;BCPNN model;Bayesian confidence propagation neural network model;custom designed logic chip;eBrain;general purpose supercomputers;human brain;hybrid memory cube;multichip custom digital supercomputer;scalable custom simulation machine;synaptic weights;technology node;Aggregates;Bandwidth;Brain modeling;Computational modeling;Delays;Random access memory;Three-dimensional displays
Electrical Engineering, Electronic Engineering, Information Engineering
IdentifiersURN: urn:nbn:se:liu:diva-114646DOI: 10.1109/ASPDAC.2014.6742953ISI: 000350791700104ScopusID: 2-s2.0-84897883326ISBN: 978-1-4799-2816-3OAI: oai:DiVA.org:liu-114646DiVA: diva2:791671
The 19th Asia and South Pacific Design Automation Conference (ASP-DAC), January 20-23, 2014, Singapore