Neurogrid: A Mixed-Analog-Digital Multichip System for Large-Scale Neural Simulations

By B. Benjamin, Peiran Gao, Emmett McQuinn, Swadesh Choudhary, A. Chandrasekaran, J. Bussat, Rodrigo Alvarez-Icaza, John V. Arthur, Paul A. Merolla, Kwabena A. Boahen. Proceedings of the IEEE, 2014

1,126 citations recorded by the source corpus.

Subfields: Cellular and Molecular Neuroscience, Electrical and Electronic Engineering

View Paper: https://www.semanticscholar.org/paper/336d333bcd1feb6d14f13a478534c808bf7e7e53

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cellular and Molecular Neuroscience

In Electrical and Electronic Engineering

Citations in the corpus

A reconfigurable on-line learning spiking neuromorphic processor comprising 256 neurons and 128K synapses (2015, result): “This approach is complementary to other recent ones that focus on accelerated neural simulations (Bruederle et al., 2011), or that target the real-time emulation of large populations of neurons but with no on-chip learning or adaptive behaviors at the synapse level (Benjamin et al., 2014).”
HFirst: A Temporal Approach to Object Recognition (2015, methodology): “…of these hardware implementation projects include the Integrate and Fire Array Transceiver (IFAT) [9], Hierarchical AER-IFAT [10], Brain Scales [11], Spiking Neural Network Architecture (SpiN-Naker) [12], Neurogrid [7], Qualcomm’s Zeroth Processor, and IBM’s TrueNorth [8] (fabricated with Samsung).”
Advancing Neuromorphic Computing With Loihi: A Survey of Results and Outlook (2021, background): “This has led to a long list of exotic chips over the past three decades, beginning with Mead and Mahowald’s silicon retina [8] through more recent chips impressively integrating up to 65 536 analog neurons and boards with one million neurons [9] and even some with synaptic plasticity [10].”
A Scalable Multicore Architecture With Heterogeneous Memory Structures for Dynamic Neuromorphic Asynchronous Processors (DYNAPs) (2017, methodology): “now have either restricted the space of possible network connectivity schemes to optimize bandwidth usage while minimizing power and latency [3], [12], or have designed systems that use large amounts of memory, silicon real-estate, and/or power, to maximize flexibility and programmability [4], [5].”
Large-scale neuromorphic computing systems (2016, methodology): “The main contributions of this paper are: (cid:129) A survey of the features of the brain that are modelled in neuromorphic systems ( section 2 ) ; (cid:129) a brief history of the early development of neuromorphic technology ( section 3 ) and more recent advances ( section 4 ) ; (cid:129) descripti”
Conversion of Continuous-Valued Deep Networks to Efficient Event-Driven Networks for Image Classification (2017, methodology): “Multi-layered spiking networks have been implemented on digital commodity platforms such as FPGAs (Neil and Liu, 2014; Gokhale et al., 2014), but spiking networks with more than tens of thousands of neurons can be implemented on large-scale neuromorphic spiking platforms such as TrueNorth (Benjamin ”
Skimming Digits: Neuromorphic Classification of Spike-Encoded Images (2016, background): “Prominent examples of these systems include SpiNNaker (Painkras et al., 2013), Neurogrid (Benjamin et al., 2014), and the TrueNorth processor from IBM (Merolla et al., 2014), which is capable of implementing one million neurons and 256 million synapses in real time and is reported to consumption of…”
Memory-Efficient Deep Learning on a SpiNNaker 2 Prototype (2018, background): “It dominates the required chip area (Benjamin et al., 2014; Schmitt et al., 2017) and power consumption (Schemmel et al., 2008; Akopyan et al., 2015) in special purpose hardware and the required memory and computation time ( Brette et al., 2007) in software simulations.”
Deep Learning With Spiking Neurons: Opportunities and Challenges (2018, background): “Hardware platforms such as SpiNNaker (Furber et al., 2014), Neurogrid (Benjamin et al., 2014), TrueNorth (Merolla et al., 2014), and BrainScaleS (Schemmel et al., 2010) have shown great potential to accelerate large-scale brain simulations.”
Opportunities for neuromorphic computing algorithms and applications (2022, background): “Both industry and academia have taken an interest in neuromorphic systems: in industry, some examples include IBM’s TrueNorth 23 and Intel’s Loihi 7 , and there are also a variety of academic efforts, including DYNAPs 24 , Neurogrid 25 , IFAT 13 and BrainScales-2 26 .”
NeuroSim: A Circuit-Level Macro Model for Benchmarking Neuro-Inspired Architectures in Online Learning (2018, methodology): “Several custom CMOS ASIC neuromorphic hardware accelerators have been developed in the recent years, such as BrainScaleS [4], Neurogrid [5], TrueNorth [6], Eyeriss [7], and DNPU [8].”