Braindrop : a mixed-signal neuromorphic architecture with a dynamical systems-based programming model

Placeholder Show Content


This thesis describes the architecture of Braindrop, a .85 square-millimeter, 4096-neuron, low-power, mixed-signal neuromorphic system. Braindrop is the first such system designed with a comprehensive set of high-level programming abstractions and a synthesis procedure for mapping them to mismatched (and temperature-sensitive) subthreshold analog hardware. This high level of abstraction stands in stark contrast to previous neuromorphic systems, which required expert knowledge (and extensive characterization) of the hardware to use. Braindrop's computations are specified as coupled nonlinear dynamical systems. This program specification is synthesized to the hardware using the Neural Engineering Framework, not just compensating for, but leveraging the fabric of mismatched analog circuit elements as dynamic computational primitives. For typical network configurations, Braindrop achieves an energy per equivalent synaptic operation of 388 fJ.


Type of resource text
Form electronic resource; remote; computer; online resource
Extent 1 online resource.
Place California
Place [Stanford, California]
Publisher [Stanford University]
Copyright date 2018; ©2018
Publication date 2018; 2018
Issuance monographic
Language English


Author Neckar, Alexander
Degree supervisor Boahen, Kwabena (Kwabena Adu)
Thesis advisor Boahen, Kwabena (Kwabena Adu)
Thesis advisor Manohar, Rajit
Thesis advisor Olukotun, Oyekunle Ayinde
Degree committee member Manohar, Rajit
Degree committee member Olukotun, Oyekunle Ayinde
Associated with Stanford University, Department of Electrical Engineering.


Genre Theses
Genre Text

Bibliographic information

Statement of responsibility Alexander Neckar.
Note Submitted to the Department of Electrical Engineering.
Thesis Thesis Ph.D. Stanford University 2018.
Location electronic resource

Access conditions

© 2018 by Alexander Smith Neckar

Also listed in

Loading usage metrics...