China
Africa
Explore chapters and articles related to this topic
Neuromorphic Platforms Comparison
Published in Paul R. Prucnal, Bhavin J. Shastri, Malvin Carl Teich, Neuromorphic Photonics, 2017
Paul R. Prucnal, Bhavin J. Shastri, Malvin Carl Teich
The neuromorphic computing community has been making vigorous efforts toward large-scale spiking neuromorphic hardware, e.g., Heidelberg HICANN chip via the FACETS/BrainScaleS projects [7] IBM TrueNorth via the DARPA SyNAPSE program [8], Stanford’s Neurogrid [9] and SpiNNaker [10] (Fig. 14.2). Many researchers are concentrating their efforts towards the long-term technical potential and functional capabilities of this hardware compared to standard digital computers.
Thermodynamic-RAM technology stack
Published in International Journal of Parallel, Emergent and Distributed Systems, 2018
M. Alexander Nugent, Timothy W. Molter
Given the success of recent advancements in machine learning algorithms combined with the hardware power dilemma, an immense pressure exists for the development of neuromorphic computer hardware. The Human Brain Project and the BRAIN Initiative with funding of over EUR 1.190 billion and USD 3 billion respectively partly aim to reverse engineer the brain in order to build brain-like hardware [5,6]. DARPA’s recent SyNAPSE program funded two large American tech companies IBM and HP as well as research giant HRL labs and aimed to develop a new type of cognitive computer similar to the form and function of a mammalian brain. The recent Nanotechnology-Inspired Grand Challenge for Future Computing in the United States [7] was formed to ‘Create a new type of computer that can proactively interpret and learn from data, solve unfamiliar problems using what it has learned, and operate with the energy efficiency of the human brain.’ CogniMem is commercialising a k-nearest neighbour application specific integrated circuit (ASIC) for pattern classification, a common machine learning task found in diverse applications [8]. Stanford’s Neurogrid, a computer board using mixed digital and analog computation to simulate a network, is yet another approach at neuromorphic hardware [9]. Manchester University’s SpiNNaker is another hardware platform utilising parallel cores to simulate biologically realistic spiking neural networks [10]. IBM’s neurosynaptic core and TrueNorth cognitive computing system resulted from the SyNAPSE program [11]. All these platforms have yet to prove utility along the path towards mass adoption, and none have yet solved the foundational problem of memory-process separation.