TrueNorth – Next Generation Cognitive Computing
In August, members of the Emerging Technologies team joined with a group of over 60 researchers from academia and government at the Silicon Valley Lab in San Jose California, for an in depth learning experience about IBM’s new neurosynaptic processing chip called TrueNorth. The three week bootcamp was organised by the Almaden Research team headed by IBM Fellow Dharmendra Modha. The purpose was to bring together a community of researchers and early adopter customers to learn about the capabilities of this exciting new technology and to begin exploring novel applications for this next generation biologically inspired computing capability.
The TrueNorth Chip
The TrueNorth chip is the result of the ambitious mission of the DARPA sponsored SyNAPSE (Systems of Neuromorphic Adaptive Plastic Scalable Electronics) project, launched in early 2009 to create a new chip with a brain-inspired computer architecture. TrueNorth has one million neurons and 256 million synapses. It is today the second largest CMOS chip IBM has ever built at 5.4 billion transistors, and has an on-chip network of 4,096 neurosynaptic cores. Yet, it only consumes 70mW during real-time operation — orders of magnitude less energy than traditional chips — and is capable of 46 billion synaptic operations per second, per watt–literally a synaptic supercomputer in your palm.
The bootcamp consisted of a series of seminars and workshops giving hands on experience of how to program the TrueNorth chip and investigating various ways in which it can be used for applications such as image and acoustic processing and text analytics. Further details can be found in the following recently published Wired Magazine article.
The prize for attending the course was that each participating institution was given the loan of one of the 48 development boards shown above. Each development board (see below) comprises a TrueNorth processor and a Xilinx ZINQ processor with a Linux operating system and a wide range of connectors and on-board sensors. The goal is to make the technology widely available creating a new research community that will explore the full potential of this new technology and rapidly develop the next generation of neuromorphic applications.
The ET team in the UK are investigating the use of the technology for a wide range of applications including autonomous drones, micro-satellites, sports video analysis and text processing. We are also investigating the use of the technology for performing distributed analytics in a ‘Fog Computing’ scenario.
Our role is to make this emerging technology available to commercial customers and to this end we are putting together an offering that includes basic and advanced training courses that will use a cloud based development environment with a full simulation capability and the capability to run neuromorphic programs on the hardware. Customers should initially be able to purchase development boards and eventually larger scale multi-chip systems.