A single TrueNorth chip consists of one million digital neurons created from 5.4 billion transistors wired together. Like the human brain, these “neurons” communicate using synapses: approximately 256 million of them.
Running at 0.8 volts, a single chip consumes 70 milliwatts of power while delivering 46 giga synaptic operations per second. Sixteen of these chips will function together, representing 16 million neurons and 4 billion synapses, with a power consumption of only 2.5 watts—orders of magnitude less than a conventional computer running interference on the same neural network.
One small step for power requirements, one giant leap for next-generation supercomputers.
Cognitive Computing in Nuclear Defense
This scalable technology will be utilized in the National Nuclear Security Administration’s (NNSA) cybersecurity work and for stewardship of the nation’s nuclear deterrent and non-proliferation initiative.
The NNSA's Advanced Simulation and Computing (ASC) program will use this system to explore a variety of fields, including machine learning and deep learning algorithms and architectures. The ASC plays a major role in the NNSA's Stockpile Stewardship Program, which ensures the safety, security and reliability of the nation's nuclear deterrent without underground testing.
The stewardship potential of this technology is only enhanced by the very low power consumption the system requires to operate.
The development of TrueNorth originated from the collaboration of Defense Advanced Research Projects Agency’s (DARPA) Systems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE) program and Cornell University.
Ongoing development of this technology will continue as a collaborative effort between Lawrence Livermore National Laboratory, IBM Research, members of the Department of Energy, and various universities.
For more information, check out this IBM infographic on cognitive computing: