At the Supercomputing 2012 conference last week, IBM Research-Almaden presented its next milestone towards fulfilling DARPA’s cognitive computing program, Systems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE).
DARPA’s SyNAPSE program was announced in 2008, and it calls for developing an electronic neuromorphic machine technology, which is basically a brain simulation that scales to biological levels using cognitive computing architecture with 10 billion neurons and 100 trillion synapses. These numbers are based on the number of estimated synapses and neurons in the human brain.
IBM states that it was able to accomplish this using its TrueNorth system running on the world’s second fastest operating supercomputer, the Lawrence Livermore National Lab (LBNL) Blue Gene/Q Sequoia that uses 96 racks with 1,572,864 processor cores, 1.5 PB memory, 98,304 MPI processes and 6,291,456 threads.
IBM and LBNL achieved a scale of 2.084 billion neurosynaptic cores containing 530 billion neurons and 100 trillion synapses running only 1542 times slower than in real time.
There is no biologically realistic simulation of a complete human brain, explains the abstract of the SC12 paper. “Computation (‘neurons’), memory (‘synapses’), and communication (‘axons,’ ‘dendrites’) are mathematically abstracted away from biological detail toward engineering goals of maximizing function (utility, applications) and minimizing cost (power, area, delay) and design complexity of hardware implementation.”
Join IBM’s Dharmendra Modha – Manager, Cognitive Computing Systems and Master Inventor – as he offers a glimpse of IBM Research’s efforts to develop a computer chip inspired by the human brain.