The Spiking Neural Network Architecture (SpiNNaker) supercomputer is capable of performing more than 200 million actions per second. Each of its multiprocessor chip is equipped with 100 million transistors.
The creators of this machine took 20 years to develop a concept, 10 years to assemble, and another 15 million pounds sterling. The project was initially funded by the Engineering and Physical Sciences Research Council (EPSRC), and now it is also supported by the European Human Brain Project (HBP). The supercomputer was first included on Friday, November 2.
SpiNNaker, developed and assembled at the School of Computer Science at the University of Manchester, can simulate more biological neurons in real time than any other machine on the planet.
This supercomputer is unique because, unlike conventional computers, it does not transfer large amounts of information from point A to point B via a standard network. Instead, it imitates the architecture of massive parallel brain communication, sending billions of small amounts of information in thousands of different directions at the same time.
Computer makers intend to model up to a billion biological neurons in real time. For comparison: the mouse brain consists of approximately 100 million neurons, and the human brain is 1000 times larger. One billion neurons are 1% of the scale of the human brain, consisting of almost 100 billion neurons, which are closely connected by approximately one quadrillion (one with 15 zeros) synapses.
What is a computer with a million nuclear processor used to imitate the brain? First of all, it helps neuroscientists better understand how this organ functions. He does this by conducting extremely large-scale simulations in real time, impossible on other machines.
SpiNNaker was also involved in controlling the SpOmnibot robot. It uses a supercomputer system to interpret visual information in real time and navigate towards certain objects, ignoring others.
“Neuroscientists can now use SpiNNaker to unravel some of the secrets of the human brain through unprecedented large-scale simulations,” said computer science professor Steve Farber. “In addition, it works as a real-time neural simulator, which allows robotics to create large-scale neural networks for mobile robots so that they can walk, talk and move.”