Scientists unite 70,000 chips to simulate human brain

New technologies such as autonomous cars require a huge increase in the computing power of computers. A new computer chip is about to offer a solution to that.

Smart cities, 5G and autonomous driving should improve the quality of life in the future. But these technologies also require a huge increase in the computing power of computers. Today’s microprocessors cannot cope with such a challenge. A new computer chip is about to change that. Uh, make that 70,000 chips. That is the number of SpiNNaker2 chips that together form an internet cloud in Germany, spread over 16 servers at the Technical University of Dresden. Not only is the number impressive, but each chip is also designed with the ultimate computer in mind: the human brain.

Brains are much better at processing of some tasks than any computer chip. The brain is both faster and more efficient. Each neuron is connected with 7,000 synapses to another neuron. The brain also saves energy by transmitting only necessary communication. It makes decisions about what information is relevant to a problem, transmits only that information and therefore consumes less energy. The SpiNNaker2 chip actually contains accelerators that simulate neurons and synapses. In that sense, it’s an artificial intelligence network in miniature format.

Christian Mayr, professor at the chair for high parallel VLSI systems and neuro-Microelectronics at the Technical University of Dresden, contributed to the development of the SpiNNaker2 chip. He used the functioning of the brain as a guide to its development. He describes himself as”strongly inspired by biology”. Therefore, this supercomputer, which brings artificial intelligence and neurology together, is clearly superior to”classic artificial intelligence (AI).”