Researchers at Western Sydney University in Australia are assembling a supercomputer designed to simulate the world's most efficient learning machine - a neuromorphic monster capable of the same 228 trillion synaptic operations per second as the human brain. As the era of artificial intelligence dawns, it is clear that this crazy technological leap is one of the most important in the history of the earth and will soon penetrate into every aspect of our lives. But it all relies on absolutely massive computing power.
In fact, on current trends, the AI servers sold by NVIDIA alone may consume more energy each year than many small countries. In a world that strives to decarbonize, such an energy load is a huge drag.
However, nature has solved this problem. The human brain is still the most advanced at present, capable of learning quickly from small amounts of messy, noisy data, or processing the equivalent of a billion billion mathematical operations per second, while consuming only 20 watts of energy.
That's why a team at Western Sydney University is building the DeepSouth neuromorphic supercomputer - the first-ever machine capable of simulating spiking neural networks at the scale of the human brain.
Professor André van Schaik, Director of the International Center for Neuromorphic Systems, said: "Our inability to simulate brain-like networks at scale has hindered our progress in understanding how the brain uses neurons to perform computations. "
DeepSouth is expected to go online in April 2024. The research team expects that it will be able to process massive amounts of data at high speeds while being much smaller and consuming much less energy than other supercomputers due to the spiking neural network method.
It has a modular and scalable design using commercially available hardware, so it can be expanded or contracted in the future to suit various tasks. The venture aims to bring AI processing closer to the way the human brain works, while also learning more about the brain and hoping to make advances relevant to other fields.
Remarkably, other researchers are approaching the same problem from the opposite direction, and some teams have now begun using real human brain tissue as part of semi-mechanical computer chips, with impressive results.
learn more:
https://www.westernsydney.edu.au/newscentre/news_centre/more_news_stories/world_first_supercomputer_capable_of_brain-scale_simulation_being_built_at_western_sydney_university