Researchers at Western Sydney University in Australia, in collaboration with tech giants Intel and Dell, are set to unveil a supercomputer named DeepSouth, designed to simulate neural networks at the scale of the human brain. The computer is capable of emulating networks of spiking neurons at a rate of 228 trillion synaptic operations per second, mirroring the estimated processing speed of the human brain. Instead of aiming to be the most powerful conventional supercomputer, DeepSouth is focused on simulating the brain’s network of neurons using a neuromorphic system, mimicking biological processes and making it more efficient and less power-hungry. The researchers hope the system will advance understanding of the brain and contribute to the development of brain-scale computing applications in fields such as sensing, biomedical research, robotics, space exploration, and large-scale AI applications.
The DeepSouth supercomputer, scheduled to become operational in April 2024, is expected to offer researchers unprecedented insights into how the human brain processes information. Neuromorphic computing, the basis of DeepSouth, involves performing numerous operations simultaneously with minimal data movement, leading to reduced energy consumption. The researchers believe that simulating spiking neural networks on standard computers using graphics processing units (GPUs) and multicore central processing units (CPUs) is slow and power-intensive, and they aim to address these limitations with their neuromorphic system. The project has the potential to significantly impact the development of advanced smart devices and more energy-efficient AI models, opening up new possibilities for research in neuroscience and AI engineering solutions.
