DeepSouth stands out due to its unique neuromorphic system, a design that emulates the biological processes of neurons. It achieves an impressive 228 trillion synaptic operations per second, a rate comparable to the human brain's capabilities. This novel approach to computing, as ICNS Director Professor Andre van Schaik explains, is tailored to function like networks of neurons, resulting in lower power consumption and increased efficiency compared to traditional supercomputers.
Traditional supercomputers, optimized for conventional computing tasks, are known for their high power usage. DeepSouth, however, represents a different approach. "Progress in our understanding of how brains compute using neurons is hampered by our inability to simulate brain-like networks at scale. Our system will change that," Professor van Schaik emphasized. The new supercomputer is not only a leap in neuromorphic computing but also a significant advancement for various applications, including sensing, biomedical, robotics, space, and large-scale AI applications.
One of the key advantages of DeepSouth is its capability for super-fast, large-scale parallel processing while consuming far less power. Mimicking the brain's efficiency, it processes massive amounts of data quickly and on a much smaller scale than other supercomputers. This scalability feature allows for both expansion and reduction, catering to different application sizes and costs.
Moreover, DeepSouth's reconfigurable nature, powered by Field Programmable Gate Arrays (FPGAs), adds to its versatility. This flexibility enables the integration of new neuron models and learning rules, a significant advancement over other neuromorphic systems with rigid, custom-designed hardware. The system will be remotely accessible, with a Python-based front end for designing neural models and networks, making it user-friendly for researchers without in-depth hardware knowledge.
Commercially available hardware forms the backbone of DeepSouth, ensuring continuous improvement independent of the design team. This approach contrasts with custom-designed neuromorphic systems, which are time-consuming and expensive to develop. The use of off-the-shelf hardware also suggests that DeepSouth's prototype could be replicated easily in data centers worldwide.
The supercomputer's potential impact on artificial intelligence is particularly noteworthy. By mimicking the brain's operations, DeepSouth is expected to enable more efficient AI processes than current models. This capability could lead to advances in smart devices, sensors for various industries, and AI applications that are less power-hungry yet more intelligent.
DeepSouth, named in homage to IBM's TrueNorth and Deep Blue systems, also acknowledges its geographical roots in Western Sydney. Its development involved collaboration with researchers from the University of Sydney, University of Melbourne, and University of Aachen, Germany, highlighting its international significance.
The supercomputer, set to become operational by April 2024, is not just a technological marvel; it is a key contributor to Western Sydney's emergence as a high-tech hub. With its unique capabilities, DeepSouth is poised to drive forward the understanding of the human brain and revolutionize the field of neuromorphic computing.
Related Links
Western Sydney University
All about the robots on Earth and beyond!
| Subscribe Free To Our Daily Newsletters |
| Subscribe Free To Our Daily Newsletters |
