Chinese researchers have developed a compact computer chip capable of modelling complex brain structures in real time, a breakthrough that could improve the diagnosis and treatment of neurological disorders and advance brain-machine interfaces, according to its developers.
The team from Peking University's School of Integrated Circuits and the Chinese Academy of Sciences published the peer-reviewed study in the journal Science on July 3.
ALSO READ | Nvidia Pushes Back Release Date For Kyber Rack Next Gen AI Computer To 2028
Built on a 40-nanometre process, the chip incorporates an artificial neural network and uses phase-change memristors that perform computation and data storage within the same memory array. This design reduces the delays and energy consumption associated with conventional chip architectures, where data must move between separate memory and processing units.
Up To 478 Times Faster Than Nvidia A100 For Brain Modelling
The device reconstructs complex brain surfaces in under half a second, with the researchers claiming it delivers between 50 and 478 times the performance of Nvidia's A100 GPU for this specific workload. It is also reported to be 3.5 to 36 times faster than existing dedicated accelerators.
Lead author Yang Yuchao, a professor at Peking University, said the chip can accurately reproduce brain folds for medical applications and could enable advances in brain-machine interfaces, disease diagnosis and, eventually, personalised digital brain twins. He added that the technology could also support neural navigation during surgery and early screening for Alzheimer's disease.
The comparison is based on Nvidia's A100 GPU, introduced in 2020, and applies only to brain-surface reconstruction rather than general-purpose AI or graphics workloads. The chip is designed specifically for neuroscience applications and is not intended to replace general-purpose GPUs.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.