Story perspectives

UT Dallas Unveils Energy-Efficient Brain-Like Computer Prototype

11/2/2025

Summary

Story summary

Researchers at the University of Texas at Dallas developed a neuromorphic computer prototype that learns patterns with fewer training computations, potentially reducing energy use.
The design integrates memory and processing to mimic brain function.
Led by Dr. Joseph S. Friedman, the team used magnetic tunnel junctions (MTJs) to form synaptic-like connections.
The prototype achieved 90% accuracy in handwritten digit classification and over 600 trillion operations per second per watt.

History

Historical Context

Neuromorphic computing, inspired by the brain's architecture, began gaining traction in the late 1980s with Carver Mead's proposals. Significant advancements followed, including MIT's 2011 chip mimicking synaptic communication and IBM's TrueNorth chip in 2014, which featured 1 million neurons. Recent developments, like the 2022 prototype from the University of Texas at Dallas, demonstrate the potential for energy-efficient computing and enhanced artificial intelligence applications.

1 of 2

Story summary

Researchers at the University of Texas at Dallas developed a neuromorphic computer prototype that learns patterns with fewer training computations, potentially reducing energy use.
The design integrates memory and processing to mimic brain function.
Led by Dr. Joseph S. Friedman, the team used magnetic tunnel junctions (MTJs) to form synaptic-like connections.
The prototype achieved 90% accuracy in handwritten digit classification and over 600 trillion operations per second per watt.

1 / 2