UT Researchers Collaborate with TSMC on Energy-Efficient AI Chips
Researchers from the University of Texas (UT) and Taiwan Semiconductor Manufacturing Company (TSMC) have collaborated for three years on developing energy-efficient chips for artificial intelligence (AI). Their work, detailed in a paper released in August, focuses on spin-orbit torque magnetic random access memory (SOT-MRAM) for AI applications. According to a news release from the Cockrell School of Engineering, SOT-MRAM technology consumes less energy than existing memory technologies and does not necessitate the use of large data centers. The primary objective of this research is to decrease the resource demands of AI without compromising the accuracy of its models. Jean Anne Incorvia, the faculty lead on the project, emphasized the need for more efficient methods to manage the increasing energy and water consumption of AI data centers. Electrical and Computer Engineering Department Ph.D. student Vivian Rogers highlighted the significant opportunity to collaborate with TSMC, a major player in the semico...
