TAIPEI (Taiwan News) — A research team from National Yang Ming Chiao Tung University (NYCU) has developed a machine learning-based temperature management system to improve chip cooling.

According to an NYCU press release on Tuesday (July 16), Professor Chen Kun-chih (陳坤志) led the team to produce a low-cost solution that can lead to better computing performance. 

The team explained that while multi-core processors are powerful, they produce much heat, per CNA. Additionally, as the number of processors increases, it becomes more difficult for them to communicate.

By applying adaptive machine learning, the system can predict which cores need to be cooled and can adjust workload accordingly, the team said. The system also includes real-time feedback of current and predicted temperatures.

Compared to traditional methods, the adaptive reinforcement learning method reduces temperature prediction errors while enhancing system performance, Chen said.

The team’s research won the IEEE TVLSI Best Paper Award and marks the first time in 30 years that a Taiwanese team has received this honor.