The work by Ph.D. student Sherif Sadiqbatcha on real-time full-chip thermal estimation based on the machine learning method for commercial multi-core CPU has been accepted by IEEE Transaction on Computer, the top journal for computer architecture.

The is significant work from VSCLAB for using machine learning to generate the full-chip map directly from the tasks running the chip. It provides more information for run-time dynamic thermal, reliability and power management. 

S. Sadiqbatcha, J. Zhang, H. Amrouch and S. X.-D. Tan, “Real-time full-chip thermal tracking: a post-silicon, machine learning perspective”, IEEE Transaction on Computers (TC),  (accepted), 2021.