This proposal addresses the fundamental challenges in todays on-chip thermal sensing and control problems. The thermal sensing technique in even the most recent processors predominantly relies on thermal diodes or sensors. Both suffer from the fixed-location problem as on-chip hot spots migrate at run time. This project aims to develop novel techniques for spectrum and high precision temperature sensing to mitigate this problem. The proposed solutions can lead to more efficient and effective on-line thermal management. The intellectual merit of the project includes the development of a fast, lean, and accurate software thermal sensor for online temperature tracking. The software adopts a highly efficient and fine-tuned numerical method to calculate temperatures at a fine granularity both temporally and spatially. Experimentation will demonstrate its superior accuracy and speed, and hence the great potential of becoming a truly competitive remedy for chip temperature sensing.

Agency
National Science Foundation (NSF)
Institute
Division of Computer and Communication Foundations (CCF)
Application #
0734339
Program Officer
Chitaranjan Das
Project Start
Project End
Budget Start
2007-04-01
Budget End
2010-03-31
Support Year
Fiscal Year
2007
Total Cost
$229,674
Indirect Cost
Name
University of Pittsburgh
Department
Type
DUNS #
City
Pittsburgh
State
PA
Country
United States
Zip Code
15213