University of California, Riverside

Department of Electrical and Computer Engineering



Prof. Tan’s paper is one of Top 10 Downloaded Articles in TODAES


Prof. Tan’s paper is one of Top 10 Downloaded Articles in TODAES
 

ACM LogoProf. Tan’s paper published in ACM on TODAES (ACM Transaction on Design Automation of Electronic Systems) in Feb. 2010 is one of Top 10 Downloaded Articles published in 2010.

D. Li, S. X.-D. Tan, E. H. Pacheco, M. Tirumala, “Parameterized architecture-level thermal modeling for multi-core microprocessors”, ACM Transaction on Design Automation of Electronic Systems (TODAES), vol. 15, no. 2, pp.1-22, February 2010

See TODAES website http://todaes.acm.org/ for details.

The paper aims to provide a new approach for building novel architecture-level dynamic thermal behavioral models to facilitate thermal-efficient multicore microprocessors at chip and package levels. The new algorithm allows fast thermal estimation and prediction for more efficient on-chip dynamic thermal management and thermal-aware design and optimization of high- performance multicore microprocessors. The significance of this paper is that dynamic thermal models is becoming more important and even imperative for leading microprocessor companies such Intel Corp and AMD due to increasing unpredictable on-chip heat and related thermal effects in multicore microprocessors (compared with existing resistor only thermal models used in the previous generations of industry microprocessors). The paper present a new dynamic thermal behavioral modeling techniques based on numerically stable and robust algorithm (an improved generalized pencil-of-functions) based on the silicon-accurate power and temperature data (measured or from field solvers). The new method can accommodate different parameter variables such as the locations of thermal sensors in a heat sink, different components (heat sink, heat spreader, core, cache, etc.), thermal conductivity of heat sink materials, and thermal boundary conditions etc. The compact behavioral models offer significant speedup over commercial thermal modeling and analysis tools such as FloTHERM on the industry examples. The proposed algorithm has been used by the authors’ industrial partner, Intel Corporation, to design advanced thermal package solutions.

More in News

More Information 

General Campus Information

University of California, Riverside
900 University Ave.
Riverside, CA 92521
Tel: (951) 827-1012

Department Information

Electrical and Computer Engineering
Suite 343 Winston Chung Hall
University of California, Riverside
Riverside, CA 92521-0429

Tel: (951) 827-2484
Fax: (951) 827-2425
E-mail: E-mail/Questions

Footer