from    
to    
search  

 


车辆与运载学院300期学术沙龙-Mechatronic Vehicle Systems Lab: Advances in Auto...
清华软件论坛第21期|Tamer zsu:Disaggregated & Heterogeneous Platform forDa...
物理系colloquium: 原子钟的发展
世纪物理情系列讲座 Manipulating Room-Temperature Polariton Condensates and Th...
报告题目:
Thermal-Aware Modeling, Analysis, Synthesis, and Run-Time Management
 报告人:
Prof. Robert Dick & Li Shang
Prof. Robert Dick (Northwestern University) 
Prof. Li Shang (Queen's University)
报告时间:
2006-08-11 10:00
报告地点:
东主楼10区309 East Main Building 10-309
主办单位:
电子工程系电路与系统研究室
  简介:

Increasing integrated circuit (IC) power consumptions and temperatures

pose serious challenges to performance, cooling cost, and lifetime

reliability. In this talk, we describe modeling, analysis, design,

synthesis, and run-time management techniques to solve these problems

systematically.

 

Accurate three-dimensional IC chip-package heat flow analysis has

typically required numerical methods that are too computationally

intensive for repeated use during IC design. We describe a comprehensive

solution that unifies spatial, time-domain, and frequency-domain

modeling techniques to efficiently and accurately solve the static,

short time scale, and long time scale IC thermal analysis problems.

Compared to existing dynamic analysis techniques, this approach achieves

more than 1,000X performance improvement with the same modeling accuracy.

 

We will also describe IC synthesis techniques that optimize temperature

and reliability via thermal-aware architectural moves and physical

design. The thermal optimization algorithm reduces peak temperature by

12.5 degrees C on average. The reliability optimization algorithm

permits an 85% increase in IC lifetime with only a 5% area cost.

 

Finally, we present a run-time thermal management technique that uses

hardware techniques such as clock gating and software techniques such as

thermal-aware process scheduling to proactively avoid thermal

emergencies, thereby improving performance by 9% compared with purely

hardware thermal management, while meeting the same constraints on peak

temperature.

 

报告人简历:

 

Li Shang received his B.E. and M.E. degrees from Tsinghua University and

his Ph.D. degree from Princeton University. He is currently an Assistant

Professor at the Department of Electrical and Computer Engineering,

Queen's University, Canada. Li has published in the areas of computer

architecture, design automation, thermal/power modeling and

optimization, reconfigurable computing, mobile computing, and

nanocomputing. He won the Best Paper Award at PDCS'02 and his

department's teaching award in 2006. He is the Walter F. Light Scholar

of Queen's University.

 

Robert P. Dick received his B.S. degree from Clarkson University and his

Ph.D. degree from Princeton University. He worked as a Visiting

Researcher at NEC Labs America, a Visiting Professor at Tsinghua

University's Department of Electronic Engineering, and is currently an

Assistant Professor at Northwestern University's Department of

Electrical Engineering and Computer Science. Robert received an NSF

CAREER award and won his department's Best Teacher of the Year award in

2004. He has published in the areas of embedded system synthesis,

embedded operating systems, mobile ad-hoc network protocols,

reliability, behavioral synthesis, data compression, and thermal

analysis of integrated circuits.

 

 

今日相关信息
 
同类别相关信息
椭球函数札记
清华信息大讲堂166讲:“逼上梁山”—...
清华论坛第66讲:3
CNN as Guided Multi-layer RECOS Tra...
Graphene and Other 2D Materials: A...
学术活动