DocumentCode :
703976
Title :
Power-efficient control of thermoelectric coolers considering distributed hot spots
Author :
Dousti, Mohammad Javad ; Pedram, Massoud
Author_Institution :
Dept. of Electr. Eng., Univ. of Southern California, Los Angeles, CA, USA
fYear :
2015
fDate :
9-13 March 2015
Firstpage :
966
Lastpage :
971
Abstract :
Thermoelectric coolers are compact devices that can target hot spots on a VLSI die. These devices are connected electrically in series and controlled together, i.e., all are ON or OFF at the same time. However, spatial and temporal distributions of hot spots on a VLSI die are non-uniform, and therefore, activating all of TECs to address one or a few localized hot spots is not economical. This traditional technique indeed leads to a significant power waste. This paper suggests that adjacent hot spots with the same thermal behavior can be grouped and controlled by a cluster of TECs. A bypass switch for each TEC cluster is added in order to allow selectively turning OFF some TEC clusters which are needed. More precisely, a clustering problem is formulated which aims to minimize the power waste due to excessive use of TECs. Due to the large number of variables in problems of interesting sizes, a greedy heuristic method for solving the problem is introduced. It is shown that the proposed heuristic can reduce the wasted power on average by 81% and also decrease the total TEC power consumption on average by 42%.
Keywords :
VLSI; energy conservation; greedy algorithms; power consumption; power distribution control; thermoelectric cooling; TEC power efficient control; TECs; bypass switch; greedy heuristic method; hot spot spatial distribution; hot spot temporal distribution; nonuniform VLSI die; power consumption; power wastage minimization; thermoelectric coolers; Benchmark testing; Cooling; Graphical models; Heating; Power demand; Switches;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Design, Automation & Test in Europe Conference & Exhibition (DATE), 2015
Conference_Location :
Grenoble
Print_ISBN :
978-3-9815-3704-8
Type :
conf
Filename :
7092528
Link To Document :
بازگشت