DocumentCode
3403948
Title
Design automation flow for voltage adaptive optimum granularity LITHE for sequential circuits
Author
Balasubramanian, Venkat Krishnan ; Hao Xu ; Vemuri, Ranga
Author_Institution
Sch. of Electron. & Comput. Syst., Univ. of Cincinnati, Cincinnati, OH, USA
fYear
2013
fDate
4-6 Sept. 2013
Firstpage
355
Lastpage
360
Abstract
Runtime Active Leakage Control (RALC) techniques like Power Gating (PG) and Reverse Body Biasing (RBB) being the primary candidates, several novel techniques such as Light Vth Hopping (LITHE) [3] with reduced substrate biasing have proven to be beneficial. The intention of LITHE is to reduce the Energy Breakeven Time (EBT) of the circuit, leading to frequent application of RBB. There are several problems that arise while integrating LITHE in a design flow. In this paper, a generic flow from RTL to spice incorporating the LITHE technique at near-optimum granularity is proposed. The problems of control transistor sizing and control signal generation is also addressed. A preprocessing clustering algorithm is applied to extend LITHE technique to sequential circuits. The entire flow has been developed using a 90 nm library developed from Synopsys PDK and has been verified on designs with varying complexities.
Keywords
logic design; sequential circuits; EBT; LITHE technique; PG; RALC technique; RBB; RBB frequent application; Synopsys PDK; control signal generation; control transistor sizing; design automation flow; energy breakeven time; generic flow; light threshold hopping; near-optimum granularity; power gating; preprocessing clustering algorithm; reduced substrate biasing; reverse body biasing; runtime active leakage control technique; sequential circuits; size 90 nm; voltage adaptive optimum granularity LITHE; Benchmark testing; Clocks; Complexity theory; Cooling; Educational institutions; Lead; Logic gates; LITHE; Leakage Reduction; Low Power; RBB;
fLanguage
English
Publisher
ieee
Conference_Titel
SOC Conference (SOCC), 2013 IEEE 26th International
Conference_Location
Erlangen
ISSN
2164-1676
Type
conf
DOI
10.1109/SOCC.2013.6749715
Filename
6749715
Link To Document