Sleep transistor sizing method using accurate delay estimation considering input vector pattern and non-linear current model

Seidai Takeda, Kyundong Kim, Hiroshi Nakamura, Kimiyoshi Usami

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Beyond deep sub-micron era, Power Gating (PG) is one of the most effective techniques to reduce leakage power of circuits. The most important issue of PG circuit design is how to decide the width of sleep transistor. Smaller total sleep transistor width provides smaller leakage power in standby mode, however, insufficient sleep transistor insertion suffers from significant performance degradation. In this paper, we present an accurate and fast gate-level delay estimation method for PG circuits and a novel sleep transistor sizing method utilizing our delay estimation for module-based PG circuits. This method achieves high accuracy within acceptable computation time utilizing accurate discharge current estimation based on delayed logic simulations with limited input vector patterns and by realizing precise current characteristics for logic gates and sleep transistors. Experimental results show that our delay estimation successfully achieves high accuracy and avoids overestimation and underestimation seen in conventional method. Also, our sleep transistor sizing method on average successfully reduces the width of sleep transistors by 40% when compared to conventional methods within an acceptable computation time.

Original languageEnglish
Pages (from-to)2499-2509
Number of pages11
JournalIEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
VolumeE94-A
Issue number12
DOIs
Publication statusPublished - 2011 Dec

Keywords

  • Delay
  • Leakage power
  • MTCMOS
  • Power gating

ASJC Scopus subject areas

  • Signal Processing
  • Computer Graphics and Computer-Aided Design
  • Electrical and Electronic Engineering
  • Applied Mathematics

Fingerprint Dive into the research topics of 'Sleep transistor sizing method using accurate delay estimation considering input vector pattern and non-linear current model'. Together they form a unique fingerprint.

  • Cite this