2002Unpublished venueRequires access

High-performance low-power CMOS circuits using multiple channel length and multiple oxide thickness

Naran Sirisantana, Lili Wei, Kaushik Roy

Open publisher page 101 citations

Abstract

Power optimization has become an important issue for high performance designs. One way to achieve low-power and high performance circuits is to use dual-threshold voltages. High threshold transistors can be used in non-critical paths to reduce the leakage power, while lower threshold voltage is used for transistors in critical path(s) to achieve high performance. This paper proposes two low power and high performance CMOS design techniques-multiple channel length (M/sub L/CMOS) and multiple oxide thickness (M/sub ox/CMOS), based on dual V/sub th/, design technique. A comprehensive algorithm for selecting and assigning optimal transistor threshold voltage, channel length and oxide thickness is given. The simulation results on ISCAS benchmark circuits show that the total power consumption can be reduced by 21% for M/sub L/CMOS at low activity. Total power savings for M/sub ox/CMOS at low and high switching activities are about 42% and 24%, respectively.

About this research paper

What this paper is about

Power optimization has become an important issue for high performance designs. One way to achieve low-power and high performance circuits is to use dual-threshold voltages. High threshold transistors can be used in non-critical paths to reduce the leakage power, while lower threshold voltage is used for transistors in critical path(s) to achieve high performance. This paper proposes two low power and high performance CMOS design techniques-multiple channel length (M/sub L/CMOS) and multiple oxide thickness (M/sub ox/CMOS), based on dual V/sub th/, design technique. A comprehensive algorithm for selecting and assigning optimal transistor threshold voltage, channel length and oxide thickness is given. The simulation results on ISCAS benchmark circuits show that the total power consumption can be reduced by 21% for M/sub L/CMOS at low activity. Total power savings for M/sub ox/CMOS at low and high switching activities are about 42% and 24%, respectively.

Why it matters

OpenAlex reports 101 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

Power optimization has become an important issue for high performance designs. One way to achieve low-power and high performance circuits is to use dual-threshold voltages. High threshold transistors can be used in non-critical paths to reduce the leakage power, while lower threshold voltage is used for transistors in critical path(s) to achieve high performance. This paper proposes two low power and high performance CMOS design techniques-multiple channel length (M/sub L/CMOS) and multiple oxide thickness (M/sub ox/CMOS), based on dual V/sub th/, design technique. A comprehensive algorithm for selecting and assigning optimal transistor threshold voltage, channel length and oxide thickness is given. The simulation results on ISCAS benchmark circuits show that the total power consumption can be reduced by 21% for M/sub L/CMOS at low activity. Total power savings for M/sub ox/CMOS at low and high switching activities are about 42% and 24%, respectively.

Key concepts: CMOS, Electronic circuit, Channel (broadcasting), Power (physics), Low-power electronics, Materials science, Length measurement, Electronic engineering

Related papers

Back to paper searchBrowse research topicsOriginal source
High-performance low-power CMOS circuits using multiple channel length and multiple oxide thickness — Research Paper | ScholarLens