VLSI Circuit Partitioning and Placement for Low Power Design

碩士 === 中原大學 === 資訊工程研究所 === 89 === Power consumption is an important issue in VLSI circuit design. In this paper, we use quadrisection algorithm based on F-M partitioning, and the gain update scheme proposed in [1] to perform our partitioning process. Then, we use iterative partitioning t...

Full description

Bibliographic Details
Main Authors: Hsien-Wen Chang, 張獻文
Other Authors: Mely Chen-Chi
Format: Others
Language:zh-TW
Published: 2001
Online Access:http://ndltd.ncl.edu.tw/handle/54530305117538899733
id ndltd-TW-089CYCU5392025
record_format oai_dc
spelling ndltd-TW-089CYCU53920252016-07-06T04:10:06Z http://ndltd.ncl.edu.tw/handle/54530305117538899733 VLSI Circuit Partitioning and Placement for Low Power Design 低功率導向之超大型積體電路分割與擺置方法 Hsien-Wen Chang 張獻文 碩士 中原大學 資訊工程研究所 89 Power consumption is an important issue in VLSI circuit design. In this paper, we use quadrisection algorithm based on F-M partitioning, and the gain update scheme proposed in [1] to perform our partitioning process. Then, we use iterative partitioning to complete placement. In the partitioning and placement process, wire load power consumption is concerned to reduce total power, and gate load power consumption is added in the constraint to obtain uniform power distribution. The experimental result shows that, concerning switching rate in the cost function computing can reduce total wire load power consumption. Except area constraint, if gate load power constraint is added, we can distribute gate load power consumption uniformly and reduce the conjunction of high switching rate nets. Further more, we also think of high switching and high fanout nets especially clock net. After the cost of clock is multiplied by a weight(>1), we can not only reduce the power of the clock net but also reduce the total power. Experimental results are shown. Mely Chen-Chi 陳美麗 2001 學位論文 ; thesis 46 zh-TW
collection NDLTD
language zh-TW
format Others
sources NDLTD
description 碩士 === 中原大學 === 資訊工程研究所 === 89 === Power consumption is an important issue in VLSI circuit design. In this paper, we use quadrisection algorithm based on F-M partitioning, and the gain update scheme proposed in [1] to perform our partitioning process. Then, we use iterative partitioning to complete placement. In the partitioning and placement process, wire load power consumption is concerned to reduce total power, and gate load power consumption is added in the constraint to obtain uniform power distribution. The experimental result shows that, concerning switching rate in the cost function computing can reduce total wire load power consumption. Except area constraint, if gate load power constraint is added, we can distribute gate load power consumption uniformly and reduce the conjunction of high switching rate nets. Further more, we also think of high switching and high fanout nets especially clock net. After the cost of clock is multiplied by a weight(>1), we can not only reduce the power of the clock net but also reduce the total power. Experimental results are shown.
author2 Mely Chen-Chi
author_facet Mely Chen-Chi
Hsien-Wen Chang
張獻文
author Hsien-Wen Chang
張獻文
spellingShingle Hsien-Wen Chang
張獻文
VLSI Circuit Partitioning and Placement for Low Power Design
author_sort Hsien-Wen Chang
title VLSI Circuit Partitioning and Placement for Low Power Design
title_short VLSI Circuit Partitioning and Placement for Low Power Design
title_full VLSI Circuit Partitioning and Placement for Low Power Design
title_fullStr VLSI Circuit Partitioning and Placement for Low Power Design
title_full_unstemmed VLSI Circuit Partitioning and Placement for Low Power Design
title_sort vlsi circuit partitioning and placement for low power design
publishDate 2001
url http://ndltd.ncl.edu.tw/handle/54530305117538899733
work_keys_str_mv AT hsienwenchang vlsicircuitpartitioningandplacementforlowpowerdesign
AT zhāngxiànwén vlsicircuitpartitioningandplacementforlowpowerdesign
AT hsienwenchang dīgōnglǜdǎoxiàngzhīchāodàxíngjītǐdiànlùfēngēyǔbǎizhìfāngfǎ
AT zhāngxiànwén dīgōnglǜdǎoxiàngzhīchāodàxíngjītǐdiànlùfēngēyǔbǎizhìfāngfǎ
_version_ 1718337523579617280