A low-power branch predictor for embedded processors

Sung Woo Jung, Gi Ho Park, Sung Bae Park

Research output: Contribution to journalArticle

Abstract

Even in embedded processors, the accuracy in a branch prediction significantly affects the performance. In designing a branch predictor, in addition to accuracy, microarchitects should consider area, delay and power consumption. We propose two techniques to reduce the power consumption; these techniques do not requires any additional storage arrays, do not incur additional delay (except just one MUX delay) and never deteriorate accuracy. One is to look up two predictions at a time by increasing the width (decreasing the depth) of the PHT (Prediction History Table). The other is to reduce unnecessary accesses to the BTB (Branch Target Buffer) by accessing the PHT in advance. Analysis results with Samsung Memory Compiler show that the proposed techniques reduce the power consumption of the branch predictor by 15-52%.

Original languageEnglish
Pages (from-to)2253-2257
Number of pages5
JournalIEICE Transactions on Information and Systems
VolumeE87-D
Issue number9
Publication statusPublished - 2004 Sep 1
Externally publishedYes

Fingerprint

Electric power utilization
Data storage equipment

Keywords

  • Branch predictor
  • Global predictor
  • Gshare
  • Low-power design
  • Microarchitecture

ASJC Scopus subject areas

  • Information Systems
  • Computer Graphics and Computer-Aided Design
  • Software

Cite this

A low-power branch predictor for embedded processors. / Jung, Sung Woo; Park, Gi Ho; Park, Sung Bae.

In: IEICE Transactions on Information and Systems, Vol. E87-D, No. 9, 01.09.2004, p. 2253-2257.

Research output: Contribution to journalArticle

Jung, Sung Woo ; Park, Gi Ho ; Park, Sung Bae. / A low-power branch predictor for embedded processors. In: IEICE Transactions on Information and Systems. 2004 ; Vol. E87-D, No. 9. pp. 2253-2257.
@article{060c44403349476d9e3c162fcef9d24c,
title = "A low-power branch predictor for embedded processors",
abstract = "Even in embedded processors, the accuracy in a branch prediction significantly affects the performance. In designing a branch predictor, in addition to accuracy, microarchitects should consider area, delay and power consumption. We propose two techniques to reduce the power consumption; these techniques do not requires any additional storage arrays, do not incur additional delay (except just one MUX delay) and never deteriorate accuracy. One is to look up two predictions at a time by increasing the width (decreasing the depth) of the PHT (Prediction History Table). The other is to reduce unnecessary accesses to the BTB (Branch Target Buffer) by accessing the PHT in advance. Analysis results with Samsung Memory Compiler show that the proposed techniques reduce the power consumption of the branch predictor by 15-52{\%}.",
keywords = "Branch predictor, Global predictor, Gshare, Low-power design, Microarchitecture",
author = "Jung, {Sung Woo} and Park, {Gi Ho} and Park, {Sung Bae}",
year = "2004",
month = "9",
day = "1",
language = "English",
volume = "E87-D",
pages = "2253--2257",
journal = "IEICE Transactions on Information and Systems",
issn = "0916-8532",
publisher = "Maruzen Co., Ltd/Maruzen Kabushikikaisha",
number = "9",

}

TY - JOUR

T1 - A low-power branch predictor for embedded processors

AU - Jung, Sung Woo

AU - Park, Gi Ho

AU - Park, Sung Bae

PY - 2004/9/1

Y1 - 2004/9/1

N2 - Even in embedded processors, the accuracy in a branch prediction significantly affects the performance. In designing a branch predictor, in addition to accuracy, microarchitects should consider area, delay and power consumption. We propose two techniques to reduce the power consumption; these techniques do not requires any additional storage arrays, do not incur additional delay (except just one MUX delay) and never deteriorate accuracy. One is to look up two predictions at a time by increasing the width (decreasing the depth) of the PHT (Prediction History Table). The other is to reduce unnecessary accesses to the BTB (Branch Target Buffer) by accessing the PHT in advance. Analysis results with Samsung Memory Compiler show that the proposed techniques reduce the power consumption of the branch predictor by 15-52%.

AB - Even in embedded processors, the accuracy in a branch prediction significantly affects the performance. In designing a branch predictor, in addition to accuracy, microarchitects should consider area, delay and power consumption. We propose two techniques to reduce the power consumption; these techniques do not requires any additional storage arrays, do not incur additional delay (except just one MUX delay) and never deteriorate accuracy. One is to look up two predictions at a time by increasing the width (decreasing the depth) of the PHT (Prediction History Table). The other is to reduce unnecessary accesses to the BTB (Branch Target Buffer) by accessing the PHT in advance. Analysis results with Samsung Memory Compiler show that the proposed techniques reduce the power consumption of the branch predictor by 15-52%.

KW - Branch predictor

KW - Global predictor

KW - Gshare

KW - Low-power design

KW - Microarchitecture

UR - http://www.scopus.com/inward/record.url?scp=5444230231&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=5444230231&partnerID=8YFLogxK

M3 - Article

VL - E87-D

SP - 2253

EP - 2257

JO - IEICE Transactions on Information and Systems

JF - IEICE Transactions on Information and Systems

SN - 0916-8532

IS - 9

ER -