Low power branch prediction for embedded application processors

Nadav Levison*, Shlomo Weiss

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Modern embedded processors used in media and communi-cation portable devices are now required to execute com- plex applications and their performance requirements are getting close to the demands of general purpose processors. The performance-per-Watt ratio is an extremely important measure in portable devices because of their limited power capacity. Branch predictors, and especially the BTB, are among the largest on-chip SRAM structures (after caches), and therefore are primary contributors to the total system power. We propose a novel micro-architectural method re- ferred to as Shifted-Index BTB with a Set-Buffer, which re- duces both dynamic and static power. Extensive simula- tions show that up to 80% reduction in dynamic power is achieved at the cost of up to 0.64% system slowdown. 58% reduction is static power is also achieved by applying low- leakage power techniques that mesh well with the Set-Buffer design.

Original languageEnglish
Title of host publicationISLPED'10 - Proceedings of the 16th ACM/IEEE International Symposium on Low-Power Electronics and Design
Pages67-72
Number of pages6
DOIs
StatePublished - 2010
Event16th ACM/IEEE International Symposium on Low-Power Electronics and Design, ISLPED'10 - Austin, TX, United States
Duration: 18 Aug 201020 Aug 2010

Publication series

NameProceedings of the International Symposium on Low Power Electronics and Design
ISSN (Print)1533-4678

Conference

Conference16th ACM/IEEE International Symposium on Low-Power Electronics and Design, ISLPED'10
Country/TerritoryUnited States
CityAustin, TX
Period18/08/1020/08/10

Keywords

  • ARM Cortex
  • BTB
  • Battery
  • Embedded
  • Mobile
  • Power

Fingerprint

Dive into the research topics of 'Low power branch prediction for embedded application processors'. Together they form a unique fingerprint.

Cite this