Energy consumption of the Jacobi Test Code on the Blue Gene/Q: does using single precision reduce energy consumption?

ePubs

The open archive for STFC research publications

DOI	10.5286/raltr.2016005
Persistent URL	http://purl.org/net/epubs/work/24764929
Record Status	Checked
Record Id	24764929
Title	Energy consumption of the Jacobi Test Code on the Blue Gene/Q: does using single precision reduce energy consumption?
Contributors	T Byrne (STFC Rutherford Appleton Lab.), M Mawson (STFC Daresbury Lab.), AD Taylor (STFC Daresbury Lab.), HS Thorne (STFC Rutherford Appleton Lab.)
Abstract	Power optimisation is one of the major requirements for the achievement of exascale systems. In this report, we will compare the power and energy requirements of the Jacobi Test Code when run in single and double precision on STFC Hartree Centre’s IBM Blue Gene/Q, Blue Joule. We will show that switching from double to single precision can give significant energy consumption savings but only when the amount of data movement is significant. As a result, algorithm/software developers should take this into account when trying to reduce the amount of energy consumed by running their codes on computers with similar underlying architectures.
Organisation	STFC , SCI-COMP
Keywords	software outlook , double precision , CCP , single precision , energy efficient computing , Blue Gene/Q Jacobi Test Code
Funding Information
Related Research Object(s):
Licence Information:	Creative Commons Attribution 3.0 Unported (CC BY 3.0)
Language	English (EN)

Type	Details	URI(s)	Local file(s)	Year
Report	RAL Technical Reports RAL-TR-2016-005. 2016.		RAL-TR-2016-005.pdf	2016