Home // COMPUTATION TOOLS 2019, The Tenth International Conference on Computational Logics, Algebras, Programming, Tools, and Benchmarking // View article

Code-level Optimization for Program Energy Consumption

Authors:
Cuijiao Fu
Depei Qian
Tianming Huang
Zhongzhi Luan

Keywords: Energy Optimization; Ineffective Memory Access; Energy-efficient Software

Abstract:
A lot of time is spent on CPU waiting for memory accesses to complete during the program is being executed, which would be longer because of data structure choice, lack of design for performance, and ineffective compiler optimization. Longer execution time means more energy consumption. To save energy, avoiding unnecessary memory accesses operations is desirable In this paper, we optimize program energy consumption by detecting and modifying the dead write, which is common inefficient memory access. Our analysis of the SPEC CPU2006 benchmarks shows that the reduction of the program running energy consumption is significant after the dead write in the code was modified. For example, the SPEC CPU2006 GCC benchmark had reduced energy consumption by up to 26.7% in some inputs and 13.5% on average. We think this energy optimization approach has tremendous benefits for the developer to develop more energy-efficient software.

Pages: 1 to 4

Copyright: Copyright (c) IARIA, 2019

Publication date: May 5, 2019

Published in: conference

ISSN: 2308-4170

ISBN: 978-1-61208-709-2

Location: Venice, Italy

Dates: from May 5, 2019 to May 9, 2019