Algorithmic-Level Approximate Computing Applied to Energy Efficient HEVC Decoding

This paper presents a novel method for applying approximate computing at the level of a complete application. The method decomposes the application into processing blocks which types define the classes of approximate computing techniques they may tolerate. By applying these approximation techniques...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on emerging topics in computing 2019-01, Vol.7 (1), p.5-17
Main Authors: Nogues, Erwan, Menard, Daniel, Pelcat, Maxime
Format: Article
Language:English
Subjects:
Algorithms
Approximate computing
Approximation algorithms
Complexity theory
Computing costs
Decoding
Energy consumption
energy efficiency
Engineering Sciences
HEVC
Quality of service
Signal and Image processing
Signal processing
Signal processing algorithms
Video compression
video decoding
Citations: Items that this one cites
Items that cite this one
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper presents a novel method for applying approximate computing at the level of a complete application. The method decomposes the application into processing blocks which types define the classes of approximate computing techniques they may tolerate. By applying these approximation techniques to the most computationally intensive blocks, drastic energy reduction can be obtained at a limited cost in terms of Quality of Service. The algorithmic-level approximate computing method is applied to a software High Efficiency Video Coding (HEVC) video decoder. The method is shown to offer multiple trade-offs between the quality of the decoded video and the energy required for the decoding process. The algorithmic-level approximate computing method offers new possibilities in terms of application energy budgeting. Energy reductions of up to 40 percent are demonstrated for a limited degradation of the application Quality of Service.
ISSN:2168-6750
2168-6750
DOI:10.1109/TETC.2016.2593644