You are here

Event Details

MP Associates, Inc.
SUNDAY September 30, 2:00pm - 6:00pm | Mollino
EVENT TYPE: TUTORIAL
SESSION 5T
A Comprehensive Analysis of Approximate Computing Techniques: From Component- to Application-Level

Speakers:
Alberto Bosio - Lyon Institute of Nanotechnology
Daniel Menard - Institut d'Électronique et de Télécommunications de Rennes
Olivier Sentieys - French Institute for Research in Computer Science and Automation
Organizers:
Alberto Bosio - Lyon Institute of Nanotechnology
Daniel Menard - Institut National des Sciences Appliquées de Rennes
Olivier Sentieys - French Institute for Research in Computer Science and Automation
A new design paradigm, Approximate Computing (AxC), has been established to investigate how computing systems can be more energy efficient, faster, and less complex. Intuitively, instead of
performing exact computation and, consequently, requiring a high amount of resources, AxC aims to selectively violate the specifications, trading accuracy off for efficiency. It has been demonstrated in the literature the effectiveness of imprecise computation for both software and hardware components implementing inexact algorithms, showing an inherent resiliency to errors.

This tutorial introduces basic and advanced topics on AxC. We intend to follow a bottom-up approach: from component- up to application-level. More in detail, we will first present the main concept and
techniques (e.g., functional approximation, voltage over-scaling). We then move to present some compile-time results in terms of energy-efficiency, area, performance versus accuracy of computations when using customized arithmetic (fixed-point, floating-point) and also try to derive some conclusions by comparing the different paradigms. The algorithmic-level approximation methods are then presented. Energy consumption can be reduced by approximating or skipping part of the computation. The concept of incremental refinement, early termination and fast decision will be detailed.

BIOGRAPHIES:
Alberto Bosio received the PhD in Computer Engineering from the Politecnico di Torino, Italy in 2006. From 2007 he is an Associate Professor at LIRMM - University of Montpellier in France. His research interests include Approximate Computing, In-Memory Computing, Test and Diagnosis of Digital circuits and systems and Reliability. He co-authored 1 book, 3 patents 35 journals, and over 120 conference papers. He will be the chair of the ETTTC from January 2018. He is a member of the IEEE.
Web: http://www.lirmm.fr/~bosio/home/

Daniel Menard received the Ph.D. and HDR degrees in Signal Processing and Telecommunications from the University of Rennes, respectively in 2002 and 2011. From 2003 to 2012 he was Associate Professor at University of Rennes in France. He is currently Full Professor at INSA Rennes. His research activities focus on the energy efficient implementation of signal and image processing applications in embedded systems. His research topics include approximate computing, fixed-point arithmetic, energy optimization in MPSoC, low power HEVC video encoding and decoding and embedded stereo-vision. He has published 25 papers in international journals and 63 papers in international conferences. He is member of the DISPS Technical Committee of the IEEE Signal Processing Society.
Web: http://dmenard.perso.insa-rennes.fr

Olivier Sentieys is a Professor at the University of Rennes and holds an Inria Research Chair on Energy-Efficient Computing Systems. He has more than 20 years of expertise in the fields of system-on-chip architectures, reconfigurable systems and their associated CAD tools, finite arithmetic effects, numerical accuracy analysis and low-power sensor networks. He authored or coauthored more than 150 journal publications or peer-reviewed conference papers and hold 6 patents. In particular, his research on methods for analytical analysis of errors in reduced-precision arithmetic and word-length optimization since 2000 with more than 50 publications, can be considered as a pioneering work in the field of approximate computing.
Web: http://people.rennes.inria.fr/Olivier.Sentieys/