Special session on Approximate Computing for IoT

The ARTEFaCT partners have organized a hot-topic special session at DATE’17 on Approximate Computing for improving power efficiency of IoT and HPC.

Power efficiency is the primary concern of IoT-related applications, both at the sensor node and on its cloud-computing counterpart. Unfortunately, achieving high efficiency and robustness requires complex and conflicting design constraints. Fortunately, the inherent error resiliency of many IoT applications allows the use of Approximate Computing techniques leading to great benefits on power efficiency while having a minimal impact on the applications. This special session has shown how Approximate Computing techniques applied at hardware and software levels can leverage the power issue both for sensor-node and HPC (High-Performance Computing) ends.

Detailed special session program

Chair:                   Christian Enz, EPFL, CH
Organizer:           Vincent Camus, EPFL, CH

Christian Enz, EPFL, CH
Olivier Sentieys, INRIA, FR
Vincent Camus, EPFL, CH
Bert Moons, KU Leuven, BE
Daniel Ménard, INSA Rennes, FR
Anshumali Shrivastava, Rice University, US

Pushing the limits of voltage over-scaling for error-resilient applications, R. Ragavan¹, B. Barrois¹², C. Killian¹ and O. Sentieys¹, ¹INRIA, France; ²University of Rennes, France

Combining structural and timing errors in overclocked Inexact Speculative Adders, X. Jiao¹, V. Camus², M. Cacciotti², Y. Jiang³, C. Enz² and R. Gupta¹
¹UC San Diego, USA; ²EPFL, Switzerland; ³Tsinghua University, China

DVAFS: Trading computational accuracy for energy through Dynamic-Voltage-Accuracy-Frequency-Scaling
B. Moons, R. Uytterhoeven, W. Dehaene and M. Verhelst, KU Leuven, Belgium

Exploiting computation skip to reduce energy consumption by approximate computing, an HEVC encoder case study, A. Mercat¹, J. Bonnot1, M. Pelcat¹², W. Hamidouche¹ and D. Menard¹
¹INSA Rennes – IETR, France, ²Institut Pascal, Clermont Ferrand, France

Location detection for navigation using IMUs with a map through coarse-grained machine learning J. Gonzales E.¹, C. Luo¹, A. Shrivastava¹, K. Palem¹, M. Yongshik², S. Noh², D. Park², S. Hong²
¹Rice University, USA; ²Seoul National University, Korea