Wed 15 Jun 2016 14:30 - 15:00 at Grand Ballroom San Rafael - Energy & Performance Chair(s): Manuel Hermenegildo

This paper introduces Input Responsive Approximation (IRA), an approach that uses a canary input — a small program input carefully constructed to capture the intrinsic properties of the original input — to automatically control how approximation is applied on an input-by-input basis for approximate programs. Motivating this approach is the observation that many of the prior techniques focusing on choosing how to approximate arrive at conservative decisions by discounting substantial differences between inputs when applying approximation. The main challenges in overcoming this limitation lie in making the choice of how to approximate both effectively (e.g., the fastest approximation that meets a particular accuracy target) and rapidly for every input. With IRA, each time the approximate program is run, a canary input is constructed and used dynamically to quickly test a spectrum of approximation alternatives. Based on these runtime tests, the approximation that best fits the desired accuracy constraints is selected and applied to the full input to produce an approximate result. We use IRA to select and parameterize mixes of four approximation techniques from the literature for a range of 13 image processing, machine learning, and data mining applications. Our results demonstrate that IRA significantly outperforms prior approaches, delivering an average of 10.2× speedup over exact execution while minimizing accuracy losses in program outputs.

Wed 15 Jun

Displayed time zone: Tijuana, Baja California change

13:30 - 15:00
Energy & PerformanceResearch Papers at Grand Ballroom San Rafael
Chair(s): Manuel Hermenegildo IMDEA Software Institute and T.U. of Madrid (UPM)
13:30
30m
Talk
Effective Padding of Multi-Dimensional Arrays to Avoid Cache Conflict Misses
Research Papers
Changwan Hong , Wenlei Bao , Albert Cohen INRIA, Sriram Krishnamoorthy Pacific Northwest National Laboratories, Louis-Noël Pouchet Ohio State University, J. Ramanujam Louisiana State University, Fabrice Rastello INRIA, France, P. Sadayappan Ohio State University
Media Attached
14:00
30m
Talk
GreenWeb: Language Extensions for Energy-Efficient Mobile Web Computing
Research Papers
Link to publication Media Attached
14:30
30m
Talk
Input Responsiveness: Using Canary Inputs to Dynamically Steer Approximation
Research Papers
Michael A. Laurenzano University of Michigan, Parker Hill , Mehrzad Samadi University of Michigan, Scott Mahlke University of Michigan, Jason Mars University of Michigan, Lingjia Tang University of Michigan
Media Attached