An introduction to approximate computing

TL;DR

This paper provides a comprehensive introduction to approximate computing, proposing a taxonomy to categorize techniques based on system-wide costs, and discusses its opportunities and challenges.

Contribution

It introduces a taxonomy for approximate computing techniques emphasizing system-level costs and discusses the unique challenges and opportunities of nondeterministic approaches.

Findings

Taxonomy categorizes approximation techniques by system cost

Highlights opportunities in energy-efficient computing

Identifies challenges in nondeterministic approximation

Abstract

Approximate computing is a research area where we investigate a wide spectrum of techniques to trade off computation accuracy for better performance or energy consumption. In this work, we provide a general introduction to approximate computing. Also, we propose a taxonomy to make it easier to discuss the merits of different approximation techniques. Our taxonomy emphasizes the expected cost of tackling approximate computing across the entire system stack. We conclude by discussing the unique opportunities as well as challenges of nondeterministic approximate computing.