Reproduce “Approximate Computing Survey, Part I: Terminology and Software & Hardware Approximation Techniques” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Vasileios Leon, Muhammad Abdullah Hanif, Giorgos Armeniakos, Xun Jiao, Muhammad Shafique, Kiamal Pekmestzi, Dimitrios Soudris
This survey paper provides a comprehensive review of Approximate Computing, establishing foundational terminology and principles while systematically classifying and detailing state-of-the-art software and hardware approximation techniques. It quantitatively analyzes these methods, demonstrating significant energy and performance gains achieved by trading off accuracy within acceptable limits across various workloads.
Key insights
- Exploiting the intrinsic error resilience of applications allows for significant trade-offs between computational accuracy and resource efficiency (power, energy, area, latency).
- A systematic, cross-layer approach is crucial for maximizing the benefits of approximation, combining techniques from application and software layers down to hardware and circuit levels.
- Different approximation techniques yield varying benefits and acceptable quality degradation, necessitating careful selection and tuning based on workload characteristics and specific application quality-of-service requirements.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.