CIRCA: Towards a modular and extensible framework for approximate circuit generation. (August 2019)

• Record Type: Journal Article
• Authors: Witschen, Linus; Awais, Muhammad; Ghasemzadeh Mohammadi, Hassan; Wiersema, Tobias; Platzner, Marco
• Is Part Of: Microelectronics and reliability. Volume 99(2019)
• Page Start: 277
• Page End: 290
• Publication Date: 2019-08
• Subjects: Approximate Computing -- Framework -- Pareto Front -- Accuracy -- Stochastic search -- Circuit synthesis; Electronic apparatus and appliances -- Reliability -- Periodicals; Miniature electronic equipment -- Periodicals; Appareils électroniques -- Fiabilité -- Périodiques; Équipement électronique miniaturisé -- Périodiques; Electronic apparatus and appliances -- Reliability; Miniature electronic equipment; Periodicals; 621.3815
• DOI: 10.1016/j.microrel.2019.04.003
• Languages: English
• ISSNs: 0026-2714

Abstract

Abstract: Existing approaches and tools for the generation of approximate circuits often lack generality and are restricted to certain circuit types, approximation techniques, and quality assurance methods. Moreover, only few tools are publicly available. This hinders the development and evaluation of new techniques for approximating circuits and their comparison to previous approaches. In this paper, we first analyze and classify related approaches and then present CIRCA, our flexible framework for search-based approximate circuit generation. CIRCA is developed with a focus on modularity and extensibility to foster comparability of different techniques. We present the architecture of CIRCA with its distinct separation into stages and processing blocks. This separation makes CIRCA flexible, allowing developers to extend it by new methods and enabling users to exchange the employed functionality quickly. To highlight CIRCA's benefits and its flexibility, we conduct a case study which focuses on the impact of the employed search method and approximation technique. We use a set of circuits from our approximate computing benchmark suite PaderBench to evaluate and compare the quality of the approximated circuits generated with the different methods. Highlights: Concept and development of the general, modular, and extensible approximation framework for approximate circuit generation CIRCA Compilation of a benchmark suite for approximate computing named PaderBench Experimental case study on the impact of different search methods and approximation techniques on the quality of approximated circuits