Automatic Approximation of Computer Systems Through Multi-objective Optimization

Mario Barbareschi (1) , Salvatore Barone (1) , Alberto Bosio (2) , Marcello Traiola (3)

1 University of Naples Federico II = Università degli studi di Napoli Federico II
2 INL - CSH - INL - Conception de Systèmes Hétérogènes
3 TARAN - Architectures matérielles spécialisées pour l’ère post loi-de-Moore

In this chapter, we address the automatic approximation of computer systems through multi-objective optimization. Firstly, we present our automatic design methodology, i.e., how we model the approximate design space to be automatically explored. The exploration is achieved through multi-objective optimization to find good trade-offs between the system efficiency and accuracy. Then, we show how the methodology is applied to the systematic and application-independent design of generic combinational logic circuits, based on non-trivial local rewriting of and-inverter graphs (AIGs). Finally, to push forward the approximation limits, we showcase the design of approximate hardware accelerators for image processing and for common machine-learning-based classification models.

Mots clés: Design space exploration; Multi-objective optimization; Discrete cosine transform; Approximate circuits; Approximate neural networks; Approximate decision trees; Approximate computing; Genetic algorithm

Domaines: Systèmes embarqués; Architectures Matérielles [cs.AR]

Mario Barbareschi, Salvatore Barone, Alberto Bosio, Marcello Traiola. Automatic Approximation of Computer Systems Through Multi-objective Optimization. Design and Applications of Emerging Computer Systems, Springer Nature Switzerland, pp.383-420, 2024, ⟨10.1007/978-3-031-42478-6_15⟩. ⟨hal-04396685⟩