Automatic adaptive approximation for stencil computations

Maxime Schmitt (1) , Philippe Helluy (2) , Cédric Bastoul (3)

1 ICube - Laboratoire des sciences de l'ingénieur, de l'informatique et de l'imagerie
2 IRMA - Institut de Recherche Mathématique Avancée
3 CAMUS - Compilation pour les Architectures MUlti-coeurS

Approximate computing is necessary to meet deadlines in some compute-intensive applications like simulation. Building them requires a high level of expertise from the application designers as well as a significant development effort. Some application programming interfaces greatly facilitate their conception but they still heavily rely on the developer's domain-specific knowledge and require many modifications to successfully generate an approximate version of the program. In this paper we present new techniques to semi-automatically discover relevant approximate computing parameters. We believe that superior compiler-user interaction is the key to improved productivity. After pinpointing the region of interest to optimize, the developer is guided by the compiler in making the best implementation choices. Static analysis and runtime monitoring are used to infer approximation parameter values for the application. We evaluated these techniques on multiple application kernels that support approximation and show that with the help of our method, we achieve similar performance as non-assisted, hand-tuned version while requiring minimal intervention from the user.

Mots clés: Approximate Computing; Compilation; Code Optimization; Application Programming Interface; Stencil Code

Domaines: Langage de programmation [cs.PL]; Performance et fiabilité [cs.PF]

Maxime Schmitt, Philippe Helluy, Cédric Bastoul. Automatic adaptive approximation for stencil computations. CC 2019 - 28th International Conference on Compiler Construction, Feb 2019, Washington, United States. pp.170-181, ⟨10.1145/3302516.3307348⟩. ⟨hal-02072737⟩