Title: Harnessing aspect-oriented programming on GPU: application to warp-level parallelism

Authors: Jonathan Passerat-Palmbach; Jonathan Caux; Pierre Schweitzer; Pridi Siregar; Claude Mazel; David R.C. Hill

Addresses: ISIMA – Institut Supérieur d'Informatique, de Modélisation et de leurs Applications, F-63173 Aubière, France; Clermont Université, Université Blaise Pascal, LIMOS, CNRS, UMR 6158, BP 10448, F-63000 Clermont-Ferrand, France ' ISIMA – Institut Supérieur d'Informatique, de Modélisation et de leurs Applications, F-63173 Aubière, France; Clermont Université, Université Blaise Pascal, LIMOS, CNRS, UMR 6158, BP 10448, F-63000 Clermont-Ferrand, France; Murex S.A.S., F-75782, Paris, France ' Clermont Université, Université Blaise Pascal, LIMOS, CNRS, UMR 6158, BP 10448, F-63000 Clermont-Ferrand, France; Clermont Université, Université Blaise Pascal, LPC, CNRS, UMR 6533, BP 10448, F-63000 Clermont-Ferrand, France ' Integrative Biocomputing, Nouvelles Structures – Place du Granier, F-35135, Chantepie, France ' ISIMA – Institut Supérieur d'Informatique, de Modélisation et de leurs Applications, F-63173 Aubière, France; Clermont Université, Université Blaise Pascal, LIMOS, CNRS, UMR 6158, BP 10448, F-63000 Clermont-Ferrand, France ' ISIMA – Institut Supérieur d'Informatique, de Modélisation et de leurs Applications, F-63173 Aubière, France; Clermont Université, Université Blaise Pascal, LIMOS, CNRS, UMR 6158, BP 10448, F-63000 Clermont-Ferrand, France

Abstract: Stochastic simulations involve multiple replications in order to build confidence intervals for their results, and designs of experiments (DOEs) to explore their parameters set. In this paper, we propose warp-level parallelism (WLP), a GPU-enabled solution to compute multiple replications in parallel (MRIP) on GPUs (graphics processing units). GPUs are intrinsically tuned to process efficiently the same operation on several data, which is not suited to parallelise MRIP or DOEs. Our approach proposes to rely on small thread groups, called warps, to perform independent computations such as replications. This approach has proved to be efficient on three classical simulation models, but originally lacked the transparency users might expect. In this work, we enhance WLP using aspect oriented programming (AOP). Our work describes the way to combine CUDA and AOP, and brings forward the techniques available to exploit AOP in a CUDA-enabled development.

Keywords: aspect-oriented programming; AOP; stochastic modelling; simulation; multiple replications in parallel; MRIP; graphics processing unit; GPU; compute unified device architecture; CUDA; warp-level parallelism; WLP.

DOI: 10.1504/IJCAET.2015.068335

International Journal of Computer Aided Engineering and Technology, 2015 Vol.7 No.2, pp.158 - 175

Published online: 31 Mar 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article