A new execution model for improving performance and flexibility of CAPE

Long Tran Van (1, 2, 3, 4) , Eric Renault (5, 1, 2, 4) , Xuan Huyen Do (6) , Viet Hai Ha (7)

1 SAMOVAR - Services répartis, Architectures, MOdélisation, Validation, Administration des Réseaux
2 TSP - RS2M - Département Réseaux et Services Multimédia Mobiles
3 HUEIC - Hue Industrial College (.)
4 CNRS - Centre National de la Recherche Scientifique
5 R3S-SAMOVAR - Réseaux, Systèmes, Services, Sécurité
6 College of Sciences [Hue] (Hue University)
7 College of Education [Hue] (Hue University)

Checkpointing-Aided Parallel Execution (CAPE) is a framework that is based on checkpointing technique and serves to automatically translates and execute OpenMP programs on distributed-memory architectures. In some comparisons with MPI, CAPE have demonstrated high-performance and the potential for fully compatibility with OpenMP on distributed-memory systems. However, it should be continued to improve the performance, flexibility, portability and capability. This paper presents the new execution model for CAPE that improves its performance and makes CAPE even more flexible

Mots clés: Computational modeling; Task analysis; Synchronization; Computer architecture; Tools; Dynamic scheduling; Adaptation models

Domaines: Réseaux et télécommunications [cs.NI]

Long Tran Van, Eric Renault, Xuan Huyen Do, Viet Hai Ha. A new execution model for improving performance and flexibility of CAPE. PDP 2018 : 26th Euromicro International Conference on Parallel, Distributed and Network-based Processing, Mar 2018, Cambridge, United Kingdom. pp.234 - 238, ⟨10.1109/PDP2018.2018.00039⟩. ⟨hal-01868501⟩