Autonomic Parallelism Adaptation for Software Transactional Memory

Naweiluo Zhou 1, 2, 3 Gwenaël Delaval 1, 2 Bogdan Robu 4 Éric Rutten 1, 2 Jean-François Méhaut 2, 3
1 CTRL-A - Control techniques for Autonomic, adaptive and Reconfigurable Computing systems
Inria Grenoble - Rhône-Alpes, LIG - Laboratoire d'Informatique de Grenoble
3 CORSE - Compiler Optimization and Run-time Systems
Inria Grenoble - Rhône-Alpes, LIG - Laboratoire d'Informatique de Grenoble
4 GIPSA-SYSCO - SYSCO
GIPSA-DA - Département Automatique
Abstract : Parallel programs need to manage the time trade-off between synchronization and computation. A high parallelism may decrease computing time but meanwhile increase synchronization cost. Software Transactional Memory (STM) has emerged as a promising technique, which bypasses locks, to address synchronization issues through transactions. A way to reduce conflicts is by adjusting parallelisms. However, there is no universal rule to decide the best parallelism for a program from an offline view. Furthermore, an offline tuning is costly and error-prone. Hence, it becomes necessary to adopt a dynamical tuning-configuration strategy to better manage a STM system. Autonomic computing offers designers a framework of methods and techniques to build systems with well-mastered behaviours. Its key idea is to implement feedback control loops to design safe, efficient and predictable controllers, which enable monitoring and adjusting controlled systems dynamically while keeping overhead low.
Type de document :
Communication dans un congrès
Conférence d’informatique en Parallélisme, Architecture et Système (COMPAS), Jul 2016, Lorient, France
Liste complète des métadonnées

https://hal.inria.fr/hal-01312786
Contributeur : N Zhou <>
Soumis le : lundi 9 mai 2016 - 10:17:24
Dernière modification le : jeudi 11 octobre 2018 - 08:48:05

Identifiants

  • HAL Id : hal-01312786, version 1

Citation

Naweiluo Zhou, Gwenaël Delaval, Bogdan Robu, Éric Rutten, Jean-François Méhaut. Autonomic Parallelism Adaptation for Software Transactional Memory. Conférence d’informatique en Parallélisme, Architecture et Système (COMPAS), Jul 2016, Lorient, France. 〈hal-01312786〉

Partager

Métriques

Consultations de la notice

434