Adapting the Polyhedral Model as a Framework for Efficient Speculative Parallelization

Abstract : In this paper, we present a Thread-Level Speculation (TLS) framework whose main feature is to be able to speculatively parallelize a sequential loop nest in various ways, by re-scheduling its iterations. The transformation to be applied is selected at runtime with the goal of minimizing the number of rollbacks and maximizing performance. We perform code transformations by applying the polyhedral model that we adapted for speculative and runtime code parallelization. For this purpose, we designed a parallel code pattern which is patched by our runtime system according to the profiling information collected on some execution samples. Adaptability is ensured by considering chunks of code of various sizes, that are launched successively, each of which being parallelized in a different manner, or run sequentially, depending on the currently observed behavior for accessing memory.
Type de document :
Communication dans un congrès
PPoPP - 17th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, Feb 2012, New Orleans, United States. ACM Press, 2012, 17th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming - PPoPP'12
Liste complète des métadonnées

https://hal.inria.fr/hal-00664353
Contributeur : Philippe Clauss <>
Soumis le : lundi 30 janvier 2012 - 14:00:02
Dernière modification le : vendredi 12 janvier 2018 - 01:10:57

Identifiants

  • HAL Id : hal-00664353, version 1

Collections

Citation

Alexandra Jimborean, Philippe Clauss, Benoit Pradelle, Luis Mastrangelo, Vincent Loechner. Adapting the Polyhedral Model as a Framework for Efficient Speculative Parallelization. PPoPP - 17th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, Feb 2012, New Orleans, United States. ACM Press, 2012, 17th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming - PPoPP'12. 〈hal-00664353〉

Partager

Métriques

Consultations de la notice

238