Profile-based Vectorization for MATLAB

Patryk Kiepas; Jaroslaw Kozlak; Claude Tadonki; Corinne Ancourt

Profile-based Vectorization for MATLAB

Patryk Kiepas ^{1, 2, 3} Jaroslaw Kozlak ³ Claude Tadonki ^{1, 2} Corinne Ancourt ^{2, 1}

1 MINES ParisTech - École nationale supérieure des mines de Paris

2 CRI - Centre de Recherche en Informatique

3 Cracovia AGH University of Mining and Metallurgy

Abstract : In recent years, MATLAB just-in-time (JIT) interpreter has improved the execution time of for-loops to a level where loops can outperform equivalent array operations in some scenarios. This has allowed systematic translation of loops to array operations, a prevalent approach for performance improvement in MATLAB, to sometimes yield a performance loss. Therefore, we propose a selective strategy to loop translation with selection criteria guided by loop profiling data. As a result, only loops with a high performance speedup potential are selected for translation to array operations. Our experimental results confirm the efficiency of our approach and illustrate the cases where systematic translation lead to performance degradation.

Keywords : vectorization selective strategy performance optimization MATLAB array language

Type de document :

Communication dans un congrès

Domaine :

Informatique [cs] / Calcul parallèle, distribué et partagé [cs.DC]

Informatique [cs] / Langage de programmation [cs.PL]

Liste complète des métadonnées

https://hal-mines-paristech.archives-ouvertes.fr/hal-01857852
Contributeur : Claire Medrala <>
Soumis le : vendredi 17 août 2018 - 14:49:02
Dernière modification le : jeudi 14 mars 2019 - 11:48:50

Profile-based Vectorization for MATLAB

Identifiants

Collections

Citation

Exporter

Métriques

105

Contact

Profile-based Vectorization for MATLAB

Identifiants

Collections

Citation

Exporter

Partager

Métriques

105

Contact