Skip to Main content Skip to Navigation
Conference papers

Exposer les caractéristiques des architectures à mémoires hétérogènes aux applications parallèles

Andrès Rubio Proaño 1
1 TADAAM - Topology-Aware System-Scale Data Management for High-Performance Computing
LaBRI - Laboratoire Bordelais de Recherche en Informatique, Inria Bordeaux - Sud-Ouest
Abstract : The complexity of memory systems has increased considerably over the past decade. Consequently, the supercomputers include memories at several levels, heterogeneous and non-uniform, with significantly different properties. Developers of scientific applications face a huge challenge: to harness the memory system efficiently to improve performance and productivity. In this work, we present an interface to manage the complexity of the memory system, composed of a set of memory attributes and an API to express and manage these various characteristics using metrics, for example bandwidth, latency and capacity. It allows runtime systems, parallel libraries and scientific applications to select the appropriate memory by expressing their needs for each allocation without having to modify the code for each platform.
Complete list of metadatas

Cited literature [11 references]  Display  Hide  Download

https://hal.inria.fr/hal-02639607
Contributor : Andrès Rubio Proaño <>
Submitted on : Thursday, May 28, 2020 - 12:29:49 PM
Last modification on : Tuesday, June 2, 2020 - 11:20:16 AM

File

paper.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02639607, version 1

Collections

Citation

Andrès Rubio Proaño. Exposer les caractéristiques des architectures à mémoires hétérogènes aux applications parallèles. COMPAS 2020 - Conférence francophone d'informatique en Parallélisme, Architecture et Système, Jun 2020, Lyon, France. ⟨hal-02639607⟩

Share

Metrics

Record views

41

Files downloads

88