High-performance code generation for stencil computations on heterogeneous multi-device architectures
| hal.structure.identifier | Département Informatique [TSP - INF] | |
| dc.contributor.author | LI, Pei | |
| hal.structure.identifier | Département Informatique [TSP - INF] | |
| dc.contributor.author | BRUNET, Elisabeth | |
| hal.structure.identifier | Efficient runtime systems for parallel architectures [RUNTIME] | |
| dc.contributor.author | NAMYST, Raymond | |
| dc.date.accessioned | 2024-04-15T09:41:41Z | |
| dc.date.available | 2024-04-15T09:41:41Z | |
| dc.date.issued | 2013 | |
| dc.date.conference | 2013-11-13 | |
| dc.identifier.uri | https://oskar-bordeaux.fr/handle/20.500.12278/197620 | |
| dc.description.abstractEn | Stencil computations are widely used in many scientific domains, and are therefore of particular interest to scientific computing research. Running stencil codes on GPGPUs and multi-devices platforms are theoretically possible for gaining better performance because of their superior floating point performance and memory bandwidth. But these architectures increase the difficulties for developers. In this paper, we propose an tool that generates OpenCL host program for stencil computing on multi-device architecture. The user provides the description of critical parameters in a XML file, then our generator will analyze the description and generate automatically the host program. Due to the data partition and data exchange strategies, the generated host program can be executed on multi-devices without changing any kernel code. The experiment shows that our tool is efficient. It can guarantee the minimum data exchanges and achieve high performance on Multi-device architecture. | |
| dc.language.iso | en | |
| dc.publisher | IEEE | |
| dc.title.en | High-performance code generation for stencil computations on heterogeneous multi-device architectures | |
| dc.type | Communication dans un congrès | |
| dc.subject.hal | Informatique [cs] | |
| bordeaux.hal.laboratories | Laboratoire Bordelais de Recherche en Informatique (LaBRI) - UMR 5800 | * |
| bordeaux.institution | Université de Bordeaux | |
| bordeaux.institution | Bordeaux INP | |
| bordeaux.institution | CNRS | |
| bordeaux.conference.title | HPPC 2013 - 15th IEEE International Conference on High Performance Computing and Communications | |
| bordeaux.country | CN | |
| bordeaux.conference.city | Zhangjiajie | |
| bordeaux.peerReviewed | oui | |
| hal.identifier | hal-00952258 | |
| hal.version | 1 | |
| hal.invited | non | |
| hal.proceedings | oui | |
| hal.conference.end | 2013-11-15 | |
| hal.popular | non | |
| hal.audience | Internationale | |
| dc.subject.it | Stencil | |
| dc.subject.it | Code generation | |
| dc.subject.it | GPGPU | |
| dc.subject.it | OpenCL | |
| hal.origin.link | https://hal.archives-ouvertes.fr//hal-00952258v1 | |
| bordeaux.COinS | ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.date=2013&rft.au=LI,%20Pei&BRUNET,%20Elisabeth&NAMYST,%20Raymond&rft.genre=unknown |
Fichier(s) constituant ce document
| Fichiers | Taille | Format | Vue |
|---|---|---|---|
|
Il n'y a pas de fichiers associés à ce document. |
|||