Geodesic
Evaluating Kernels on Xeon Phi to accelerate Gysela application
Guillaume Latu1 ; Matthieu Haefele2, 3 ; Julien Bigot3 ; Virginie Grandgirard1 ; Thomas Cartier-Michaud1 ; Fabien Rozar1, 3
1CEA, IRFM, F-13108 Saint-Paul-lez-Durance
2IPP, Boltzmannstrasse 2, D-85748 Garching
3Maison de la Simulation, CEA/CNRS/Inria/Univ. Paris-Sud/Univ. de Versailles, F-91191, Gif-sur-Yvette Cedex
ESAIM. Proceedings, Tome 53 (2016), pp. 211-231
Cet article a éte moissonné depuis la source EDP Sciences

Voir la notice de l'article

This work describes the challenges presented by porting parts of the Gysela code to the Intel Xeon Phi coprocessor, as well as techniques used for optimization, vectorization and tuning that can be applied to other applications. We evaluate the performance of some generic micro-benchmark on Phi versus Intel Sandy Bridge. Several interpolation kernels useful for the Gysela application are analyzed and the performances are shown. Some memory-bound and compute-bound kernels are accelerated by a factor 2 on the Phi device compared to Sandy architecture. Nevertheless, it is hard, if not impossible, to reach a large fraction of the peak performance on the Phi device, especially for real-life applications as Gysela. A collateral benefit of this optimization and tuning work is that the execution time of Gysela (using 4D advections) has decreased on a standard architecture such as Intel Sandy Bridge.
DOI : 10.1051/proc/201653013

Guillaume Latu  1   ; Matthieu Haefele  2 , 3   ; Julien Bigot  3   ; Virginie Grandgirard  1   ; Thomas Cartier-Michaud  1   ; Fabien Rozar  1 , 3

1 CEA, IRFM, F-13108 Saint-Paul-lez-Durance
2 IPP, Boltzmannstrasse 2, D-85748 Garching
3 Maison de la Simulation, CEA/CNRS/Inria/Univ. Paris-Sud/Univ. de Versailles, F-91191, Gif-sur-Yvette Cedex 
@article{EP_2016_53_a13,
     author = {Guillaume Latu and Matthieu Haefele and Julien Bigot and Virginie Grandgirard and Thomas Cartier-Michaud and Fabien Rozar},
     title = {Evaluating {Kernels} on {Xeon} {Phi} to accelerate {Gysela} application},
     journal = {ESAIM. Proceedings},
     pages = {211--231},
     year = {2016},
     volume = {53},
     doi = {10.1051/proc/201653013},
     language = {en},
     url = {http://geodesic.mathdoc.fr/articles/10.1051/proc/201653013/}
}
TY  - JOUR
AU  - Guillaume Latu
AU  - Matthieu Haefele
AU  - Julien Bigot
AU  - Virginie Grandgirard
AU  - Thomas Cartier-Michaud
AU  - Fabien Rozar
TI  - Evaluating Kernels on Xeon Phi to accelerate Gysela application
JO  - ESAIM. Proceedings
PY  - 2016
SP  - 211
EP  - 231
VL  - 53
UR  - http://geodesic.mathdoc.fr/articles/10.1051/proc/201653013/
DO  - 10.1051/proc/201653013
LA  - en
ID  - EP_2016_53_a13
ER  - 
%0 Journal Article
%A Guillaume Latu
%A Matthieu Haefele
%A Julien Bigot
%A Virginie Grandgirard
%A Thomas Cartier-Michaud
%A Fabien Rozar
%T Evaluating Kernels on Xeon Phi to accelerate Gysela application
%J ESAIM. Proceedings
%D 2016
%P 211-231
%V 53
%U http://geodesic.mathdoc.fr/articles/10.1051/proc/201653013/
%R 10.1051/proc/201653013
%G en
%F EP_2016_53_a13
Guillaume Latu; Matthieu Haefele; Julien Bigot; Virginie Grandgirard; Thomas Cartier-Michaud; Fabien Rozar. Evaluating Kernels on Xeon Phi to accelerate Gysela application. ESAIM. Proceedings, Tome 53 (2016), pp. 211-231. doi: 10.1051/proc/201653013

Cité par Sources :