Geodesic
Optimizing the parallel scheme of the Poisson solver for the reduced kinetic code TERESA
Thomas Cartier-Michaud1 ; Philippe Ghendrih1 ; Virginie Grandgirard1 ; Guillaume Latu1
1 CEA Cadarache, IRFM, F-13108 Saint-Paul-lez-Durance, France
ESAIM. Proceedings, Tome 43 (2013), pp. 274-294.

Voir la notice de l'article provenant de la source EDP Sciences

The parallelization performance of the TERESA Code for Trapped Element REduction in Semi lagrangian Approach is analyzed. TERESA is a kinetic code in four dimensions (4D), two in "real" space and two "energy" coordinates. It addresses the turbulent evolution of the distribution of ions governed by uctuations of the electric potential. The numerical scheme is split into four steps: the 4D advection of the distribution function with the Vlasov equation, the computation of the charge density, the calculation of the electric potential using the quasineutrality asymptotic limit of the Poisson equation and finally, the two high frequency averages of the electric potential in order to compute the advection field. Starting from an initial standard parallelization scheme we find that the Poisson solver accounts for most of the execution time of TERESA (up to 90% at aimed meshes) and that its scaling performance is poor. Two improvements of the Poisson solver have been implemented and tested. These show that performance levelling are due to interference phenomena when addressing computation resources as well as communication overheads when the loading of the CPUs with MPI processes / OpenMP threads is not optimized. Tests on two computers, Rheticus et Helios, with very similar architecture, allows one to better diagnose the interference phenomena. These developments lead to reduction by up to a factor 100 of the execution time and extends the scalability properties up to 1024 cores for large meshes cases.
DOI : 10.1051/proc/201343017

Thomas Cartier-Michaud 1 ; Philippe Ghendrih 1 ; Virginie Grandgirard 1 ; Guillaume Latu 1

1 CEA Cadarache, IRFM, F-13108 Saint-Paul-lez-Durance, France 
@article{EP_2013_43_a17,
     author = {Thomas Cartier-Michaud and Philippe Ghendrih and Virginie Grandgirard and Guillaume Latu},
     title = {Optimizing the parallel scheme of the {Poisson} solver for the reduced kinetic code {TERESA}},
     journal = {ESAIM. Proceedings},
     pages = {274--294},
     publisher = {mathdoc},
     volume = {43},
     year = {2013},
     doi = {10.1051/proc/201343017},
     language = {en},
     url = {http://geodesic.mathdoc.fr/articles/10.1051/proc/201343017/}
}
TY  - JOUR
AU  - Thomas Cartier-Michaud
AU  - Philippe Ghendrih
AU  - Virginie Grandgirard
AU  - Guillaume Latu
TI  - Optimizing the parallel scheme of the Poisson solver for the reduced kinetic code TERESA
JO  - ESAIM. Proceedings
PY  - 2013
SP  - 274
EP  - 294
VL  - 43
PB  - mathdoc
UR  - http://geodesic.mathdoc.fr/articles/10.1051/proc/201343017/
DO  - 10.1051/proc/201343017
LA  - en
ID  - EP_2013_43_a17
ER  - 
%0 Journal Article
%A Thomas Cartier-Michaud
%A Philippe Ghendrih
%A Virginie Grandgirard
%A Guillaume Latu
%T Optimizing the parallel scheme of the Poisson solver for the reduced kinetic code TERESA
%J ESAIM. Proceedings
%D 2013
%P 274-294
%V 43
%I mathdoc
%U http://geodesic.mathdoc.fr/articles/10.1051/proc/201343017/
%R 10.1051/proc/201343017
%G en
%F EP_2013_43_a17
Thomas Cartier-Michaud; Philippe Ghendrih; Virginie Grandgirard; Guillaume Latu. Optimizing the parallel scheme of the Poisson solver for the reduced kinetic code TERESA. ESAIM. Proceedings, Tome 43 (2013), pp. 274-294. doi : 10.1051/proc/201343017. http://geodesic.mathdoc.fr/articles/10.1051/proc/201343017/

Cité par Sources :