Geodesic
Using a scheduler for efficient data exchange on hybrid supercomputers with massively-parallel accelerators
Numerical methods and programming, Tome 14 (2013) no. 4, pp. 122-134

Voir la notice de l'article provenant de la source Math-Net.Ru

An optimization of communications within multi-level parallelization based on a combination of MPI, OpenMP and OpenCL is proposed to fit all kinds of modern supercomputer architectures including hybrid systems with GPU and Intel Xeon Phi accelerators. A general-purpose scheduler that simplifies a heterogeneous implementation is discussed. The scheduler controls queues of computing and communication OpenCL tasks. It uses an interdependency graph of a target computing algorithm as input. The use of the scheduler is demonstrated on an example of a finite-volume CFD algorithm for unstructured meshes. In particular, the scheduler has been used to simplify an implementation of an overlapped communication scheme. The implementation of the CFD algorithm with MPI and CPU-GPU communications overlapped with computations is described and its parallel efficiency is demonstrated.
Keywords: gas dynamics, scheduler, parallel computing, GPU, OpenCL, MPI. 
@article{VMP_2013_14_4_a16,
     author = {P. B. Bogdanov and A. A. Efremov and A. V. Gorobets and S. A. Sukov},
     title = {Using a scheduler for efficient data exchange on hybrid supercomputers with massively-parallel accelerators},
     journal = {Numerical methods and programming},
     pages = {122--134},
     publisher = {mathdoc},
     volume = {14},
     number = {4},
     year = {2013},
     language = {ru},
     url = {http://geodesic.mathdoc.fr/item/VMP_2013_14_4_a16/}
}
TY  - JOUR
AU  - P. B. Bogdanov
AU  - A. A. Efremov
AU  - A. V. Gorobets
AU  - S. A. Sukov
TI  - Using a scheduler for efficient data exchange on hybrid supercomputers with massively-parallel accelerators
JO  - Numerical methods and programming
PY  - 2013
SP  - 122
EP  - 134
VL  - 14
IS  - 4
PB  - mathdoc
UR  - http://geodesic.mathdoc.fr/item/VMP_2013_14_4_a16/
LA  - ru
ID  - VMP_2013_14_4_a16
ER  - 
%0 Journal Article
%A P. B. Bogdanov
%A A. A. Efremov
%A A. V. Gorobets
%A S. A. Sukov
%T Using a scheduler for efficient data exchange on hybrid supercomputers with massively-parallel accelerators
%J Numerical methods and programming
%D 2013
%P 122-134
%V 14
%N 4
%I mathdoc
%U http://geodesic.mathdoc.fr/item/VMP_2013_14_4_a16/
%G ru
%F VMP_2013_14_4_a16
P. B. Bogdanov; A. A. Efremov; A. V. Gorobets; S. A. Sukov. Using a scheduler for efficient data exchange on hybrid supercomputers with massively-parallel accelerators. Numerical methods and programming, Tome 14 (2013) no. 4, pp. 122-134. http://geodesic.mathdoc.fr/item/VMP_2013_14_4_a16/