PEPC

Principal Investigator: 

P. Gibbon
Simulation Laboratory Plasma Physics,
Jülich Supercomputing Centre
Email: pepc@fz-juelich.de

Other application users/developers: 

Imperial College London, University of Illinois, FZ Jülich, MPIfR Bonn, TU Chemnitz, Institute of Physics ASCR Prague and many more.

Scientific area: 
N-body simulations (Coulomb, Gravitational, …) beam-plasma interaction, vortex dynamics, gravitational interaction, MD
Abstract: 
PEPC is a tree code for solving the N-body problem. It is not restricted to Coulomb systems but also handles gravitation and hydrodynamics using the vortex method as well as smooth particle hydrodynamics (SPH). PEPC is a non-recursive version of the Barnes-Hut algorithm with a level-by-level approach to both tree construction and traversals. The parallel version is a hybrid MPI/PThreads implementation of the Warren-Salmon 'Hashed Oct-Tree' scheme. The long range interactions are computed using multipole groupings of distant particles to reduce computational effort. The Barnes-Hut algorithm is well suited to dynamic, nonlinear problems and can be combined with multiple-timestep integrators.
Scalability: 
294912 cores on BlueGene/P 458752 cores on BlueGene/Q The code has been qualified for being part of the High-Q Club which gathers the highest scaling codes on JUQUEEN (IBM BG/Q @ JSC, Germany) with a scaling on the full system i.e 458,752 cores (1,668,196 parallel threads) on BlueGene/Q
Versions: 
MPI
Other: 
pthreads, OmpSs
Tested on platforms: 

BlueGene/P, BlueGene/Q, x86, ARM

Portable to all existing HPC platforms

 

Images: 
Any other information: 

[1] M. Winkel, R. Speck, H. Hübner, L. Arnold, R. Krause, P. Gibbon
"A massively parallel, multi-disciplinary Barnes-Hut tree code for
extreme-scale N-body simulations"
Computer Physics Communications, Vol. 183 (4), April 2012