|Name:||(09) Particle-in-Cell Plasma Simulation on CPUs, GPUs & Xeon Phi Coprocessors|
|Time:||Monday, June 23, 2014
05:18 pm - 05:25 pm
CCL - Congress Center Leipzig
|Speaker:||Sergey Bastrakov, N.I. Lobachevsky University of Nizhni Novgorod|
|Abstract:||Simulation of plasma dynamics with the Particle-in-Cell method is one of the currently high-demand areas of computational physics. Solving up-to-date problems requires large-scale plasma simulation. Given the growing popularity of GPGPU and the advent of Intel Xeon Phi coprocessors there is an interest in high-performance implementation of the method for heterogeneous systems. PICADOR is a tool for three-dimensional plasma simulation using the Particle-in-Cell method. Features of PICADOR include FDTD and NDF field solvers, Boris particle pusher, CIC and TSC particle form factors, Esirkepov current deposition, ionization, and moving frame. The code is extendable in terms of additional stages, is capable of using modern cluster systems with CPUs, GPUs and Xeon Phi coprocessors, and supports dynamic load balancing.
On a simulation of dense plasma in double precision PICADOR achieves 15 nanoseconds per particle update on 8-core Intel Xeon E2690 CPU with 99% strong scaling efficiency on shared memory, which is competitive to the state-of-the-art. A Xeon Phi coprocessor in the native mode outperforms a Xeon E2690 CPU with 12 nanoseconds per particle update on the same benchmark. A heterogeneous Xeon + Xeon Phi simulation achieves 6.7 nanoseconds per particle update, slightly outperforming 2 Xeon CPUs. Implementation on Kepler-generation GPUs achieves up to 10x speedup over 8 CPU cores in single precision.
Sergey Bastrakov, Iosif Meyerov, Igor Surmin, Evgeny Efimenko, Arkady Gonoskov, Alexander Malyshev & Mikhail Shyriaev, N.I. Lobachevsky State University of Nizhni Novgorod
Sergey Bastrakov: email@example.com