|
|||||
|
|
Particle-in-cell plasma simulation on heterogeneous cluster systems |
|
| Authors: | S Bastrakov R Donchenko A Gonoskov E Efimenko A Malyshev I Meyerov I Surmin |
| Institution: | 1. Lobachevsky State University of Nizhni Novgorod, Russia;2. Institute of Applied Physics, Russian Academy of Sciences, Nizhni Novgorod, Russia;1. Center for Mathematical Plasma Astrophysics, KU Leuven, Celestpp’nenlaan 200B, 3001, Leuven, Belgium;2. Dipartimento di Elettronica e Telecomunicazioni, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy;3. Scuola di Ingegneria, Università della Basilicata, via dell’Ateneo Lucano, 10, 85100 Potenza, Italy;4. Dipartimento Energia, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy;1. LIDYL, CEA, CNRS, Université Paris-Saclay, CEA Saclay, 91 191 Gif-sur-Yvette, France;2. Lawrence Berkeley National Laboratory, Berkeley, CA, USA |
| Abstract: | This paper offers an approach to high-performance implementation of the Particle-in-Cell (PIC) algorithm for GPU-enabled heterogeneous cluster systems. We present Picador — a tool for three dimensional plasma simulation based on the fully relativistic PIC algorithm aimed at research of high intensity laser-matter interaction. It scales up to at least 2048 CPU cores with 85% strong scaling efficiency and 512 GPUs with 64% weak scaling efficiency, achieving up to 28% of the peak performance on the CPU and 14% on the GPU, respectively. |
| Keywords: | |
| 本文献已被 ScienceDirect 等数据库收录！ | |