On the Parallelization of the Particle Method for Hybrid Supercomputers

On the Parallelization of the Particle Method for Hybrid Supercomputers

An implementation of the particle method on a hybrid cluster with graphics processing units (GPU) for modeling bulk generation, propagation, and scattering of electrons in a self-consistent electromagnetic field is presented. For synchronization of recording of results and data exchange in the calcu...

Full description

Saved in:
Bibliographic Details
Published in:Doklady. Mathematics Vol. 106; no. 1; pp. 225 - 229
Main Authors: Chetverushkin, B. N., Markov, M. B., Uskov, R. V.
Format: Journal Article
Language:English
Published: Moscow Pleiades Publishing 01-08-2022
Springer Nature B.V
Subjects:
Data exchange
Electromagnetic fields
Graphics processing units
Mathematics
Mathematics and Statistics
Supercomputers
Synchronism
particle method
GPU
synchronization of calculations
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:An implementation of the particle method on a hybrid cluster with graphics processing units (GPU) for modeling bulk generation, propagation, and scattering of electrons in a self-consistent electromagnetic field is presented. For synchronization of recording of results and data exchange in the calculation of particle parameters, solutions are proposed that reduce GPU RAM usage.
ISSN:1064-5624
1531-8362
DOI:10.1134/S1064562422040068