Efficient parallelization method for large scale beam dynamics simulations in linear accelerators

Large scale beam dynamics simulations are important to support the design and operations of an accelerator. The beam dynamics code TRACK, originally developed at Argonne National Laboratory (ANL) has become a general beam dynamics code for hadron linacs and has found wide applications worldwide. In order to combine the unique features of TRACK with large scale and fast parallel computing we have recently developed parallel version of the code: PTRACK It has been successfully benchmarked on different platforms: BG/L and Jazz at ANL, Iceberg at ARSC and Seaborg at NERSC. Large scale end-to- end simulations of the FNAL proton driver has been performed, where 100 millions particles were tracked. The actual parallel version has the potential of simulating 10
⁹
particles on 10 racks with 20,480 processors of BG/L at ANL, which will be available in 2008. This paper focus on efficient parallelization method been used in PTRACK.