Argonne National Laboratory

Scalable Parallel Scripting for Scientific Computing

TitleScalable Parallel Scripting for Scientific Computing
Publication TypeJournal Article
Year of Publication2010
AuthorsBeckman, PH, Foster, IT, Wilde, M, Raicu, I
JournalSciDAC Review
Date Published03/2010

Researchers at the University of Chicago and Argonne National Laboratory have been extending the time-tested programming technique of scripting into new realms of performance. Through the Swift parallel scripting language, they aim to enhance scientific productivity by enabling scripts to execute many copies of ordinary application programs at very high degrees of parallelism. Parallel Swift scripts can run with little or no change across a range of platforms from multicore desktops to the largest petascale systems available. Scientists are using Swift in a broad range of disciplines to productively leverage highly parallel resources such as the Blue Gene/P in the Argonne Leadership Computing Facility.