Architecture, implementation and parallelization of the software to search for periodic

Document #:

LIGO-P1400097-v1

Document type:

P - Publications

Abstract:

The parallelization, design and scalability of the PolGrawAllSky code to search for periodic gravitational waves from rotating neutron stars is discussed. The code is based on an efficient implementation of the F -statistic using the Fast Fourier Transform algorithm. To perform such an analysis of data from the advanced gravitational wave detectors LIGO and Virgo that will start operating at the end of 2015, hundreds of millions of CPU hours will be required - the implementation on high performance computers is therefore mandatory. We have parallelized parts of code using the Message Passing Interface library, implemented a task-farming mechanism for combining the searches at dierent sky-positions and frequency bands into one extremely scalable program. We have used the MPI I/O library to escape bottlenecks, when writing the generated data into file system. This allowed to develop a highly scalable computation code, which would enable the data analysis at large scales on acceptable time scales.
Benchmarking of the code on a Cray XE6 system was performed to show efficiency of our parallelization concept and to demonstrate scaling up to 50 thousand cores in parallel.

Files in Document:

• C:\Documents and Settings\SONY\Moje dokumenty\Dropbox\...pdf (366.5 kB)

Topics:

• Continuous Wave

Authors:

• Andrzej Krolak

Keywords:

Parallelization MPI MPI I/O HPC gravitational waves F -statistic multi-level parallelism Farm skeletons

Notes and Changes:

Published in Computer Physics Communications Volume 188, March 2015, Pages 167–176.