A Comparison of Parallel Profiling Tools for Programs utilizing the FFT

Thumbnail
Name:
ixpug_paper_4.pdf
Size:
554.8Kb
Format:
PDF
Description:
Accepted Manuscript
Download
Type
Conference Paper
Authors
Leu, Brian
Aseeri, Samar
Muite, Benson
KAUST Department
Computer, Electrical and Mathematical Science and Engineering (CEMSE) Division
Extreme Computing Research Center
Date
2021-01-06
Online Publication Date
2021-01-06
Print Publication Date
2021-01-20
Permanent link to this record
http://hdl.handle.net/10754/667636

Metadata
Show full item record
Abstract
Performance monitoring is an important component of code optimization. Performance monitoring is also important for the beginning user, but can be difficult to configure appropriately. The overhead of the performance monitoring tools Craypat, FPMP, mpiP, Scalasca and TAU, are measured using default configurations likely to be choosen by a novice user and shown to be small when profiling Fast Fourier Transform based solvers for the Klein Gordon equation based on 2decomp&FFT and on FFTE. Performance measurements help explain that despite FFTE having a more efficient parallel algorithm, it is not always faster than 2decom&FFT because the complied single core FFT is not as fast as that in FFTW which is used in 2decomp&FFT.
Citation
Leu, B., Aseeri, S., & Muite, B. (2021). A Comparison of Parallel Profiling Tools for Programs utilizing the FFT. The International Conference on High Performance Computing in Asia-Pacific Region Companion. doi:10.1145/3440722.3440881
Sponsors
We thank José Gracia, Chirstoph Niethammer, Sameer Shende, Daisuke Takahashi and Brian Wylie for helpful discussions. B.L.’s work was primarily done while affiliated with the University of Michigan. B.K.M. was partially supported by HPC Europa 3 (INFRAIA-2016-1-730897) and the Estonian Center for Excellence in IT (TK148),and his work was mostly done while affiliated with the Institute of Computer Science at the University of Tartu. The computational resources used to build and test the programs were: •Shaheen and Shaheen II operated by the KAUST Supercomputing Laboratory •Hazelhen at HLRS. •K computer that was operated by RIKEN. •Rocket and Vedur operated by the University of Tartu HPCcenter. •Mira that was operated by the Argonne Leadership Computing Facility, which is a DOE Office of Science User Facility supported under Contract DE-AC02-06CH11357.
Publisher
Association for Computing Machinery (ACM)
Conference/Event name
2021 International Conference on High Performance Computing in Asia-Pacific Region Workshops, HPC Asia 2021
ISBN
9781450383035
DOI
10.1145/3440722.3440881
Additional Links
https://dl.acm.org/doi/10.1145/3440722.3440881
Collections
Conference Papers; Extreme Computing Research Center; Computer, Electrical and Mathematical Science and Engineering (CEMSE) Division