Assessing the performance of the FEniCS Project on Graviton3

Reference : Assessing the performance of the FEniCS Project on Graviton3


	Document type :	Scientific Presentations in Universities or Research Centers : Scientific presentation in universities or research centers
	Discipline(s) :	Physical, chemical, mathematical & earth Sciences : Multidisciplinary, general & others Engineering, computing & technology : Computer science
	Focus Areas :	Computational Sciences
	To cite this reference:	http://hdl.handle.net/10993/55560

Title :	Assessing the performance of the FEniCS Project on Graviton3
Language :	English
Author, co-author :	Habera, Michal [University of Luxembourg > > >]
	Hale, Jack [University of Luxembourg > Faculty of Science, Technology and Medicine (FSTM) > Department of Engineering (DoE) >]
Publication date :	6-Jun-2023
Audience :	International
Event name :	UL HPC School
Event date :	06-06-2023 to 08-06-2023
Event organizer :	UL HPC Team
Event place (city) :	Luxembourg
Event country :	Luxembourg
Keywords :	[en] HPC ; Amazon ; Graviton 3 ; ARM ; FEniCS
Abstract :	[en] We assess the performance of the FEniCS Project finite element solver on AWS Graviton3 instances, including compiler auto-vectorisation and MPI runs. Our results show that Graviton3 is competitive or better performing for than state-of-the-art x86-64 HPC systems for various standard finite element methods.
Funders :	Fonds National de la Recherche - FnR
Name of the research project :	COAT
Permalink :	http://hdl.handle.net/10993/55560
Commentary :	This project received material resources from Amazon AWS.

File(s) associated to this reference

Fulltext file(s):

	File	Commentary	Version	Size	Access
Open access	HABERA_HALE_graviton3_ul_hpc_day.pdf		Publisher postprint	6.5 MB	View/Open

All documents in ORBi^lu are protected by a user license.