Heterogeneous Programming for the Homogeneous Majority

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

162 Downloads (Pure)

Abstract

In order to take advantage of the burgeoning diversity in processors at the frontier of supercomputing, the HPC community is migrating and improving codes to utilise heterogeneous nodes, where accelerators, principally GPUs, are highly prevalent in top-tier supercomputer designs. Programs therefore need to embrace at least some of the complexities of heterogeneous architectures. Parallel programming models have evolved to express heterogeneous paradigms whilst providing mechanisms for writing portable, performant programs. History shows that technologies first introduced at the frontier percolate down to local workhorse systems. However, we expect there will always be a mix of systems, some heterogeneous, but some remaining as homogeneous CPU systems. Thus it is important to ensure codes adapted for heterogeneous systems continue to run efficiently on CPUs. In this study, we explore how well widely used heterogeneous programming models perform on CPU-only platforms, and survey the performance portability they offer on the latest CPU architectures.
Original languageEnglish
Title of host publication2022 IEEE/ACM International Workshop on Performance, Portability and Productivity in HPC (P3HPC)
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Pages1-13
Number of pages13
ISBN (Electronic)978-1-6654-6021-7
ISBN (Print)978-1-6654-6022-4
DOIs
Publication statusE-pub ahead of print - 30 Jan 2023
Event2022 International Workshop on Performance Portability and Productivity (P3HPC) -
Duration: 13 Nov 202213 Nov 2022

Publication series

Name
PublisherIEEE
ISSN (Print)2831-3895
ISSN (Electronic)2831-3909

Conference

Conference2022 International Workshop on Performance Portability and Productivity (P3HPC)
Period13/11/2213/11/22

Bibliographical note

Funding Information:
For the purpose of open access, the author(s) has applied a Creative Commons Attribution (CC BY) license to any Accepted Manuscript version arising. This work was in part funded by EPSRC through the Strategic Partnership in Computational Science for Advanced Simulation and Modelling of Engineering Systems (ASiMoV) project, EP/S005072/1.

Funding Information:
This work used the Isambard 2 UK National Tier-2 HPC Service (http://gw4.ac.uk/isambard/) operated by GW4 and the UK Met Office, and funded by EPSRC (EP/T022078/1). The University of Bristol is an Intel oneAPI Center of Excellence, which helped support this work. Thanks to AWS for supporting access to Graviton 3.

Publisher Copyright:
© 2022 IEEE.

Fingerprint

Dive into the research topics of 'Heterogeneous Programming for the Homogeneous Majority'. Together they form a unique fingerprint.

Cite this