Experience and Analysis of Scalable High-Fidelity Computational Fluid Dynamics on Modular Supercomputing Architectures

The never-ending computational demand from simulations of turbulence makes computational fluid dynamics (CFD) a prime application use case for current and future exascale systems. High-order finite element methods, such as the spectral element method, have been gaining traction as they offer high pe...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Karp, Martin, Suarez, Estela, Meinke, Jan H, Andersson, Måns I, Schlatter, Philipp, Markidis, Stefano, Jansson, Niclas
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Karp, Martin
Suarez, Estela
Meinke, Jan H
Andersson, Måns I
Schlatter, Philipp
Markidis, Stefano
Jansson, Niclas
description The never-ending computational demand from simulations of turbulence makes computational fluid dynamics (CFD) a prime application use case for current and future exascale systems. High-order finite element methods, such as the spectral element method, have been gaining traction as they offer high performance on both multicore CPUs and modern GPU-based accelerators. In this work, we assess how high-fidelity CFD using the spectral element method can exploit the modular supercomputing architecture at scale through domain partitioning, where the computational domain is split between a Booster module powered by GPUs and a Cluster module with conventional CPU nodes. We investigate several different flow cases and computer systems based on the modular supercomputing architecture (MSA). We observe that for our simulations, the communication overhead and load balancing issues incurred by incorporating different computing architectures are seldom worthwhile, especially when I/O is also considered, but when the simulation at hand requires more than the combined global memory on the GPUs, utilizing additional CPUs to increase the available memory can be fruitful. We support our results with a simple performance model to assess when running across modules might be beneficial. As MSA is becoming more widespread and efforts to increase system utilization are growing more important our results give insight into when and how a monolithic application can utilize and spread out to more than one module and obtain a faster time to solution.
doi_str_mv 10.48550/arxiv.2405.05640
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2405_05640</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2405_05640</sourcerecordid><originalsourceid>FETCH-LOGICAL-a670-b6daf908c55e5f7856ceb3629d8eb7fe8d72e4aa79740f09c8f405b52a5395823</originalsourceid><addsrcrecordid>eNotkLFugzAURb10qNJ-QKf6B6AuYGxGRJOmUqIOyY4e9nNiyRhkIAp_X0o73eXeK51DyMs7izPJOXuDcLe3OMkYjxnPM_ZIbtt7j8GiV0jBa1p6cPNgB9oZelLgoHFI9_ZyjXZWo7PjTKuu7acRRtstXbpzk9X0Y_bQWrXMPD12enIQ6GlantVatv5Cy6CudkQ1TgGHJ_JgwA34_J8bct5tz9U-Onx_flXlIYJcsKjJNZiCScU5ciMkzxU2aZ4UWmIjDEotEswARCEyZlihpFnQGp4ATwsuk3RDXv9uV_C6D7aFMNe_AupVQPoDuFlYYQ</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Experience and Analysis of Scalable High-Fidelity Computational Fluid Dynamics on Modular Supercomputing Architectures</title><source>arXiv.org</source><creator>Karp, Martin ; Suarez, Estela ; Meinke, Jan H ; Andersson, Måns I ; Schlatter, Philipp ; Markidis, Stefano ; Jansson, Niclas</creator><creatorcontrib>Karp, Martin ; Suarez, Estela ; Meinke, Jan H ; Andersson, Måns I ; Schlatter, Philipp ; Markidis, Stefano ; Jansson, Niclas</creatorcontrib><description>The never-ending computational demand from simulations of turbulence makes computational fluid dynamics (CFD) a prime application use case for current and future exascale systems. High-order finite element methods, such as the spectral element method, have been gaining traction as they offer high performance on both multicore CPUs and modern GPU-based accelerators. In this work, we assess how high-fidelity CFD using the spectral element method can exploit the modular supercomputing architecture at scale through domain partitioning, where the computational domain is split between a Booster module powered by GPUs and a Cluster module with conventional CPU nodes. We investigate several different flow cases and computer systems based on the modular supercomputing architecture (MSA). We observe that for our simulations, the communication overhead and load balancing issues incurred by incorporating different computing architectures are seldom worthwhile, especially when I/O is also considered, but when the simulation at hand requires more than the combined global memory on the GPUs, utilizing additional CPUs to increase the available memory can be fruitful. We support our results with a simple performance model to assess when running across modules might be beneficial. As MSA is becoming more widespread and efforts to increase system utilization are growing more important our results give insight into when and how a monolithic application can utilize and spread out to more than one module and obtain a faster time to solution.</description><identifier>DOI: 10.48550/arxiv.2405.05640</identifier><language>eng</language><subject>Computer Science - Distributed, Parallel, and Cluster Computing ; Computer Science - Mathematical Software ; Physics - Fluid Dynamics</subject><creationdate>2024-05</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2405.05640$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2405.05640$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Karp, Martin</creatorcontrib><creatorcontrib>Suarez, Estela</creatorcontrib><creatorcontrib>Meinke, Jan H</creatorcontrib><creatorcontrib>Andersson, Måns I</creatorcontrib><creatorcontrib>Schlatter, Philipp</creatorcontrib><creatorcontrib>Markidis, Stefano</creatorcontrib><creatorcontrib>Jansson, Niclas</creatorcontrib><title>Experience and Analysis of Scalable High-Fidelity Computational Fluid Dynamics on Modular Supercomputing Architectures</title><description>The never-ending computational demand from simulations of turbulence makes computational fluid dynamics (CFD) a prime application use case for current and future exascale systems. High-order finite element methods, such as the spectral element method, have been gaining traction as they offer high performance on both multicore CPUs and modern GPU-based accelerators. In this work, we assess how high-fidelity CFD using the spectral element method can exploit the modular supercomputing architecture at scale through domain partitioning, where the computational domain is split between a Booster module powered by GPUs and a Cluster module with conventional CPU nodes. We investigate several different flow cases and computer systems based on the modular supercomputing architecture (MSA). We observe that for our simulations, the communication overhead and load balancing issues incurred by incorporating different computing architectures are seldom worthwhile, especially when I/O is also considered, but when the simulation at hand requires more than the combined global memory on the GPUs, utilizing additional CPUs to increase the available memory can be fruitful. We support our results with a simple performance model to assess when running across modules might be beneficial. As MSA is becoming more widespread and efforts to increase system utilization are growing more important our results give insight into when and how a monolithic application can utilize and spread out to more than one module and obtain a faster time to solution.</description><subject>Computer Science - Distributed, Parallel, and Cluster Computing</subject><subject>Computer Science - Mathematical Software</subject><subject>Physics - Fluid Dynamics</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotkLFugzAURb10qNJ-QKf6B6AuYGxGRJOmUqIOyY4e9nNiyRhkIAp_X0o73eXeK51DyMs7izPJOXuDcLe3OMkYjxnPM_ZIbtt7j8GiV0jBa1p6cPNgB9oZelLgoHFI9_ZyjXZWo7PjTKuu7acRRtstXbpzk9X0Y_bQWrXMPD12enIQ6GlantVatv5Cy6CudkQ1TgGHJ_JgwA34_J8bct5tz9U-Onx_flXlIYJcsKjJNZiCScU5ciMkzxU2aZ4UWmIjDEotEswARCEyZlihpFnQGp4ATwsuk3RDXv9uV_C6D7aFMNe_AupVQPoDuFlYYQ</recordid><startdate>20240509</startdate><enddate>20240509</enddate><creator>Karp, Martin</creator><creator>Suarez, Estela</creator><creator>Meinke, Jan H</creator><creator>Andersson, Måns I</creator><creator>Schlatter, Philipp</creator><creator>Markidis, Stefano</creator><creator>Jansson, Niclas</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20240509</creationdate><title>Experience and Analysis of Scalable High-Fidelity Computational Fluid Dynamics on Modular Supercomputing Architectures</title><author>Karp, Martin ; Suarez, Estela ; Meinke, Jan H ; Andersson, Måns I ; Schlatter, Philipp ; Markidis, Stefano ; Jansson, Niclas</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a670-b6daf908c55e5f7856ceb3629d8eb7fe8d72e4aa79740f09c8f405b52a5395823</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Science - Distributed, Parallel, and Cluster Computing</topic><topic>Computer Science - Mathematical Software</topic><topic>Physics - Fluid Dynamics</topic><toplevel>online_resources</toplevel><creatorcontrib>Karp, Martin</creatorcontrib><creatorcontrib>Suarez, Estela</creatorcontrib><creatorcontrib>Meinke, Jan H</creatorcontrib><creatorcontrib>Andersson, Måns I</creatorcontrib><creatorcontrib>Schlatter, Philipp</creatorcontrib><creatorcontrib>Markidis, Stefano</creatorcontrib><creatorcontrib>Jansson, Niclas</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Karp, Martin</au><au>Suarez, Estela</au><au>Meinke, Jan H</au><au>Andersson, Måns I</au><au>Schlatter, Philipp</au><au>Markidis, Stefano</au><au>Jansson, Niclas</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Experience and Analysis of Scalable High-Fidelity Computational Fluid Dynamics on Modular Supercomputing Architectures</atitle><date>2024-05-09</date><risdate>2024</risdate><abstract>The never-ending computational demand from simulations of turbulence makes computational fluid dynamics (CFD) a prime application use case for current and future exascale systems. High-order finite element methods, such as the spectral element method, have been gaining traction as they offer high performance on both multicore CPUs and modern GPU-based accelerators. In this work, we assess how high-fidelity CFD using the spectral element method can exploit the modular supercomputing architecture at scale through domain partitioning, where the computational domain is split between a Booster module powered by GPUs and a Cluster module with conventional CPU nodes. We investigate several different flow cases and computer systems based on the modular supercomputing architecture (MSA). We observe that for our simulations, the communication overhead and load balancing issues incurred by incorporating different computing architectures are seldom worthwhile, especially when I/O is also considered, but when the simulation at hand requires more than the combined global memory on the GPUs, utilizing additional CPUs to increase the available memory can be fruitful. We support our results with a simple performance model to assess when running across modules might be beneficial. As MSA is becoming more widespread and efforts to increase system utilization are growing more important our results give insight into when and how a monolithic application can utilize and spread out to more than one module and obtain a faster time to solution.</abstract><doi>10.48550/arxiv.2405.05640</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2405.05640
ispartof
issn
language eng
recordid cdi_arxiv_primary_2405_05640
source arXiv.org
subjects Computer Science - Distributed, Parallel, and Cluster Computing
Computer Science - Mathematical Software
Physics - Fluid Dynamics
title Experience and Analysis of Scalable High-Fidelity Computational Fluid Dynamics on Modular Supercomputing Architectures
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-28T23%3A30%3A30IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Experience%20and%20Analysis%20of%20Scalable%20High-Fidelity%20Computational%20Fluid%20Dynamics%20on%20Modular%20Supercomputing%20Architectures&rft.au=Karp,%20Martin&rft.date=2024-05-09&rft_id=info:doi/10.48550/arxiv.2405.05640&rft_dat=%3Carxiv_GOX%3E2405_05640%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true