Loading…
Optimizing Single-Source Graph Execution on NUMA Machines
Graphs are data structures capable of representing problems from different domains, such as logistics and social networks. However, these massive graphs stored in high-performance computing (HPC) servers start processing from distinct source vertices (i.e., single-source: such as a different user or...
Saved in:
Main Authors: | , , , |
---|---|
Format: | Conference Proceeding |
Language: | English |
Subjects: | |
Online Access: | Request full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
cited_by | |
---|---|
cites | |
container_end_page | 6 |
container_issue | |
container_start_page | 1 |
container_title | |
container_volume | |
creator | Rocha, Hiago Mayk G. de A. Querol, Vicenc Beltran Lorenzon, Arthur F. Beck, Antonio Carlos S. |
description | Graphs are data structures capable of representing problems from different domains, such as logistics and social networks. However, these massive graphs stored in high-performance computing (HPC) servers start processing from distinct source vertices (i.e., single-source: such as a different user or message in a social network). Therefore, the amount of vertices and the structure of the sub-graphs to be processed will also change depending on the source, highly influencing the graph algorithm behavior and performance. In this paper, we propose GraphNroll, a framework to extract the full potential of these single-source graph algorithms by exploiting the fact that the majority of the HPC servers are on Non-Uniform Memory Access (NUMA) architectures and, therefore, highly dependent of thread and data mapping. GraphNroll adjusts the thread and data mapping on NUMA machines by using only the embeddings of the source vertices (i.e., vector representations that encode the graph topology) to find the best configuration in terms of thread/data mapping. For that, GraphNroll learns over such embeddings to build a machine learning model capable of predicting the mapping solutions for a new algorithm execution without any profiling. Our results show that GraphNroll is, on average, 8% (up to 18%) faster than the Linux OS Default and Random solutions on three NUMA machines while reducing energy consumption. |
doi_str_mv | 10.1109/SBESC60926.2023.10324068 |
format | conference_proceeding |
fullrecord | <record><control><sourceid>ieee_CHZPO</sourceid><recordid>TN_cdi_ieee_primary_10324068</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>10324068</ieee_id><sourcerecordid>10324068</sourcerecordid><originalsourceid>FETCH-LOGICAL-i204t-361d1bbe73c56f08f3870c6c54e14b19600331ff4ce2259610eb0a2a6a2658883</originalsourceid><addsrcrecordid>eNo1j1FLwzAUhaMgOGb_gQ_5A603uUmaPM5S52BzD3XPI423LrJ1pe1A_fUWVDicA9-BA4cxLiATAtxD9VhWhQEnTSZBYiYApQJjr1jicmdRAzolwF6zmZyaNLdO3bJkGD4AYOJaOzVjbtuN8RS_Y_vOq8mOlFbnSx-IL3vfHXj5SeEyxnPLJ73sNgu-8eEQWxru2E3jjwMlfzlnu6fytXhO19vlqlis0yhBjSka8SbqmnIM2jRgG7Q5BBO0IqFq4QwAomgaFUhK7YwAqsFLb7w02lqLc3b_uxuJaN_18eT7r_3_W_wBeMBHrA</addsrcrecordid><sourcetype>Publisher</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype></control><display><type>conference_proceeding</type><title>Optimizing Single-Source Graph Execution on NUMA Machines</title><source>IEEE Xplore All Conference Series</source><creator>Rocha, Hiago Mayk G. de A. ; Querol, Vicenc Beltran ; Lorenzon, Arthur F. ; Beck, Antonio Carlos S.</creator><creatorcontrib>Rocha, Hiago Mayk G. de A. ; Querol, Vicenc Beltran ; Lorenzon, Arthur F. ; Beck, Antonio Carlos S.</creatorcontrib><description>Graphs are data structures capable of representing problems from different domains, such as logistics and social networks. However, these massive graphs stored in high-performance computing (HPC) servers start processing from distinct source vertices (i.e., single-source: such as a different user or message in a social network). Therefore, the amount of vertices and the structure of the sub-graphs to be processed will also change depending on the source, highly influencing the graph algorithm behavior and performance. In this paper, we propose GraphNroll, a framework to extract the full potential of these single-source graph algorithms by exploiting the fact that the majority of the HPC servers are on Non-Uniform Memory Access (NUMA) architectures and, therefore, highly dependent of thread and data mapping. GraphNroll adjusts the thread and data mapping on NUMA machines by using only the embeddings of the source vertices (i.e., vector representations that encode the graph topology) to find the best configuration in terms of thread/data mapping. For that, GraphNroll learns over such embeddings to build a machine learning model capable of predicting the mapping solutions for a new algorithm execution without any profiling. Our results show that GraphNroll is, on average, 8% (up to 18%) faster than the Linux OS Default and Random solutions on three NUMA machines while reducing energy consumption.</description><identifier>EISSN: 2324-7894</identifier><identifier>EISBN: 9798350394108</identifier><identifier>DOI: 10.1109/SBESC60926.2023.10324068</identifier><language>eng</language><publisher>IEEE</publisher><subject>Graph's High-Level Features ; NUMA Machines ; Single-Source Algorithms ; Thread and Data Mapping</subject><ispartof>2023 XIII Brazilian Symposium on Computing Systems Engineering (SBESC), 2023, p.1-6</ispartof><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/10324068$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,780,784,789,790,27925,54555,54932</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/10324068$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Rocha, Hiago Mayk G. de A.</creatorcontrib><creatorcontrib>Querol, Vicenc Beltran</creatorcontrib><creatorcontrib>Lorenzon, Arthur F.</creatorcontrib><creatorcontrib>Beck, Antonio Carlos S.</creatorcontrib><title>Optimizing Single-Source Graph Execution on NUMA Machines</title><title>2023 XIII Brazilian Symposium on Computing Systems Engineering (SBESC)</title><addtitle>SBESC</addtitle><description>Graphs are data structures capable of representing problems from different domains, such as logistics and social networks. However, these massive graphs stored in high-performance computing (HPC) servers start processing from distinct source vertices (i.e., single-source: such as a different user or message in a social network). Therefore, the amount of vertices and the structure of the sub-graphs to be processed will also change depending on the source, highly influencing the graph algorithm behavior and performance. In this paper, we propose GraphNroll, a framework to extract the full potential of these single-source graph algorithms by exploiting the fact that the majority of the HPC servers are on Non-Uniform Memory Access (NUMA) architectures and, therefore, highly dependent of thread and data mapping. GraphNroll adjusts the thread and data mapping on NUMA machines by using only the embeddings of the source vertices (i.e., vector representations that encode the graph topology) to find the best configuration in terms of thread/data mapping. For that, GraphNroll learns over such embeddings to build a machine learning model capable of predicting the mapping solutions for a new algorithm execution without any profiling. Our results show that GraphNroll is, on average, 8% (up to 18%) faster than the Linux OS Default and Random solutions on three NUMA machines while reducing energy consumption.</description><subject>Graph's High-Level Features</subject><subject>NUMA Machines</subject><subject>Single-Source Algorithms</subject><subject>Thread and Data Mapping</subject><issn>2324-7894</issn><isbn>9798350394108</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>2023</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><recordid>eNo1j1FLwzAUhaMgOGb_gQ_5A603uUmaPM5S52BzD3XPI423LrJ1pe1A_fUWVDicA9-BA4cxLiATAtxD9VhWhQEnTSZBYiYApQJjr1jicmdRAzolwF6zmZyaNLdO3bJkGD4AYOJaOzVjbtuN8RS_Y_vOq8mOlFbnSx-IL3vfHXj5SeEyxnPLJ73sNgu-8eEQWxru2E3jjwMlfzlnu6fytXhO19vlqlis0yhBjSka8SbqmnIM2jRgG7Q5BBO0IqFq4QwAomgaFUhK7YwAqsFLb7w02lqLc3b_uxuJaN_18eT7r_3_W_wBeMBHrA</recordid><startdate>20231121</startdate><enddate>20231121</enddate><creator>Rocha, Hiago Mayk G. de A.</creator><creator>Querol, Vicenc Beltran</creator><creator>Lorenzon, Arthur F.</creator><creator>Beck, Antonio Carlos S.</creator><general>IEEE</general><scope>6IE</scope><scope>6IL</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIL</scope></search><sort><creationdate>20231121</creationdate><title>Optimizing Single-Source Graph Execution on NUMA Machines</title><author>Rocha, Hiago Mayk G. de A. ; Querol, Vicenc Beltran ; Lorenzon, Arthur F. ; Beck, Antonio Carlos S.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-i204t-361d1bbe73c56f08f3870c6c54e14b19600331ff4ce2259610eb0a2a6a2658883</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Graph's High-Level Features</topic><topic>NUMA Machines</topic><topic>Single-Source Algorithms</topic><topic>Thread and Data Mapping</topic><toplevel>online_resources</toplevel><creatorcontrib>Rocha, Hiago Mayk G. de A.</creatorcontrib><creatorcontrib>Querol, Vicenc Beltran</creatorcontrib><creatorcontrib>Lorenzon, Arthur F.</creatorcontrib><creatorcontrib>Beck, Antonio Carlos S.</creatorcontrib><collection>IEEE Electronic Library (IEL) Conference Proceedings</collection><collection>IEEE Proceedings Order Plan All Online (POP All Online) 1998-present by volume</collection><collection>IEEE Xplore All Conference Proceedings</collection><collection>IEEE Xplore</collection><collection>IEEE Proceedings Order Plans (POP All) 1998-Present</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Rocha, Hiago Mayk G. de A.</au><au>Querol, Vicenc Beltran</au><au>Lorenzon, Arthur F.</au><au>Beck, Antonio Carlos S.</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Optimizing Single-Source Graph Execution on NUMA Machines</atitle><btitle>2023 XIII Brazilian Symposium on Computing Systems Engineering (SBESC)</btitle><stitle>SBESC</stitle><date>2023-11-21</date><risdate>2023</risdate><spage>1</spage><epage>6</epage><pages>1-6</pages><eissn>2324-7894</eissn><eisbn>9798350394108</eisbn><abstract>Graphs are data structures capable of representing problems from different domains, such as logistics and social networks. However, these massive graphs stored in high-performance computing (HPC) servers start processing from distinct source vertices (i.e., single-source: such as a different user or message in a social network). Therefore, the amount of vertices and the structure of the sub-graphs to be processed will also change depending on the source, highly influencing the graph algorithm behavior and performance. In this paper, we propose GraphNroll, a framework to extract the full potential of these single-source graph algorithms by exploiting the fact that the majority of the HPC servers are on Non-Uniform Memory Access (NUMA) architectures and, therefore, highly dependent of thread and data mapping. GraphNroll adjusts the thread and data mapping on NUMA machines by using only the embeddings of the source vertices (i.e., vector representations that encode the graph topology) to find the best configuration in terms of thread/data mapping. For that, GraphNroll learns over such embeddings to build a machine learning model capable of predicting the mapping solutions for a new algorithm execution without any profiling. Our results show that GraphNroll is, on average, 8% (up to 18%) faster than the Linux OS Default and Random solutions on three NUMA machines while reducing energy consumption.</abstract><pub>IEEE</pub><doi>10.1109/SBESC60926.2023.10324068</doi><tpages>6</tpages></addata></record> |
fulltext | fulltext_linktorsrc |
identifier | EISSN: 2324-7894 |
ispartof | 2023 XIII Brazilian Symposium on Computing Systems Engineering (SBESC), 2023, p.1-6 |
issn | 2324-7894 |
language | eng |
recordid | cdi_ieee_primary_10324068 |
source | IEEE Xplore All Conference Series |
subjects | Graph's High-Level Features NUMA Machines Single-Source Algorithms Thread and Data Mapping |
title | Optimizing Single-Source Graph Execution on NUMA Machines |
url | http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-06T16%3A34%3A19IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-ieee_CHZPO&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=proceeding&rft.atitle=Optimizing%20Single-Source%20Graph%20Execution%20on%20NUMA%20Machines&rft.btitle=2023%20XIII%20Brazilian%20Symposium%20on%20Computing%20Systems%20Engineering%20(SBESC)&rft.au=Rocha,%20Hiago%20Mayk%20G.%20de%20A.&rft.date=2023-11-21&rft.spage=1&rft.epage=6&rft.pages=1-6&rft.eissn=2324-7894&rft_id=info:doi/10.1109/SBESC60926.2023.10324068&rft.eisbn=9798350394108&rft_dat=%3Cieee_CHZPO%3E10324068%3C/ieee_CHZPO%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-i204t-361d1bbe73c56f08f3870c6c54e14b19600331ff4ce2259610eb0a2a6a2658883%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_id=info:pmid/&rft_ieee_id=10324068&rfr_iscdi=true |