Performance and Power: Systematic Evaluation of AI Workloads on Accelerators with CARAML
The rapid advancement of machine learning (ML) technologies has driven the development of specialized hardware accelerators designed to facilitate more efficient model training. This paper introduces the CARAML benchmark suite, which is employed to assess performance and energy consumption during th...
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| 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 | John, Chelsea Maria Nassyr, Stepan Penke, Carolin Herten, Andreas |
| description | The rapid advancement of machine learning (ML) technologies has driven the
development of specialized hardware accelerators designed to facilitate more
efficient model training. This paper introduces the CARAML benchmark suite,
which is employed to assess performance and energy consumption during the
training of transformer-based large language models and computer vision models
on a range of hardware accelerators, including systems from NVIDIA, AMD, and
Graphcore. CARAML provides a compact, automated, extensible, and reproducible
framework for assessing the performance and energy of ML workloads across
various novel hardware architectures. The design and implementation of CARAML,
along with a custom power measurement tool called jpwr, are discussed in
detail. |
| doi_str_mv | 10.48550/arxiv.2409.12994 |
| format | Article |
| fullrecord | <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2409_12994</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2409_12994</sourcerecordid><originalsourceid>FETCH-arxiv_primary_2409_129943</originalsourceid><addsrcrecordid>eNpjYJA0NNAzsTA1NdBPLKrILNMzMjGw1DM0srQ04WSICEgtSssvyk3MS05VSMxLUQjIL08tslIIriwuSc1NLMlMVnAtS8wpBbLy8xTy0xQcPRXC84uyc_ITU4oVgEKOycmpOalFiSX5RcUK5ZklGQrOjkGOvj48DKxpiTnFqbxQmptB3s01xNlDF-yG-IKizNzEosp4kFviwW4xJqwCANbHPsQ</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Performance and Power: Systematic Evaluation of AI Workloads on Accelerators with CARAML</title><source>arXiv.org</source><creator>John, Chelsea Maria ; Nassyr, Stepan ; Penke, Carolin ; Herten, Andreas</creator><creatorcontrib>John, Chelsea Maria ; Nassyr, Stepan ; Penke, Carolin ; Herten, Andreas</creatorcontrib><description>The rapid advancement of machine learning (ML) technologies has driven the
development of specialized hardware accelerators designed to facilitate more
efficient model training. This paper introduces the CARAML benchmark suite,
which is employed to assess performance and energy consumption during the
training of transformer-based large language models and computer vision models
on a range of hardware accelerators, including systems from NVIDIA, AMD, and
Graphcore. CARAML provides a compact, automated, extensible, and reproducible
framework for assessing the performance and energy of ML workloads across
various novel hardware architectures. The design and implementation of CARAML,
along with a custom power measurement tool called jpwr, are discussed in
detail.</description><identifier>DOI: 10.48550/arxiv.2409.12994</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Distributed, Parallel, and Cluster Computing ; Computer Science - Hardware Architecture ; Computer Science - Learning ; Computer Science - Performance</subject><creationdate>2024-09</creationdate><rights>http://creativecommons.org/licenses/by-sa/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2409.12994$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2409.12994$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>John, Chelsea Maria</creatorcontrib><creatorcontrib>Nassyr, Stepan</creatorcontrib><creatorcontrib>Penke, Carolin</creatorcontrib><creatorcontrib>Herten, Andreas</creatorcontrib><title>Performance and Power: Systematic Evaluation of AI Workloads on Accelerators with CARAML</title><description>The rapid advancement of machine learning (ML) technologies has driven the
development of specialized hardware accelerators designed to facilitate more
efficient model training. This paper introduces the CARAML benchmark suite,
which is employed to assess performance and energy consumption during the
training of transformer-based large language models and computer vision models
on a range of hardware accelerators, including systems from NVIDIA, AMD, and
Graphcore. CARAML provides a compact, automated, extensible, and reproducible
framework for assessing the performance and energy of ML workloads across
various novel hardware architectures. The design and implementation of CARAML,
along with a custom power measurement tool called jpwr, are discussed in
detail.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Distributed, Parallel, and Cluster Computing</subject><subject>Computer Science - Hardware Architecture</subject><subject>Computer Science - Learning</subject><subject>Computer Science - Performance</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNpjYJA0NNAzsTA1NdBPLKrILNMzMjGw1DM0srQ04WSICEgtSssvyk3MS05VSMxLUQjIL08tslIIriwuSc1NLMlMVnAtS8wpBbLy8xTy0xQcPRXC84uyc_ITU4oVgEKOycmpOalFiSX5RcUK5ZklGQrOjkGOvj48DKxpiTnFqbxQmptB3s01xNlDF-yG-IKizNzEosp4kFviwW4xJqwCANbHPsQ</recordid><startdate>20240919</startdate><enddate>20240919</enddate><creator>John, Chelsea Maria</creator><creator>Nassyr, Stepan</creator><creator>Penke, Carolin</creator><creator>Herten, Andreas</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20240919</creationdate><title>Performance and Power: Systematic Evaluation of AI Workloads on Accelerators with CARAML</title><author>John, Chelsea Maria ; Nassyr, Stepan ; Penke, Carolin ; Herten, Andreas</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-arxiv_primary_2409_129943</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Distributed, Parallel, and Cluster Computing</topic><topic>Computer Science - Hardware Architecture</topic><topic>Computer Science - Learning</topic><topic>Computer Science - Performance</topic><toplevel>online_resources</toplevel><creatorcontrib>John, Chelsea Maria</creatorcontrib><creatorcontrib>Nassyr, Stepan</creatorcontrib><creatorcontrib>Penke, Carolin</creatorcontrib><creatorcontrib>Herten, Andreas</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>John, Chelsea Maria</au><au>Nassyr, Stepan</au><au>Penke, Carolin</au><au>Herten, Andreas</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Performance and Power: Systematic Evaluation of AI Workloads on Accelerators with CARAML</atitle><date>2024-09-19</date><risdate>2024</risdate><abstract>The rapid advancement of machine learning (ML) technologies has driven the
development of specialized hardware accelerators designed to facilitate more
efficient model training. This paper introduces the CARAML benchmark suite,
which is employed to assess performance and energy consumption during the
training of transformer-based large language models and computer vision models
on a range of hardware accelerators, including systems from NVIDIA, AMD, and
Graphcore. CARAML provides a compact, automated, extensible, and reproducible
framework for assessing the performance and energy of ML workloads across
various novel hardware architectures. The design and implementation of CARAML,
along with a custom power measurement tool called jpwr, are discussed in
detail.</abstract><doi>10.48550/arxiv.2409.12994</doi><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | DOI: 10.48550/arxiv.2409.12994 |
| ispartof | |
| issn | |
| language | eng |
| recordid | cdi_arxiv_primary_2409_12994 |
| source | arXiv.org |
| subjects | Computer Science - Artificial Intelligence Computer Science - Distributed, Parallel, and Cluster Computing Computer Science - Hardware Architecture Computer Science - Learning Computer Science - Performance |
| title | Performance and Power: Systematic Evaluation of AI Workloads on Accelerators with CARAML |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-07T13%3A24%3A18IST&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=Performance%20and%20Power:%20Systematic%20Evaluation%20of%20AI%20Workloads%20on%20Accelerators%20with%20CARAML&rft.au=John,%20Chelsea%20Maria&rft.date=2024-09-19&rft_id=info:doi/10.48550/arxiv.2409.12994&rft_dat=%3Carxiv_GOX%3E2409_12994%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 |