{"links":{},"id":227129,"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00227129","sets":["1164:2240:11176:11310"]},"path":["11310"],"owner":"44499","recid":"227129","title":["深層学習モデルにおける電力最適化に向けた消費電力特性および電力制御手法の評価"],"pubdate":{"attribute_name":"公開日","attribute_value":"2023-07-27"},"_buckets":{"deposit":"16dad717-6b78-449e-80ec-ae1db6c6ff0c"},"_deposit":{"id":"227129","pid":{"type":"depid","value":"227129","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"深層学習モデルにおける電力最適化に向けた消費電力特性および電力制御手法の評価","author_link":["604612","604614","604613","604611"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"深層学習モデルにおける電力最適化に向けた消費電力特性および電力制御手法の評価"},{"subitem_title":"Characterization of Power Consumption for Deep Learning Workloads and Power Cap Evaluation Towards Optimization","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"機械学習","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2023-07-27","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"日本アイ・ビー・エム株式会社"},{"subitem_text_value":"日本アイ・ビー・エム株式会社"},{"subitem_text_value":"日本アイ・ビー・エム株式会社"},{"subitem_text_value":"日本アイ・ビー・エム株式会社"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"IBM Research-Tokyo","subitem_text_language":"en"},{"subitem_text_value":"IBM Research-Tokyo","subitem_text_language":"en"},{"subitem_text_value":"IBM Research-Tokyo","subitem_text_language":"en"},{"subitem_text_value":"IBM Research-Tokyo","subitem_text_language":"en"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_publisher":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"情報処理学会","subitem_publisher_language":"ja"}]},"publish_status":"0","weko_shared_id":-1,"item_file_price":{"attribute_name":"Billing file","attribute_type":"file","attribute_value_mlt":[{"url":{"url":"https://ipsj.ixsq.nii.ac.jp/record/227129/files/IPSJ-HPC23190018.pdf","label":"IPSJ-HPC23190018.pdf"},"date":[{"dateType":"Available","dateValue":"2025-07-27"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-HPC23190018.pdf","filesize":[{"value":"1.8 MB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"660","billingrole":"5"},{"tax":["include_tax"],"price":"330","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"14"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"ede866b2-ad65-40ed-888d-b81ddc2553a0","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2023 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"中澤, 里奈"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Sunyanan, Choochotkaew"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Marcelo, Amaral"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"千葉, 立寛"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN10463942","subitem_source_identifier_type":"NCID"}]},"item_4_textarea_12":{"attribute_name":"Notice","attribute_value_mlt":[{"subitem_textarea_value":"SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc."}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_18gh","resourcetype":"technical report"}]},"item_4_source_id_11":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"2188-8841","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"深層学習モデルの学習や推論は，GPU をはじめとするアクセラレータを大量に消費する．そのため，深層学習モデルサイズの増加が著しい昨今において，実行時に消費する電力をいかに最適化するかは非常に重要な課題となっている．BERT や GPT 等様々なモデルが登場し，モデルの性能特性の対する比較はよく行われている一方，これらのモデルがどのような電力消費特性を有しているかは明らかではない．また，様々な深層学習フレームワークの登場に伴い，同一のモデルであっても異なるフレームワークでの実行が可能であり，クラウドを含めた様々な実行環境における電力消費特性および電力制御手法を調べることは重要である．本稿では，MLPerf をベンチマークとして，クラウド環境下での PyTorch や TensorFlow など異なる深層学習フレームワークを用いた際の消費電力の評価およびモデルごとの電力消費特性の評価を行う．さらに，電力制御手法の性能に対する影響についても考察する．","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"8","bibliographic_titles":[{"bibliographic_title":"研究報告ハイパフォーマンスコンピューティング（HPC）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2023-07-27","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"18","bibliographicVolumeNumber":"2023-HPC-190"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"created":"2025-01-19T01:26:26.581573+00:00","updated":"2025-01-19T12:16:04.052147+00:00"}