{"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00227125","sets":["1164:2240:11176:11310"]},"path":["11310"],"owner":"44499","recid":"227125","title":["CUDA Fortran＋MIG＋UVMを用いたBLR行列QR分解の大規模高速化"],"pubdate":{"attribute_name":"公開日","attribute_value":"2023-07-27"},"_buckets":{"deposit":"d0d00bb2-3876-4d64-90ff-2de3bfe7645b"},"_deposit":{"id":"227125","pid":{"type":"depid","value":"227125","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"CUDA Fortran＋MIG＋UVMを用いたBLR行列QR分解の大規模高速化","author_link":["604602","604601","604599","604600","604603"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"CUDA Fortran＋MIG＋UVMを用いたBLR行列QR分解の大規模高速化"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"数値計算","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":"九州大学情報基盤研究開発センター"},{"subitem_text_value":"海洋研究開発機構"},{"subitem_text_value":"名古屋大学情報基盤センター"},{"subitem_text_value":"東京工業大学学術国際情報センター"},{"subitem_text_value":"Sandia National Laboratories"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Sandia National Laboratories","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/227125/files/IPSJ-HPC23190014.pdf","label":"IPSJ-HPC23190014.pdf"},"date":[{"dateType":"Available","dateValue":"2025-07-27"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-HPC23190014.pdf","filesize":[{"value":"1.2 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":"becd487d-1ce4-43df-a326-6d65de941eb3","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":"大島, 聡史"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"伊田, 明弘"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"河合, 直聡"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"横田, 理央"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"山崎, 市太郎"}],"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":"CPU や GPU の演算性能の向上に対してメモリ容量やメモリ速度の向上が遅れている今日，行列を近似して計算を行う近似行列計算法への注目が高まっている．我々は近似行列の一種であるブロック低ランク (Block Low Rank，BLR) 行列に対する QR 分解 (BLR-QR) に取り組んでいる．特に GPU を用いた高速な BLR-QR の実現を目指しており，既に過去の研究においてMulti-Instance GPU (MIG) を用いた BLR-QR の高速化に成功している．本稿ではその際に生じた幾つかの課題を解決するため，CUDA Fortran と MIG と Unified Virtual Memory (UVM) を用いた実装を行い，性能を評価した．その結果，Fortran90 プログラムに対して C/C++ および CUDA C を使わずに GPU 化を行うことができ，GPU に適しない小規模 なGPU カーネルであっても MIG により高性能を得ることが可能であり，UVM を使うことで GPU に収まらないメモリサイズでも容易に実行可能であることが確認できた．この実装（実行）方法は BLR-QR 以外のアプリケーションにとっても有用であることが期待される．","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":"14","bibliographicVolumeNumber":"2023-HPC-190"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"id":227125,"updated":"2025-01-19T12:16:08.317283+00:00","links":{},"created":"2025-01-19T01:26:26.348711+00:00"}