{"id":2007750,"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:02007750","sets":["1164:4179:1771495534039:1771495585049"]},"path":["1771495585049"],"owner":"80578","recid":"2007750","title":["大規模言語モデルによるCPUアーキテクチャごとの特性を考慮したコンパイラ最適化"],"pubdate":{"attribute_name":"PubDate","attribute_value":"2026-02-28"},"_buckets":{"deposit":"c53f0eb0-62ec-4318-8d40-ef1405f1cf4c"},"_deposit":{"id":"2007750","pid":{"type":"depid","value":"2007750","revision_id":0},"owners":[80578],"status":"published","created_by":80578},"item_title":"大規模言語モデルによるCPUアーキテクチャごとの特性を考慮したコンパイラ最適化","author_link":[],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"大規模言語モデルによるCPUアーキテクチャごとの特性を考慮したコンパイラ最適化","subitem_title_language":"ja"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"LLM 応用","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2026-02-28","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"奈良先端科学技術大学院大学"},{"subitem_text_value":"奈良先端科学技術大学院大学"},{"subitem_text_value":"奈良先端科学技術大学院大学"},{"subitem_text_value":"奈良先端科学技術大学院大学"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Nara Institute of Science and Technology","subitem_text_language":"en"},{"subitem_text_value":"Nara Institute of Science and Technology","subitem_text_language":"en"},{"subitem_text_value":"Nara Institute of Science and Technology","subitem_text_language":"en"},{"subitem_text_value":"Nara Institute of Science and Technology","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/2007750/files/IPSJ-NL26267024.pdf","label":"IPSJ-NL26267024.pdf"},"date":[{"dateType":"Available","dateValue":"2028-02-28"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-NL26267024.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":"23"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"2a2b5fb1-9db8-4ce3-8dc4-363c914970ee","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2026 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"門谷,拓能"}]},{"creatorNames":[{"creatorName":"坂井,優介"}]},{"creatorNames":[{"creatorName":"上垣外,英剛"}]},{"creatorNames":[{"creatorName":"渡辺,太郎"}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN10115061","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":"2188-8779","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"大規模言語モデル（LLM）を用いることで、コンパイラによって最適化されたコードをさらに効率化できることが知られている。一方で、その最適化効果がどのアーキテクチャにおいてどの程度発現するのか、また性能劣化や数値誤差を引き起こす条件などについては、十分に明らかではない。本研究ではPolyBench/Cを対象とし、LLMにより最適化されたコードを、AVX2やAVX-512などSIMD機能の異なる複数のx86 CPUノード上で実行することで、実行時間、機械語命令列長、ならびに数値的一致性の評価を行う。本稿では、特にLLM生成コードが-O3最適化を上回る条件や、アーキテクチャの違いに依存した性能差、ならびに性能が悪化する典型的なパターンを示す。さらに、LLMが提示するループ展開、タイル化、手動ベクトル化といった最適化パターンが各アーキテクチャにおいてどのように振る舞うかを分析し、今後構築を目指す体系的なベンチマーク基盤の方向性についても議論する。","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"8","bibliographic_titles":[{"bibliographic_title":"研究報告自然言語処理（NL）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2026-02-28","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"24","bibliographicVolumeNumber":"2026-NL-267"}]},"relation_version_is_last":true,"weko_creator_id":"80578"},"updated":"2026-02-19T10:21:12.179639+00:00","created":"2026-02-19T10:21:08.058616+00:00","links":{}}