{"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00016588","sets":["934:935:949:953"]},"path":["953"],"owner":"1","recid":"16588","title":["統計情報によるオブジェクト配置の最適化"],"pubdate":{"attribute_name":"公開日","attribute_value":"2006-02-15"},"_buckets":{"deposit":"5a88fc8f-aa91-47a1-ac50-8ae7346d81d7"},"_deposit":{"id":"16588","pid":{"type":"depid","value":"16588","revision_id":0},"owners":[1],"status":"published","created_by":1},"item_title":"統計情報によるオブジェクト配置の最適化","author_link":["0","0"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"統計情報によるオブジェクト配置の最適化"},{"subitem_title":"Optimized Object Placement Based on Statistics","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"発表概要","subitem_subject_scheme":"Other"}]},"item_type_id":"3","publish_date":"2006-02-15","item_3_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"日本電信電話株式会社ネットワークサービスシステム研究所"},{"subitem_text_value":"NTT ソフトウェア株式会社"},{"subitem_text_value":"日本電信電話株式会社ネットワークサービスシステム研究所"}]},"item_3_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Network Service System Laboratory NTT Corporation","subitem_text_language":"en"},{"subitem_text_value":"NTT Software","subitem_text_language":"en"},{"subitem_text_value":"Network Service System Laboratory NTT Corporation","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/16588/files/IPSJ-TPRO4702019.pdf"},"date":[{"dateType":"Available","dateValue":"2008-02-15"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-TPRO4702019.pdf","filesize":[{"value":"34.5 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"0","billingrole":"5"},{"tax":["include_tax"],"price":"0","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"15"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"230f9c03-e095-4f60-a4b2-70edca46e8c4","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2006 by the Information Processing Society of Japan"}]},"item_3_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"田中, 慎司"},{"creatorName":"山田, 哲靖"},{"creatorName":"八尾, 宏"}],"nameIdentifiers":[{}]}]},"item_3_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Shinji, Tanaka","creatorNameLang":"en"},{"creatorName":"Tetsuyasu, Yamada","creatorNameLang":"en"},{"creatorName":"Hiroshi, Yao","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_3_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AA11464814","subitem_source_identifier_type":"NCID"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_6501","resourcetype":"journal article"}]},"item_3_source_id_11":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"1882-7802","subitem_source_identifier_type":"ISSN"}]},"item_3_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"JavaVM のヒープ上には多数のオブジェクトが存在するが，それらのすべてが平均的にアクセスされるわけではなく，それぞれのオブジェクトに対するアクセス頻度には偏りがある．我々の調査では，J2EE のようなアプリケーションでのオブジェクトアクセス傾向を解析した結果では，50%のアクセスがヒープ上の8%のオブジェクトに集中していた．この偏りを利用し，高い頻度でアクセスされるオブジェクトをまとめることで，プロセッサのデータキャッシュやTLB バッファの利用効率が向上することが期待できる．本発表では，各オブジェクトのアクセス頻度を計測することでオブジェクト配置を最適化する手法を提案する．まず，ヒープを高頻度でアクセスされるオブジェクト向けのヒープと低頻度でアクセスされるオブジェクト向けのヒープに分割する．アプリケーション起動後に，アプリケーションのオブジェクトのアクセス傾向を，オブジェクトの生成命令である“オブジェクトの生成サイト” ごとに観測する．その結果をもとに，オブジェクト生成サイトを高頻度アクセスオブジェクト生成サイトと低頻度アクセスオブジェクト生成サイトに分け，以後のオブジェクト生成時に，対応するヒープにオブジェクトを生成する．大量に生成したオブジェクトに偏りのあるアクセスを行うマイクロベンチマークによる評価結果では，1.2 倍から最大で3 倍の性能改善効果が得られた．","subitem_description_type":"Other"}]},"item_3_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"There are lots of objects in the heap of JavaVM. Applications usually accesses them partially, not impartially. From our investigation about object access tendency in applications like J2EE, 50% of accesses are aimed at 8% of objects. Using this tendency, it can be expected that efficiency of usage of data caches and TLB buffers in processors become higher by making high frequency accessed objects close. In this presentation, we propose object location optimization based on statistics of object access frequency. At first, the heap is divided into a heap for objects accessed frequently and a heap for objects accessed rarely. Then, JavaVM observes how often the application accesses to each “object creating site,” which is the instruction of creating objects. Each site is categorized to the site creates frequently accessed object or the site creates rarely accessed object. Finally JavaVM creates objects in the appropriate heap based on the category of the site. From results of a evaluation with our micro benchmark, that creates massive objects and that accesses them partially, the throughput are improved by 1.2 times to 3 times.","subitem_description_type":"Other"}]},"item_3_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"109","bibliographic_titles":[{"bibliographic_title":"情報処理学会論文誌プログラミング（PRO）"}],"bibliographicPageStart":"109","bibliographicIssueDates":{"bibliographicIssueDate":"2006-02-15","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"SIG2(PRO28)","bibliographicVolumeNumber":"47"}]},"relation_version_is_last":true,"weko_creator_id":"1"},"id":16588,"updated":"2025-01-22T23:47:40.545178+00:00","links":{},"created":"2025-01-18T22:49:49.116956+00:00"}