{"updated":"2025-01-22T03:10:15.627852+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00060362","sets":["1164:1579:1639:1643"]},"path":["1643"],"owner":"2","recid":"60362","title":["クラスタ型並列計算機におけるImplicit Co - schedulingの性能評価"],"pubdate":{"attribute_name":"公開日","attribute_value":"1998-03-05"},"_buckets":{"deposit":"d9d5b4f3-08e7-454b-8787-fce23c236b71"},"_deposit":{"id":"60362","pid":{"type":"depid","value":"60362","revision_id":0},"owners":[2],"status":"published","created_by":2},"item_title":"クラスタ型並列計算機におけるImplicit Co - schedulingの性能評価","author_link":["0","0"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"クラスタ型並列計算機におけるImplicit Co - schedulingの性能評価"},{"subitem_title":"Evaluation of Implicit Co - scheduling on Clustered Parallel Computer","subitem_title_language":"en"}]},"item_type_id":"4","publish_date":"1998-03-05","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"東京工業大学"},{"subitem_text_value":"東京工業大学"},{"subitem_text_value":"新情報処理開発機構"},{"subitem_text_value":"新情報処理開発機構"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Tokyo Institute of Technology","subitem_text_language":"en"},{"subitem_text_value":"Tokyo Institute of Technology","subitem_text_language":"en"},{"subitem_text_value":"Real World Computing Partnership","subitem_text_language":"en"},{"subitem_text_value":"Real World Computing Partnership","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/60362/files/IPSJ-ARC97128008.pdf"},"date":[{"dateType":"Available","dateValue":"2000-03-05"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-ARC97128008.pdf","filesize":[{"value":"676.9 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"660","billingrole":"5"},{"tax":["include_tax"],"price":"330","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"16"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"dc605dac-6768-452e-900c-e24a536d07f3","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 1998 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"福地, 健太郎"},{"creatorName":"松岡, 聡"},{"creatorName":"堀, 敦史"},{"creatorName":"石川, 裕"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Kentarou, Fukuchi","creatorNameLang":"en"},{"creatorName":"Satoshi, Matsuoka","creatorNameLang":"en"},{"creatorName":"Atsushi, Hori","creatorNameLang":"en"},{"creatorName":"Yutaka, Ishikawa","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN10096105","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_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"Implicit co?schedulingはBerkeley NOWプロジェクトで提案された、大域スケジューラーを持たず、オーバーヘッドが少い・実装が容易である等の利点を持つ並列ジョブスケジューリング技法である。これまでの性能評価では実行時間の対ギャングスケジューリング比にして0.6?1.6程度の性能が出るとされているが、実用的なアプリケーションでの性能評価はなされていない。本研究では、大規模高性能クラスター上で、NAS並列ベンチマークを用いる事で、implicit co?schedulingの実践的な性能を測定した。その結果、FT,CGにおいて実行時間の対ギャングスケジューリング比にして最大2.3倍という結果を得ており、Berkeleyの評価が再現しなかった。これは、ネットワークの混雑等が原因と予測され、現在追試中である。","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"Implicit co-scheduling is a parallel job scheduling methodology proposed by the UC Berkeley NOW project, and embodies favorable characterlistics such as lack of global schedulers, low overhead, and easy implementation. Previous literatures have claimed that overhead versus traditional gang schedulers was about a factor or 0.6 to 1.6; however, evaluations were not performed using real-life workloads. We have implemented an implicit co-scheduler on a large-scale, high-performance cluster, and used NAS parallel benchmarks to measure effective performance. There, we found that for FT and CG, the overhead versus gang scheduling can be as high as factor of 2.3, negating the Berkeley results. We conjecture that this is due to excessive network traffic, but are still in the process of perforoming additional experiments.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"48","bibliographic_titles":[{"bibliographic_title":"情報処理学会研究報告計算機アーキテクチャ(ARC)"}],"bibliographicPageStart":"43","bibliographicIssueDates":{"bibliographicIssueDate":"1998-03-05","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"18(1997-ARC-128)","bibliographicVolumeNumber":"1998"}]},"relation_version_is_last":true,"weko_creator_id":"2"},"created":"2025-01-18T23:22:56.626731+00:00","id":60362,"links":{}}