{"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00012672","sets":["581:703:711"]},"path":["711"],"owner":"1","recid":"12672","title":["ウェーブフロント型並列処理における分散メモリ型並列計算機の通信機構の評価"],"pubdate":{"attribute_name":"公開日","attribute_value":"1999-05-15"},"_buckets":{"deposit":"f7293b1b-5242-4ba4-b0d8-b4d15882542d"},"_deposit":{"id":"12672","pid":{"type":"depid","value":"12672","revision_id":0},"owners":[1],"status":"published","created_by":1},"item_title":"ウェーブフロント型並列処理における分散メモリ型並列計算機の通信機構の評価","author_link":["0","0"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"ウェーブフロント型並列処理における分散メモリ型並列計算機の通信機構の評価"},{"subitem_title":"Evaluation of Communication Mechanisms for Distributed Memory Parallel Computers in Wavefront Computation","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"特集:並列処理","subitem_subject_scheme":"Other"}]},"item_type_id":"2","publish_date":"1999-05-15","item_2_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"電子技術総合研究所情報アーキテクチャ部/電気通信大学大学院情報システム学研究科"},{"subitem_text_value":"電子技術総合研究所情報アーキテクチャ部"},{"subitem_text_value":"電子技術総合研究所情報アーキテクチャ部"},{"subitem_text_value":"電子技術総合研究所情報アーキテクチャ部"},{"subitem_text_value":"電子技術総合研究所情報アーキテクチャ部"},{"subitem_text_value":"電子技術総合研究所情報アーキテクチャ部/筑波大学電子・情報工学系"},{"subitem_text_value":"電気通信大学大学院情報システム学研究科"}]},"item_2_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Computer Science Division, Electrotechnical Laboratory/Graduate School of Information Systems, The University of Electro - Communications","subitem_text_language":"en"},{"subitem_text_value":"Computer Science Division, Electrotechnical Laboratory","subitem_text_language":"en"},{"subitem_text_value":"Computer Science Division, Electrotechnical Laboratory","subitem_text_language":"en"},{"subitem_text_value":"Computer Science Division, Electrotechnical Laboratory","subitem_text_language":"en"},{"subitem_text_value":"Computer Science Division, Electrotechnical Laboratory","subitem_text_language":"en"},{"subitem_text_value":"Computer Science Division, Electrotechnical Laboratory/Institute of Information Sciences and Electronics, University of Tsukuba","subitem_text_language":"en"},{"subitem_text_value":"Graduate School of Information Systems, The University of Electro-Communications","subitem_text_language":"en"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"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/12672/files/IPSJ-JNL4005036.pdf"},"date":[{"dateType":"Available","dateValue":"2001-05-15"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-JNL4005036.pdf","filesize":[{"value":"1.5 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":"8"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"c566cd62-2fce-4f45-a038-3499d024e764","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 1999 by the Information Processing Society of Japan"}]},"item_2_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"坂根, 広史"},{"creatorName":"児玉, 祐悦"},{"creatorName":"建部修見"},{"creatorName":"小池, 汎平"},{"creatorName":"山名, 早人"},{"creatorName":"山口, 喜教"},{"creatorName":"弓場, 敏嗣"}],"nameIdentifiers":[{}]}]},"item_2_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Hirofumi, Sakane","creatorNameLang":"en"},{"creatorName":"Yuetsu, Kodama","creatorNameLang":"en"},{"creatorName":"Osamu, Tatebe","creatorNameLang":"en"},{"creatorName":"Hanpei, Koike","creatorNameLang":"en"},{"creatorName":"Hayato, Yamane","creatorNameLang":"en"},{"creatorName":"Yoshinori, Yamaguchi","creatorNameLang":"en"},{"creatorName":"Toshitsugu, Yuba","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_2_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN00116647","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_2_source_id_11":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"1882-7764","subitem_source_identifier_type":"ISSN"}]},"item_2_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"本論文では 分散メモリ型並列計算機において 同期・通信の支援機構が行列問題の並列処理性能に与える影響について議論し それらが有効となる条件・要因を モデルと実験によって定量的に明らかにする. LU分解法の代入部に現れる三角方程式の求解では 互いに依存性のない計算要素がイテレーション間にまたがっており その並列性はウェーブフロント状に抽出できる. この問題を 並列性を自然に利用する細粒度アルゴリズムと ブロック化による粗粒度アルゴリズムで表し 並列計算機EM-XとAP1000+に実装した. 最初に予備実験によって これらの計算機が持つ同期・通信機構の特徴をパラメータによって表した. 次に アルゴリズムの性質をモデル化し 通信オーバヘッドに起因する性能上限と 並列度の制限による有効PE台数を理論的に示した. 問題サイズが小さい場合 あるいは十分なPE台数が利用できる場合は 高い並列度が得られる細粒度アルゴリズムが有望である. ただし細粒度アルゴリズムで高い性能を得るには 通信起動のオーバヘッドが十分小さいことが必要であり EM-Xがこの要件を満たす. 逆に 問題サイズが十分大きいか 比較的少ないPE台数しか与えられない場合は粗粒度アルゴリズムの方が良い. この場合は通信性能より逐次演算性能が重要となり AP1000+が優位性を示す.","subitem_description_type":"Other"}]},"item_2_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"In this paper, we discuss efficient parallel execution of a dense-matrix problem considering trade-offs between fine-grain and coarse-grain communication in distributed memory machines. The solution of the triangular system of equations involves data dependencies between consecutive iterations in the outer-loop. The dependencies can be naturally solved and processed in parallel by wavefront computation. Two ways of parallelizing are presented; the element-wise fine-grain approach and the coarse-grain approach. We implemented these algorithms on both EM-X and AP 1000+. Fine-grain support mechanisms of the EM-X had a great effect on the performance of the element-wise method for relatively small problem size, while employed RISC processors of the AP1000+ brought high performance of the coarse-grain method for larger size.","subitem_description_type":"Other"}]},"item_2_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"2292","bibliographic_titles":[{"bibliographic_title":"情報処理学会論文誌"}],"bibliographicPageStart":"2281","bibliographicIssueDates":{"bibliographicIssueDate":"1999-05-15","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"5","bibliographicVolumeNumber":"40"}]},"relation_version_is_last":true,"item_2_alternative_title_2":{"attribute_name":"その他タイトル","attribute_value_mlt":[{"subitem_alternative_title":"性能評価"}]},"weko_creator_id":"1"},"id":12672,"updated":"2025-01-23T01:32:52.654306+00:00","links":{},"created":"2025-01-18T22:46:57.941463+00:00"}