{"@context":{"@vocab":"https://cir.nii.ac.jp/schema/1.0/","rdfs":"http://www.w3.org/2000/01/rdf-schema#","dc":"http://purl.org/dc/elements/1.1/","dcterms":"http://purl.org/dc/terms/","foaf":"http://xmlns.com/foaf/0.1/","prism":"http://prismstandard.org/namespaces/basic/2.0/","cinii":"http://ci.nii.ac.jp/ns/1.0/","datacite":"https://schema.datacite.org/meta/kernel-4/","ndl":"http://ndl.go.jp/dcndl/terms/","jpcoar":"https://github.com/JPCOAR/schema/blob/master/2.0/"},"@id":"https://cir.nii.ac.jp/crid/1050845762823319296.json","@type":"Article","productIdentifier":[{"identifier":{"@type":"URI","@value":"https://ipsj.ixsq.nii.ac.jp/records/18391"}},{"identifier":{"@type":"NDL_BIB_ID","@value":"7463779"}},{"identifier":{"@type":"URI","@value":"http://id.ndl.go.jp/bib/7463779"}},{"identifier":{"@type":"URI","@value":"https://ndlsearch.ndl.go.jp/books/R000000004-I7463779"}},{"identifier":{"@type":"NAID","@value":"110002769844"}}],"resourceType":"学術雑誌論文(journal article)","dc:title":[{"@value":"数値計算ポリシー入力型グラムシュミット直交化ライブラリの異種混合計算機環境における性能評価"},{"@language":"en","@value":"Performance Evaluation of the Gram-Schmidt Orthogonalization Library with Numerical Policy Interface on Heterogeneous Platforms"},{"@language":"ja-Kana","@value":"スウチ ケイサン ポリシー ニュウリョクガタ グラムシュミット チョッコウカ ライブラリ ノ イシュ コンゴウ ケイサンキ カンキョウ ニ オケル セイノウ ヒョウカ"}],"dc:language":"ja","description":[{"type":"Other","notation":[{"@value":"報告者らは，数値計算ポリシー入力型の行列計算ライブラリ方式を提案している．本方式は，行列計算処理において，同じ機能を持つが異なる計算精度，計算時間となるアルゴリズムを同時に実行し，数値計算ポリシーとして与えられたユーザの要求精度を満たす最短処理時間での計算結果をユーザに引き渡す．本報告では，本方式の異種混合計算機環境における有効性を評価した．特に，ユーザがプログラム開発環境としてのPC から大規模計算実行環境としてのスーパコンピュータにプログラム実行環境を移した際に，満足な性能を得られていなかったグラムシュミット直交化処理ついて本方式を適用した結果，本方式は最大約4 倍の性能向上を達成することができた．"}]},{"type":"Other","notation":[{"@value":"We have proposed the method of the matrix library with numerical policy interface. The method executes multiple algorithms that have the same function but have a different accuracy and a different computation time, and returns the result that is executed fastest and satisfies the required accuracy of a numerical policy given by user. In this paper, we evaluate the method on the heterogeneous platforms. We especially treat the Gram-Schmidt orthogonalization that was originally executed in small size computations on PCs and that, however, often attains poor performance in large scale computations on supercomputers. Performance evaluation shows that the method of the matrix computation library with numerical policy achieves about 4 times better performance on a supercomputer than the method only tuned in small size computations on a PC."}]}],"creator":[{"@id":"https://cir.nii.ac.jp/crid/1070845762823319296","@type":"Researcher","foaf:name":[{"@value":"直野, 健"},{"@value":"猪貝光祥"},{"@value":"木立, 啓之"}]},{"@id":"https://cir.nii.ac.jp/crid/1070865265057889920","@type":"Researcher","foaf:name":[{"@value":"Ken, Naono"},{"@value":"Mitsuyoshi, Igai"},{"@value":"Hiroyuki, Kidachi"}]}],"publication":{"publicationIdentifier":[{"@type":"ISSN","@value":"18827829"},{"@type":"NDL_BIB_ID","@value":"000000163559"},{"@type":"ISSN","@value":"03875806"},{"@type":"LISSN","@value":"03875806"},{"@type":"NCID","@value":"AA11833852"}],"prism:publicationName":[{"@value":"情報処理学会論文誌コンピューティングシステム（ACS）"}],"dc:publisher":[{"@value":"情報処理学会"}],"prism:publicationDate":"2005-08-15","prism:volume":"46","prism:number":"SIG12(ACS11)","prism:startingPage":"279","prism:endingPage":"288"},"url":[{"@id":"https://ipsj.ixsq.nii.ac.jp/records/18391"},{"@id":"http://id.ndl.go.jp/bib/7463779"},{"@id":"https://ndlsearch.ndl.go.jp/books/R000000004-I7463779"},{"@id":"http://id.nii.ac.jp/1001/00018391/"}],"foaf:topic":[{"@id":"https://cir.nii.ac.jp/all?q=%E6%95%B0%E5%80%A4%E8%A8%88%E7%AE%97","dc:title":"数値計算"}],"dcterms:subject":[{"subjectScheme":"Other","notation":[{"@value":"数値計算"}]}],"relatedProduct":[{"@id":"https://cir.nii.ac.jp/crid/1390282680743410304","@type":"Article","relationType":["isCitedBy"],"jpcoar:relatedTitle":[{"@language":"ja","@value":"オープンスーパーコンピューター時代の数値計算ライブラリについて(インダストリアルマテリアル)"},{"@language":"en","@value":"Numerical Library for Open-Supercomputer(Industrial Materials)"},{"@value":"インダストリアルマテリアル オープンスーパーコンピューター時代の数値計算ライブラリについて"},{"@language":"ja-Kana","@value":"インダストリアルマテリアル オープンスーパーコンピューター ジダイ ノ スウチ ケイサン ライブラリ ニ ツイテ"}]},{"@id":"https://cir.nii.ac.jp/crid/1520009409524678272","@type":"Article","relationType":["isCitedBy"],"jpcoar:relatedTitle":[{"@value":"行列積を用いた古典Gram-Schmidt直交化の並列化手法の検討"},{"@language":"ja-Kana","@value":"ギョウレツ セキ オ モチイタ コテン Gram Schmidt チョッコウカ ノ ヘイレツカ シュホウ ノ ケントウ"}]},{"@id":"https://cir.nii.ac.jp/crid/1520009409525657088","@type":"Article","relationType":["isCitedBy"],"jpcoar:relatedTitle":[{"@value":"数値計算アルゴリズム性能情報DBの自動チューニング技術への適用の検討"},{"@language":"ja-Kana","@value":"スウチ ケイサン アルゴリズム セイノウ ジョウホウ DB ノ ジドウ チューニング ギジュツ エノ テキヨウ ノ ケントウ"}]}],"dataSourceIdentifier":[{"@type":"IRDB","@value":"oai:irdb.nii.ac.jp:02902:0003113194"},{"@type":"NDL_SEARCH","@value":"oai:ndlsearch.ndl.go.jp:R000000004-I7463779"},{"@type":"CIA","@value":"110002769844"}]}