{"id":174128,"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00174128","sets":["1164:2240:8543:8882"]},"path":["8882"],"owner":"11","recid":"174128","title":["HPCにおけるHSAの性能評価"],"pubdate":{"attribute_name":"公開日","attribute_value":"2016-08-01"},"_buckets":{"deposit":"c91d0f0f-4cdc-4a06-893b-8c2b2711bbd0"},"_deposit":{"id":"174128","pid":{"type":"depid","value":"174128","revision_id":0},"owners":[11],"status":"published","created_by":11},"item_title":"HPCにおけるHSAの性能評価","author_link":["356216","356215","356217","356214"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"HPCにおけるHSAの性能評価"},{"subitem_title":"Effectiveness of Heterogeneous System Architecture in High Performance Computing","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"GPU","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2016-08-01","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"東京大学"},{"subitem_text_value":"東京大学"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"University of Tokyo","subitem_text_language":"en"},{"subitem_text_value":"University of Tokyo","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/174128/files/IPSJ-HPC16155014.pdf","label":"IPSJ-HPC16155014.pdf"},"date":[{"dateType":"Available","dateValue":"2018-08-01"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-HPC16155014.pdf","filesize":[{"value":"627.0 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":"14"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"7193d02b-c0b8-405b-82c7-bfcae4f5b675","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2016 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"石村, 脩"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"吉本, 芳英"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Osamu, Ishimura","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Yoshihide, Yoshimoto","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN10463942","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":"2188-8841","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"今日のHigh Performance Computing (HPC) では，処理の高速化のため，General Purpose computing on GPU (GPGPU) が頻繁に用いられている．しかし，一般的にこれらで用いられている GPU は CPU に汎用バスを介して接続されているため，CPU と GPU の間のデータ転送や処理の切り替えが遅く，粒度の細かい並列処理には向かない．一方で近年開発が進められている Heterogeneous System Architecture (HSA) では，汎用バスを介したデータ転送ではなく CPU と GPU で仮想空間を含めて統合されたメモリによるデータ共有 (Heterogeneous Uniform Memory Access) やカーネルモードへのコンテキストスイッチをせずに GPU にジョブを渡すことを可能とする機構 (Heterogeneous Queuing) が用意され，この問題への対応がなされていると主張されている．しかし，HSA が HPC において，実際にどの程度の効果を持つものであるか検証した先行研究は存在しない．そこで本研究では，HSA を採用した APU (Godavari) の性能評価を，データのアクセス遅延・バンド幅，および GPU のジョブの起動遅延に注目して行った．","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"Recently, General Purpose computing on GPU (GPGPU) has frequently been used as an acceleration technique in High Performance Computing (HPC). Most of the GPUs used in GPGPU are discrete GPUs which are connected to CPUs via general purpose buses. Therefore, they are not effective for fine grained parallelism because of their large latency in a context switch and slow data transfer. Heterogeneous System Architecture (HSA), which has been developed recently, provides unified memory access including the virtual space between CPU and GPU without the transfer (Heterogeneous Uniform Memory Access) and job queuing without kernel context switch (Heterogeneous Queuing) in order to address this problem. However, there is no previous research on the effectiveness of HSA in HPC. In this paper, from the viewpoint of latency and bandwidth in data access and job queuing latency, we evaluated the performance of Godavari that is an APU implements HSA.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"6","bibliographic_titles":[{"bibliographic_title":"研究報告ハイパフォーマンスコンピューティング（HPC）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2016-08-01","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"14","bibliographicVolumeNumber":"2016-HPC-155"}]},"relation_version_is_last":true,"weko_creator_id":"11"},"updated":"2025-01-20T07:01:01.886223+00:00","created":"2025-01-19T00:44:21.667397+00:00","links":{}}