WEKO3
アイテム
{"_buckets": {"deposit": "424e588e-3d8d-4d93-ac69-9436f2cbfe2e"}, "_deposit": {"created_by": 3, "id": "1920", "owners": [3], "pid": {"revision_id": 0, "type": "depid", "value": "1920"}, "status": "published"}, "_oai": {"id": "oai:uec.repo.nii.ac.jp:00001920", "sets": ["2"]}, "author_link": ["6356", "6357", "6358", "6359", "6360", "6361"], "control_number": "1920", "item_10003_biblio_info_30": {"attribute_name": "書誌情報", "attribute_value_mlt": [{"bibliographicIssueDates": {"bibliographicIssueDate": "2011-09-09", "bibliographicIssueDateType": "Issued"}, "bibliographicPageEnd": "234", "bibliographicPageStart": "228", "bibliographic_titles": [{"bibliographic_title": "Second International Conference on Networking and Computing", "bibliographic_titleLang": "en"}]}]}, "item_10003_description_29": {"attribute_name": "内容記述", "attribute_value_mlt": [{"subitem_description": "Numerical simulation for visual processing of thehuman brain is one of time-consuming applications. This papershows acceleration techniques for a simulation program of thevisual processing. We parallelize convolution calculations, whichare core operations, which the simulation program requests, on aGPU-accelerated PC cluster. Our implementation includes threeimprovement points. Firstly, we consider efficient data mappingonto global and shared memories1 of the GPU. Secondly, multipleconvolutions for the same input data are computed by eachnode’s GPU, referred to as package execution. Finally, an input2-dimensional image is divided into regions and convolutions forthese regions are executed in parallel utilizing MPI (MessagePassing Interface). Our experimental results show a linearspeedup up to 12 nodes in the PC cluster for the convolutionprogram. We also show the effects of the package executionand reduced communication on NVIDIA tesla C1060 and C2070,respectively.", "subitem_description_type": "Other"}]}, "item_10003_publisher_31": {"attribute_name": "出版者", "attribute_value_mlt": [{"subitem_publisher": "IEEE"}]}, "item_creator": {"attribute_name": "著者", "attribute_type": "creator", "attribute_value_mlt": [{"creatorNames": [{"creatorName": "Junichi, Ohmura", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "6356", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Akira, Egashira", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "6357", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Shunji, Satoh", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "6358", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Takefumi, Miyoshi", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "6359", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Hidetsugu, Irie", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "6360", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Tsutomu, Yoshinaga", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "6361", "nameIdentifierScheme": "WEKO"}]}]}, "item_files": {"attribute_name": "ファイル情報", "attribute_type": "file", "attribute_value_mlt": [{"accessrole": "open_date", "date": [{"dateType": "Available", "dateValue": "2016-09-15"}], "displaytype": "detail", "download_preview_message": "", "file_order": 0, "filename": "9000000555.pdf", "filesize": [{"value": "1.5 MB"}], "format": "application/pdf", "future_date_message": "", "is_thumbnail": false, "licensetype": "license_free", "mimetype": "application/pdf", "size": 1500000.0, "url": {"label": "9000000555.pdf", "url": "https://uec.repo.nii.ac.jp/record/1920/files/9000000555.pdf"}, "version_id": "aa1c4227-cc99-4730-9636-54d2a81bfa78"}]}, "item_keyword": {"attribute_name": "キーワード", "attribute_value_mlt": [{"subitem_subject": "Parallel computing", "subitem_subject_language": "en", "subitem_subject_scheme": "Other"}, {"subitem_subject": "CUDA", "subitem_subject_language": "en", "subitem_subject_scheme": "Other"}, {"subitem_subject": "GPU", "subitem_subject_language": "en", "subitem_subject_scheme": "Other"}, {"subitem_subject": "MPI", "subitem_subject_language": "en", "subitem_subject_scheme": "Other"}, {"subitem_subject": "Numerical simulation", "subitem_subject_language": "en", "subitem_subject_scheme": "Other"}, {"subitem_subject": "Convolution", "subitem_subject_language": "en", "subitem_subject_scheme": "Other"}, {"subitem_subject": "Visual neuron systemsimulation", "subitem_subject_language": "en", "subitem_subject_scheme": "Other"}]}, "item_language": {"attribute_name": "言語", "attribute_value_mlt": [{"subitem_language": "eng"}]}, "item_resource_type": {"attribute_name": "資源タイプ", "attribute_value_mlt": [{"resourcetype": "conference paper", "resourceuri": "http://purl.org/coar/resource_type/c_5794"}]}, "item_title": "Multi-GPU Acceleration of Optical Flow Computation in Visual Functional Simulation", "item_titles": {"attribute_name": "タイトル", "attribute_value_mlt": [{"subitem_title": "Multi-GPU Acceleration of Optical Flow Computation in Visual Functional Simulation", "subitem_title_language": "en"}]}, "item_type_id": "10003", "owner": "3", "path": ["2"], "permalink_uri": "https://uec.repo.nii.ac.jp/records/1920", "pubdate": {"attribute_name": "PubDate", "attribute_value": "2016-09-15"}, "publish_date": "2016-09-15", "publish_status": "0", "recid": "1920", "relation": {}, "relation_version_is_last": true, "title": ["Multi-GPU Acceleration of Optical Flow Computation in Visual Functional Simulation"], "weko_shared_id": -1}
Multi-GPU Acceleration of Optical Flow Computation in Visual Functional Simulation
https://uec.repo.nii.ac.jp/records/1920
https://uec.repo.nii.ac.jp/records/192001bdc8da-a3a4-40ab-b050-140ac3eb9e8b
名前 / ファイル | ライセンス | アクション |
---|---|---|
![]() |
|
Item type | 会議発表論文 / Conference Paper(1) | |||||
---|---|---|---|---|---|---|
公開日 | 2016-09-15 | |||||
タイトル | ||||||
言語 | en | |||||
タイトル | Multi-GPU Acceleration of Optical Flow Computation in Visual Functional Simulation | |||||
言語 | ||||||
言語 | eng | |||||
キーワード | ||||||
言語 | en | |||||
主題Scheme | Other | |||||
主題 | Parallel computing | |||||
キーワード | ||||||
言語 | en | |||||
主題Scheme | Other | |||||
主題 | CUDA | |||||
キーワード | ||||||
言語 | en | |||||
主題Scheme | Other | |||||
主題 | GPU | |||||
キーワード | ||||||
言語 | en | |||||
主題Scheme | Other | |||||
主題 | MPI | |||||
キーワード | ||||||
言語 | en | |||||
主題Scheme | Other | |||||
主題 | Numerical simulation | |||||
キーワード | ||||||
言語 | en | |||||
主題Scheme | Other | |||||
主題 | Convolution | |||||
キーワード | ||||||
言語 | en | |||||
主題Scheme | Other | |||||
主題 | Visual neuron systemsimulation | |||||
資源タイプ | ||||||
資源タイプ識別子 | http://purl.org/coar/resource_type/c_5794 | |||||
資源タイプ | conference paper | |||||
著者 |
Junichi, Ohmura
× Junichi, Ohmura× Akira, Egashira× Shunji, Satoh× Takefumi, Miyoshi× Hidetsugu, Irie× Tsutomu, Yoshinaga |
|||||
内容記述 | ||||||
内容記述タイプ | Other | |||||
内容記述 | Numerical simulation for visual processing of thehuman brain is one of time-consuming applications. This papershows acceleration techniques for a simulation program of thevisual processing. We parallelize convolution calculations, whichare core operations, which the simulation program requests, on aGPU-accelerated PC cluster. Our implementation includes threeimprovement points. Firstly, we consider efficient data mappingonto global and shared memories1 of the GPU. Secondly, multipleconvolutions for the same input data are computed by eachnode’s GPU, referred to as package execution. Finally, an input2-dimensional image is divided into regions and convolutions forthese regions are executed in parallel utilizing MPI (MessagePassing Interface). Our experimental results show a linearspeedup up to 12 nodes in the PC cluster for the convolutionprogram. We also show the effects of the package executionand reduced communication on NVIDIA tesla C1060 and C2070,respectively. | |||||
書誌情報 |
en : Second International Conference on Networking and Computing p. 228-234, 発行日 2011-09-09 |
|||||
出版者 | ||||||
出版者 | IEEE |