| Item type |
Trans(1) |
| 公開日 |
2016-02-26 |
| タイトル |
|
|
タイトル |
Abstraction of Space Partitioning for Spatial Computation |
| タイトル |
|
|
言語 |
en |
|
タイトル |
Abstraction of Space Partitioning for Spatial Computation |
| 言語 |
|
|
言語 |
eng |
| キーワード |
|
|
主題Scheme |
Other |
|
主題 |
[発表概要] |
| 資源タイプ |
|
|
資源タイプ識別子 |
http://purl.org/coar/resource_type/c_6501 |
|
資源タイプ |
journal article |
| 著者所属 |
|
|
|
Department of Information and Communication Engineering, Graduate School of Information Science and Technology, The University of Tokyo |
| 著者所属 |
|
|
|
Department of Information and Communication Engineering, Graduate School of Information Science and Technology, The University of Tokyo |
| 著者所属(英) |
|
|
|
en |
|
|
Department of Information and Communication Engineering, Graduate School of Information Science and Technology, The University of Tokyo |
| 著者所属(英) |
|
|
|
en |
|
|
Department of Information and Communication Engineering, Graduate School of Information Science and Technology, The University of Tokyo |
| 著者名 |
Shigeyuki, Sato
Kenjiro, Taura
|
| 著者名(英) |
Shigeyuki, Sato
Kenjiro, Taura
|
| 論文抄録 |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Space-partitioning trees, each of which holds particles in some space, are extensively used in various applications. Parallel programming with space-partitioning trees is an important issue because they are often used for spatial computation on large-scale data that motivates parallel computing. High-level computational patterns that abstract spatial computation on space-partitioning trees will make parallel programming easy, simple, and productive. Although they have tree structures, spatial computation on them is inherently interaction/correlation computation on two sets rather than tree computation. Existing patterns based only on tree structures are therefore insufficient to abstract it. In this study, we present a pattern based on cross edges between two space-partitioning trees. It abstracts computations between subspaces/particles as cross edges. We have implemented our pattern as a C++ template library at the proof-of-concept level. We present our library API, describe its affinity with parallel implementation, and also report the experimental overhead of sequential execution. |
| 論文抄録(英) |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Space-partitioning trees, each of which holds particles in some space, are extensively used in various applications. Parallel programming with space-partitioning trees is an important issue because they are often used for spatial computation on large-scale data that motivates parallel computing. High-level computational patterns that abstract spatial computation on space-partitioning trees will make parallel programming easy, simple, and productive. Although they have tree structures, spatial computation on them is inherently interaction/correlation computation on two sets rather than tree computation. Existing patterns based only on tree structures are therefore insufficient to abstract it. In this study, we present a pattern based on cross edges between two space-partitioning trees. It abstracts computations between subspaces/particles as cross edges. We have implemented our pattern as a C++ template library at the proof-of-concept level. We present our library API, describe its affinity with parallel implementation, and also report the experimental overhead of sequential execution. |
| 書誌レコードID |
|
|
収録物識別子タイプ |
NCID |
|
収録物識別子 |
AA11464814 |
| 書誌情報 |
情報処理学会論文誌プログラミング（PRO）
巻 9,
号 1,
p. 16-16,
発行日 2016-02-26
|
| ISSN |
|
|
収録物識別子タイプ |
ISSN |
|
収録物識別子 |
1882-7802 |
| 出版者 |
|
|
言語 |
ja |
|
出版者 |
情報処理学会 |
IPSJ-TPRO0901011.pdf (27.4 kB)
