| Item type |
SIG Technical Reports(1) |
| 公開日 |
2023-11-28 |
| タイトル |
|
|
タイトル |
Performance Evaluation of File-based and Memory-based Workflows on a Heterogeneous Vector System |
| タイトル |
|
|
言語 |
en |
|
タイトル |
Performance Evaluation of File-based and Memory-based Workflows on a Heterogeneous Vector System |
| 言語 |
|
|
言語 |
eng |
| キーワード |
|
|
主題Scheme |
Other |
|
主題 |
分散システム |
| 資源タイプ |
|
|
資源タイプ識別子 |
http://purl.org/coar/resource_type/c_18gh |
|
資源タイプ |
technical report |
| 著者所属 |
|
|
|
Graduate School of Information Sciences, Tohoku University |
| 著者所属 |
|
|
|
Cyberscience Center, Tohoku University／Graduate School of Information Sciences, Tohoku University |
| 著者所属 |
|
|
|
Cyberscience Center, Tohoku University |
| 著者所属 |
|
|
|
Cyberscience Center, Tohoku University／Graduate School of Information Sciences, Tohoku University |
| 著者所属(英) |
|
|
|
en |
|
|
Graduate School of Information Sciences, Tohoku University |
| 著者所属(英) |
|
|
|
en |
|
|
Cyberscience Center, Tohoku University / Graduate School of Information Sciences, Tohoku University |
| 著者所属(英) |
|
|
|
en |
|
|
Cyberscience Center, Tohoku University |
| 著者所属(英) |
|
|
|
en |
|
|
Cyberscience Center, Tohoku University / Graduate School of Information Sciences, Tohoku University |
| 著者名 |
Yutsen, Tseng
Keichi, Takahashi
Yoichi, Shimomura
Hiroyuki, Takizawa
|
| 著者名(英) |
Yutsen, Tseng
Keichi, Takahashi
Yoichi, Shimomura
Hiroyuki, Takizawa
|
| 論文抄録 |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Heterogeneous computing systems composed of multiple types of processors each designed for a particular task are widely deployed. One challenge in heterogeneous computing is how to efficiently exchange data between applications that run on different processors. Conventionally, file-based workflows, in which applications exchange data through a file system, have been used. However, due to the storage I/O bottleneck in recent years, memory-based workflows, in which applications exchange data directly between their memory spaces, have shown excellent performance on homogeneous systems. Nevertheless, memory-based workflows have not been implemented or evaluated on heterogeneous vector systems, and thus it is unknown whether their advantages hold. This work implements an in-situ visualization workflow using both file-based and memory-based approaches and explores their performance characteristics under different task assignments, output size, and output frequency on a heterogeneous vector system. |
| 論文抄録(英) |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Heterogeneous computing systems composed of multiple types of processors each designed for a particular task are widely deployed. One challenge in heterogeneous computing is how to efficiently exchange data between applications that run on different processors. Conventionally, file-based workflows, in which applications exchange data through a file system, have been used. However, due to the storage I/O bottleneck in recent years, memory-based workflows, in which applications exchange data directly between their memory spaces, have shown excellent performance on homogeneous systems. Nevertheless, memory-based workflows have not been implemented or evaluated on heterogeneous vector systems, and thus it is unknown whether their advantages hold. This work implements an in-situ visualization workflow using both file-based and memory-based approaches and explores their performance characteristics under different task assignments, output size, and output frequency on a heterogeneous vector system. |
| 書誌レコードID |
|
|
収録物識別子タイプ |
NCID |
|
収録物識別子 |
AN10463942 |
| 書誌情報 |
研究報告ハイパフォーマンスコンピューティング（HPC）
巻 2023-HPC-192,
号 31,
p. 1-7,
発行日 2023-11-28
|
| ISSN |
|
|
収録物識別子タイプ |
ISSN |
|
収録物識別子 |
2188-8841 |
| Notice |
|
|
|
SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc. |
| 出版者 |
|
|
言語 |
ja |
|
出版者 |
情報処理学会 |
IPSJ-HPC23192031.pdf (1.1 MB)
