| Item type |
SIG Technical Reports(1) |
| 公開日 |
2021-03-08 |
| タイトル |
|
|
タイトル |
Energy Aware Scheduler of Single/Multi-node Jobs Exploiting Node Heterogeneity |
| タイトル |
|
|
言語 |
en |
|
タイトル |
Energy Aware Scheduler of Single/Multi-node Jobs Exploiting Node Heterogeneity |
| 言語 |
|
|
言語 |
eng |
| キーワード |
|
|
主題Scheme |
Other |
|
主題 |
ジョブスケジューラー |
| 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 Informatics, Kyoto University |
| 著者所属 |
|
|
|
Academic Center of Computing and Media Studies, Kyoto University |
| 著者所属 |
|
|
|
Academic Center of Computing and Media Studies, Kyoto University |
| 著者所属(英) |
|
|
|
en |
|
|
Graduate School of Informatics, Kyoto University |
| 著者所属(英) |
|
|
|
en |
|
|
Academic Center of Computing and Media Studies, Kyoto University |
| 著者所属(英) |
|
|
|
en |
|
|
Academic Center of Computing and Media Studies, Kyoto University |
| 著者名 |
Jiacheng, Zhou
Keiichiro, Fukazawa
Hiroshi, Nakashima
|
| 著者名(英) |
Jiacheng, Zhou
Keiichiro, Fukazawa
Hiroshi, Nakashima
|
| 論文抄録 |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Modern CPUs suffer from power efficiency heterogeneity, which can result in additional energy cost or performance loss. On the other hand, future supercomputers are expected to be power constraint. This report focuses on energy aware scheduling algorithms target on two situations. In single-node situation, workload consists of various single-node jobs, Combinatorial Optimization Algorithm saves energy by calculating a local optimal allocation plan with KM algorithm. In multi-node situation, power cap causes load unbalancing in multi-node jobs. Sliding Window Algorithm targets on reducing such unbalancing by sliding window. Proposed algorithms are evaluated in the simulation and real supercomputer environment. In single-node situation, Combinatorial Optimization Algorithm achieved up to 2.92% saving. For the multi-node situation, workload is designed based on real historic workload, and up to 5.36% saving was achieved by Sliding Window Algorithm. |
| 論文抄録(英) |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Modern CPUs suffer from power efficiency heterogeneity, which can result in additional energy cost or performance loss. On the other hand, future supercomputers are expected to be power constraint. This report focuses on energy aware scheduling algorithms target on two situations. In single-node situation, workload consists of various single-node jobs, Combinatorial Optimization Algorithm saves energy by calculating a local optimal allocation plan with KM algorithm. In multi-node situation, power cap causes load unbalancing in multi-node jobs. Sliding Window Algorithm targets on reducing such unbalancing by sliding window. Proposed algorithms are evaluated in the simulation and real supercomputer environment. In single-node situation, Combinatorial Optimization Algorithm achieved up to 2.92% saving. For the multi-node situation, workload is designed based on real historic workload, and up to 5.36% saving was achieved by Sliding Window Algorithm. |
| 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 2021-HPC-178,
号 13,
p. 1-12,
発行日 2021-03-08
|
| 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-HPC21178013.pdf (7.8 MB)
