| Project/Area Number |
26540049
|
|---|
| Research Category |
Grant-in-Aid for Challenging Exploratory Research
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
High performance computing
|
|---|
| Research Institution | Tohoku University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
Uno Atsuya 理化学研究所, 計算科学研究機構, チームヘッド (10359218)
|
|---|
| Co-Investigator(Renkei-kenkyūsha) |
Kobayashi Hiroaki 東北大学, サイバーサイエンスセンター, 教授 (40205480)
Egawa Ryusuke 東北大学, サイバーサイエンスセンター, 准教授 (80374990)
Sato Yukinori 東京工業大学, 学術国際情報センター, 特任講師 (30452113)
|
|---|
| Project Period (FY) |
2014-04-01 – 2016-03-31
|
| Project Status |
Completed (Fiscal Year 2015)
|
| Budget Amount *help |
¥3,640,000 (Direct Cost: ¥2,800,000、Indirect Cost: ¥840,000)
Fiscal Year 2015: ¥2,080,000 (Direct Cost: ¥1,600,000、Indirect Cost: ¥480,000)
Fiscal Year 2014: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
|
|---|
| Keywords | 高性能計算 / 耐障害性 / チェックポイントリスタート |
|---|
| Outline of Final Research Achievements |
Assuming that the state of an application is periodically saved during its execution, we have considered an automatic tuning method for the frequency of saving the state to a hierarchical storage system, and also have discussed a way for reducing the time for writing the state to the storage. A promising approach to the reduction is to speculatively write data that will be written in the future at a high probability. Hence, one technical issue is how to predict such data. For the prediction, we need to analyze memory access patterns of the target application. Hence, we have developed a performance analysis tool for the purpose. The validity and effectiveness of these proposed methods are evaluated based on job scheduling simulation of a large-scale computing system.
|
