Development of a Parallel Language for Graph Search Applications in Massively Parallel Environments
| Project/Area Number |
17K00099
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Software
|
|---|
| Research Institution | Kyoto Tachibana University (2021)
Kyoto University (2017-2020) |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2017-04-01 – 2022-03-31
|
| Project Status |
Completed (Fiscal Year 2021)
|
| Budget Amount *help |
¥4,550,000 (Direct Cost: ¥3,500,000、Indirect Cost: ¥1,050,000)
Fiscal Year 2020: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2019: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2018: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2017: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
|
|---|
| Keywords | プログラミング言語 / 並列・分散処理 / 高性能計算 / 数値計算 / データマイニング / 高性能計算アプリケーション / 知識発見とデータマイニング / 探索・論理・推論アルゴリズム |
|---|
| Outline of Final Research Achievements |
We improved our proposed task-palallel language, called Tascell, and implementations of sevelal application programs using Tascell, as a research project to develop a parallel language that can deliver good performance in distributed memory environemts for practical graph applications. We improved strategies of task assignment to Tascell workers, namely strategies of how a thief worker chooses a victim worker at a work-steal attempt. We also improved our MPI-based implementation of inter-node communication in Tascell. Using the imporoved implementation of Tascell, we developed and tuned several task-parallel applications such as graph mining and construction of Hierarcical matrices, and we achieved good performance on distributed memroy systems including a supercomputer.
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究で対象としている，ワーカ間の知識共有を伴う実用的なグラフ探索・生成アルゴリズムの分散環境での高性能並列化の例は世界的にも少なく，そのような目標を見据えた並列言語のワークスティール戦略や知識共有支援機能の研究は，スーパーコンピュータ等の高性能計算環境の応用範囲を広げる意味でも有意義なものである．たとえば，階層型行列の生成処理は既存の実装では効率良い並列化が困難であったものを，本研究で並列化に成功したものである．本研究で開発したアプリケーション以外にも，SATソルバ，ソーシャルネットワークの分析，ゲーム木探索等の応用が考えられる．
|
Report
(6 results)
Research Products
(42 results)
