2022 Fiscal Year Final Research Report
Principles and Practice of Massively-Parallel Computing Based on Tropical Algebra
Project/Area Number |
20K21794
|
Research Category |
Grant-in-Aid for Challenging Research (Exploratory)
|
Allocation Type | Multi-year Fund |
Review Section |
Medium-sized Section 60:Information science, computer engineering, and related fields
|
Research Institution | Osaka University |
Principal Investigator |
Ino Fumihiko 大阪大学, 大学院情報科学研究科, 教授 (90346172)
|
Project Period (FY) |
2020-07-30 – 2023-03-31
|
Keywords | トロピカル代数 / 最適化問題 / 高速化 / アクセラレータ |
Outline of Final Research Achievements |
We tried to develop a fast method for solving combinatorial optimization problems with an emerging algebraic system, called tropical algebraic system, on computational accelerators, such as the graphics processing unit (GPU). In more detail, we developed not only an optimization technique specific to tropical algebraic system but also a data compression/decompression method for accelerating GPU applications executed on thousands of threads. Their effectiveness was evaluated with practical GPU applications, such as all-pairs shortest path search and quantum circuit simulation.
|
Free Research Field |
高性能計算
|
Academic Significance and Societal Importance of the Research Achievements |
人工知能技術の隆盛に象徴されるように,GPUによる計算の高速化は技術のブレークスルーに不可欠な手段として定着している.トロピカル代数系に特有の最適化技術は,道路網やSNSだけでなく,生命情報科学における生体配列の解析に対して貢献でき,適用範囲は広い.また,ライブラリとして実現したデータ圧縮技術は,煩雑なGPUプログラミングの労力を軽減でき,超並列計算機による研究開発の敷居を低下できるものと期待される.
|