Research on Software Autotuning Mechanism that evolves to unknown computing environments

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K12033
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: High performance computing
• Research Institution: The University of Tokyo
• Principal Investigator: SUDA Reiji 東京大学, 大学院情報理工学系研究科, 教授 (40251392)
• Co-Investigator(Kenkyū-buntansha): 滝沢 寛之 東北大学, サイバーサイエンスセンター, 教授 (70323996)
• Co-Investigator(Renkei-kenkyūsha): YASUGI Masahiro 九州工業大学, 大学院情報工学研究院, 教授 (30273759)
• Research Collaborator: KATAGIRI Takahiro 名古屋大学, 情報基盤センター, 教授 (40345434)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 自動チューニング / 高性能計算 / コード変換 / 最適化

Outline of Final Research Achievements

Autotuning is a technology aiming to attain good execution performance on various computational environments by preparing variabilities within software and letting the software itself control the variabilities. In this research, we aimed to develop methodology to infuse variabilities and control mechanism which are unintended or even unknown to existing codes, to attain autotuning even if novel computational environments and novel variabilities become newly known.
We have shown that, by using Xevolver, which is developed by our team members, we can infuse variabilities and autotuning mechanisms which is unknown to the original code. However, it became clear that we need to fully analyze the original code before applying such infusions.

Free Research Field

高性能・並列数値計算