Co-operation Between Approximate Networks and Parallel Algorithms(Fostering Joint International Research)

2017 Fiscal Year Final Research Report

• Project/Area Number: 16KK0009
• Research Category: Fund for the Promotion of Joint International Research (Fostering Joint International Research)
• Allocation Type: Multi-year Fund
• Research Field: Information network
• Research Institution: National Institute of Informatics
• Principal Investigator: Koibuchi Michihiro 国立情報学研究所, アーキテクチャ科学研究系, 准教授 (40413926)
• Research Collaborator: CASANOVA Henri ハワイ大学マノア校, Information and Computer Science Department, 教授 ; MATSUTANI Hiroki 慶應義塾大学, 理工学部, 准教授
• Project Period (FY): 2017
• Keywords: 相互結合網 / Approximate Computing / 計算機システム / フォトニックネットワーク / ハイパフォーマンスコンピューティング

Outline of Final Research Achievements

Existing large-scale parallel computers have supported high accurate computation required by traditional scientific simulation. However, recently, big-data low-accuracy data processing and parallel computation would received attention as emerging applications. They do not require high-precise computation and communication. In this context, we extend approximate computing networks that leave bit errors. On approximate computing networks with special supports, we have developed parallel algorithms in order to obtain the acceptable quality of results.

Free Research Field

情報学