A study of visualization method for communication behavior of network used in large scale parallel computer

2015 Fiscal Year Final Research Report

• Project/Area Number: 25330146
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: High performance computing
• Research Institution: Tokyo University of Technology
• Principal Investigator: ISHIHATA Hiroaki 東京工科大学, コンピュータサイエンス学部, 教授 (90468885)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: 可視化 / 通信ネットワーク / ネットワークトポロジ

Outline of Final Research Achievements

In this study, we developed the visualization technique of the network congestion in large scale parallel computer. The displayed communication features are extracted from raw statistic data of network communication. Raw data from network simulator and real large scale parallel computer and k-means method is used to extract features. The results are visualized on the picture of network topology by coloring depends on their amounts of data transfer on each ports of the switch in the network. Using this, we can recognize the difference of the communication pattern.

Free Research Field

コンピュータアーキテクチャ