Performance prediction of massively parallel computation by Bayesian inference

• Kohashi Hisashi

  Department of Mechanical and Physical Engineering, Tottori University

• Iwamoto Harumichi

  Department of Mechanical and Physical Engineering, Tottori University

• Fukaya Takeshi

  Information Initiative Center, Hokkaido University

• Yamamoto Yusaku

  Department of Communication Engineering and Informatics, The University of Electro-Communications

• Hoshi Takeo

  Department of Mechanical and Physical Engineering, Tottori University

Abstract

A performance prediction method for massively parallel computation is proposed. The method is based on performance modeling and Bayesian inference to predict elapsed time $T$ as a function of the number of used nodes $P$ ($T=T(P)$). The focus is on extrapolation for larger values of $P$ from the perspective of application researchers. The proposed method has several improvements over the method developed in a previous paper, and application to real-symmetric generalized eigenvalue problem shows promising prediction results. The method is generalizable and applicable to many other computations.

Journal

• JSIAM Letters

  JSIAM Letters 14 (0), 13-16, 2022

  The Japan Society for Industrial and Applied Mathematics

Keywords

• parallel algorithm
• Bayesian inference
• performance modeling
• generalized eigenvalue problem
• Monte Carlo method

Details

• CRID: 1390010292486973184
• NII Article ID: 130008164520
• DOI: 10.14495/jsiaml.14.13
• ISSN: 18830617
• ISSN: 18830609
• Web Site: https://www.jstage.jst.go.jp/article/jsiaml/14/0/14_13/_pdf
• Text Lang: en
• Data Source:
  • JaLC
  • Crossref
  • CiNii Articles
  • KAKEN
• Abstract License Flag: Disallowed