Knowledge engineering for automatic parallel performance diagnosis

Abstract

Abstract Scientific parallel programs often undergo significant performance tuning before meeting their performance expectation. Performance tuning naturally involves a diagnostic process—locating performance bugs that make a program inefficient and explaining them in terms of high‐level program design. We present a systematic approach to generating performance knowledge for automatically diagnosing parallel programs. Our approach exploits program semantics and parallelism found in parallel programming patterns to search for and define bugs. The approach addresses how to extract the expert knowledge required for performance diagnosis from parallel patterns and represents this knowledge in a manner such that the diagnostic process can be automated. We demonstrate the effectiveness of our knowledge‐engineering approach through a case study. Our experience diagnosing divide‐and‐conquer programs shows that pattern‐based performance knowledge can provide effective guidance for locating and defining performance bugs at a high level of program abstraction. Copyright © 2006 John Wiley & Sons, Ltd.

Cite this paper

@article{li2006knowledge,
  title = {{Knowledge engineering for automatic parallel performance diagnosis}},
  author = {Li, L. and Malony, Allen D.},
  journal = {Concurrency and Computation Practice and Experience},
  volume = {19},
  number = {11},
  pages = {1497--1515},
  year = {2006},
  doi = {10.1002/cpe.1127}
}

Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.

Related papers