Roofline analysis with Cray performance analysis tools (CrayPat) and roofline-based performance projections for a future architecture

Research output: Contribution to journalArticle

Abstract

The roofline analysis model is a visually intuitive performance model used to understand hardware performance limitations as well as potential benefits of optimizations for science and engineering applications. Intel Advisor has provided a useful roofline analysis feature since its version 2017 update 2, but it is not widely compatible with other compilers and chip-architectures. As an alternative, we have employed Cray Performance Analysis Tools (CrayPat) that are more flexible for multiple compilers and architectures. First, we present our procedure for measuring a reliable computational intensity for roofline analysis. We performed several numerical studies for validation via manually derived reference data as well as data from Intel Advisor. Second, we provide roofline analysis results on Blue Waters for several HPC benchmarks and sparse linear algebra libraries. In addition, we present an example of roofline-based performance projection for a future system.

Original languageEnglish (US)
Article numbere4963
JournalConcurrency Computation
Volume31
Issue number16
DOIs
StatePublished - Aug 25 2019

Fingerprint

Performance Analysis
Projection
Compiler
Linear algebra
Performance Model
Model Analysis
Engineering Application
Hardware
Numerical Study
Intuitive
Chip
Update
Benchmark
Water
Optimization
Alternatives
Architecture
Libraries

Keywords

  • CrayPat
  • high-performance computing
  • performance projection
  • profiling
  • roofline performance analysis

ASJC Scopus subject areas

  • Software
  • Theoretical Computer Science
  • Computer Science Applications
  • Computer Networks and Communications
  • Computational Theory and Mathematics

Cite this

@article{913811c522ad48a693dcf36d76828a0a,
title = "Roofline analysis with Cray performance analysis tools (CrayPat) and roofline-based performance projections for a future architecture",
abstract = "The roofline analysis model is a visually intuitive performance model used to understand hardware performance limitations as well as potential benefits of optimizations for science and engineering applications. Intel Advisor has provided a useful roofline analysis feature since its version 2017 update 2, but it is not widely compatible with other compilers and chip-architectures. As an alternative, we have employed Cray Performance Analysis Tools (CrayPat) that are more flexible for multiple compilers and architectures. First, we present our procedure for measuring a reliable computational intensity for roofline analysis. We performed several numerical studies for validation via manually derived reference data as well as data from Intel Advisor. Second, we provide roofline analysis results on Blue Waters for several HPC benchmarks and sparse linear algebra libraries. In addition, we present an example of roofline-based performance projection for a future system.",
keywords = "CrayPat, high-performance computing, performance projection, profiling, roofline performance analysis",
author = "Kwack, {Jae Hyuk} and Galen Arnold and Celso Mendes and Bauer, {Gregory H.}",
year = "2019",
month = "8",
day = "25",
doi = "10.1002/cpe.4963",
language = "English (US)",
volume = "31",
journal = "Concurrency Computation",
issn = "1532-0626",
publisher = "John Wiley and Sons Ltd",
number = "16",

}

TY - JOUR

T1 - Roofline analysis with Cray performance analysis tools (CrayPat) and roofline-based performance projections for a future architecture

AU - Kwack, Jae Hyuk

AU - Arnold, Galen

AU - Mendes, Celso

AU - Bauer, Gregory H.

PY - 2019/8/25

Y1 - 2019/8/25

N2 - The roofline analysis model is a visually intuitive performance model used to understand hardware performance limitations as well as potential benefits of optimizations for science and engineering applications. Intel Advisor has provided a useful roofline analysis feature since its version 2017 update 2, but it is not widely compatible with other compilers and chip-architectures. As an alternative, we have employed Cray Performance Analysis Tools (CrayPat) that are more flexible for multiple compilers and architectures. First, we present our procedure for measuring a reliable computational intensity for roofline analysis. We performed several numerical studies for validation via manually derived reference data as well as data from Intel Advisor. Second, we provide roofline analysis results on Blue Waters for several HPC benchmarks and sparse linear algebra libraries. In addition, we present an example of roofline-based performance projection for a future system.

AB - The roofline analysis model is a visually intuitive performance model used to understand hardware performance limitations as well as potential benefits of optimizations for science and engineering applications. Intel Advisor has provided a useful roofline analysis feature since its version 2017 update 2, but it is not widely compatible with other compilers and chip-architectures. As an alternative, we have employed Cray Performance Analysis Tools (CrayPat) that are more flexible for multiple compilers and architectures. First, we present our procedure for measuring a reliable computational intensity for roofline analysis. We performed several numerical studies for validation via manually derived reference data as well as data from Intel Advisor. Second, we provide roofline analysis results on Blue Waters for several HPC benchmarks and sparse linear algebra libraries. In addition, we present an example of roofline-based performance projection for a future system.

KW - CrayPat

KW - high-performance computing

KW - performance projection

KW - profiling

KW - roofline performance analysis

UR - http://www.scopus.com/inward/record.url?scp=85053813348&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85053813348&partnerID=8YFLogxK

U2 - 10.1002/cpe.4963

DO - 10.1002/cpe.4963

M3 - Article

AN - SCOPUS:85053813348

VL - 31

JO - Concurrency Computation

JF - Concurrency Computation

SN - 1532-0626

IS - 16

M1 - e4963

ER -