figshare
Browse
PosterSIAMCSE2019.pdf (1.7 MB)

Automated Performance Analysis with PETSc

Download (1.7 MB)
journal contribution
posted on 2019-02-23, 02:17 authored by Matthew KnepleyMatthew Knepley, Justin Chang, Albert Cowie
Comparing the performance of different discretizations or complex solvers is a challenging task, since pure flop rates or rooflines shed little light. Our recently developed TAS system for performance comparison incorporates error measures into the performance analysis, so that we can compare different solvers, discretizations, and models for the same problem. PETSc has been augmented to produce the data for this analysis.

Funding

U.S. DOE Contract DE-AC02-06CH11357

NSF SI2-SSI: 1450339

History

Usage metrics

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC