Skip to content

pymgrit/performance_model

Repository files navigation

Parallel-in-time performance analysis

Python tool for task graph based performance analysis of parallel-in-time methods. Currently the methods

  • Parareal [1]
  • Parallel full approximation scheme in space and time (PFASST) [2]
  • Multigrid reduction in time (MGRIT) [3]

are supported.

The model was compared to the following parallel-in-time libraries:

References

[1] Jacques-Louis Lions, Yvon Maday, and Gabriel Turinici. “Résolution d’EDP par un schéma en temps “pararéel””. In: Comptes Rendus de l’Académie des Sciences. Série I. Mathématique 332.7 (2001), pp. 661–668.

[2] Matthew Emmett and Michael Minion. “Toward an efficient parallel in time method for partial differ- ential equations”. In: Communications in Applied Mathematics and Computational Science 7.1 (2012), pp. 105 –132.

[3] Robert D. Falgout, Stephanie Friedhoff, Tzanio Kolev, Scott MacLachlan, and Jacob B. Schroder. “Parallel time integration with multigrid”. In: SIAM Journal on Scientific Computing 36.6 (2014), pp. C635–C661.

[4] https://github.com/Parallel-in-Time/PararealF90

[5] https://github.com/libpfasst/LibPFASST

[6] https://github.com/Parallel-in-Time/pySDC

[7] https://www.llnl.gov/casc/xbraid

[8] https://github.com/pymgrit/pymgrit