Publications

BibTex format

@article{Betteridge:2021:10.1109/MCSE.2021.3085102,
author = {Betteridge, JD and Farrell, PE and Ham, DA},
doi = {10.1109/MCSE.2021.3085102},
journal = {Computing in Science and Engineering},
pages = {8--17},
title = {Code generation for productive, portable, and scalable finite element simulation in Firedrake},
url = {http://dx.doi.org/10.1109/MCSE.2021.3085102},
volume = {23},
year = {2021}
}

Download

RIS format (EndNote, RefMan)

TY  - JOUR
AB  - Creating scalable, high performance PDE-based simulations requires a suitablecombination of discretizations, differential operators, preconditioners andsolvers. The required combination changes with the application and with theavailable hardware, yet software development time is a severely limitedresource for most scientists and engineers. Here we demonstrate that generatingsimulation code from a high-level Python interface provides an effectivemechanism for creating high performance simulations from very few lines of usercode. We demonstrate that moving from one supercomputer to another can requiresignificant algorithmic changes to achieve scalable performance, but that thecode generation approach enables these algorithmic changes to be achieved withminimal development effort.
AU  - Betteridge,JD
AU  - Farrell,PE
AU  - Ham,DA
DO  - 10.1109/MCSE.2021.3085102
EP  - 17
PY  - 2021///
SN  - 1521-9615
SP  - 8
TI  - Code generation for productive, portable, and scalable finite element simulation in Firedrake
T2  - Computing in Science and Engineering
UR  - http://dx.doi.org/10.1109/MCSE.2021.3085102
UR  - http://arxiv.org/abs/2104.08012v1
UR  - https://ieeexplore.ieee.org/document/9447889
UR  - http://hdl.handle.net/10044/1/89116
VL  - 23
ER  -

Download

ProfessorDavidHam

Contact

Location

Summary

Citation

BibTex format

RIS format (EndNote, RefMan)