Imperial College London


Faculty of EngineeringDepartment of Aeronautics

Reader in Aeronautics



+44 (0)20 7594 1975p.vincent




211City and Guilds BuildingSouth Kensington Campus






BibTex format

author = {Vincent, PE and Witherden, FD and Vermeire and Park, JS and Iyer},
publisher = {IEEE},
title = {Towards green aviation with Python at petascale},
url = {},

RIS format (EndNote, RefMan)

AB - Accurate simulation of unsteady turbulentflow is critical for improved design of greener aircraftthat are quieter and more fuel-efficient. We demonstrateapplication of PyFR, a Python based computational fluiddynamics solver, to petascale simulation of such flowproblems. Rationale behind algorithmic choices, whichoffer increased levels of accuracy and enable sustainedcomputation at up to 58% of peak DP-FLOP/s on unstruc-tured grids, will be discussed in the context of modernhardware. A range of software innovations will also bedetailed, including use of runtime code generation, whichenables PyFR to efficiently target multiple platforms,including heterogeneous systems, via a single implemen-tation. Finally, results will be presented from a full-scale simulation of flow over a low-pressure turbine bladecascade, along with weak/strong scaling statistics from thePiz Daint and Titan supercomputers, and performancedata demonstrating sustained computation at up to 13.7DP-PFLOP/s.
AU - Vincent,PE
AU - Witherden,FD
AU - Vermeire
AU - Park,JS
AU - Iyer
TI - Towards green aviation with Python at petascale
UR -
ER -