Future proofing and performance optimisation of explicit finite difference solvers using the OpenSBLI framework. Satya Pramod Jammy. University of Southampton


Seminar in the Many-core series. No Booking required. Open to all. Lunch provided

Satya Pramod Jammy, a post doctoral researcher at the University of Southampton. He will present a talk entitled ‘Future proofing and performance optimisation of explicit finite difference solvers using the OpenSBLI framework’.

Abstract

In this talk, a new framework, OpenSBLI will be introduced, which unlike the majority of existing finite difference models, allows users to specify the equations they want to solve in Einstein notation, and details of the numerical methods, at an abstract level. From this specification, the C code that performs the discretisation is automatically generated. By coupling with the OPS execution framework, the generated code is then tailored towards a desired backend to enable the efficient execution of the model on a wide variety of compute hardware. Then we discuss the ease at which algorithmical changes can be made in OpenSBLI and show the performance of several numerical algorithms, characterised by varying degrees of memory and computational intensity on a variety of architectures such as multicore CPUs, GPUs and some initial results on the Intel Xeon Phi Knights Landing coprocessor.

About our Speaker

Satya Pramod Jammy is currently as a post doctoral researcher at the University of Southampton, working on future proofing of Southampton SBLI code and optimisation on various compute architectures from a numerical modelling point of view. He holda a PhD from University of Surrey, with a thesis titled “Direct Numerical simulation of turbulent vortices”. His research interests include Shock-wave boundary layer interactions and Numerical frameworks for DNS and LES of compressible flows.