OpenSBLI
A framework for the automated derivation of finite difference solvers from high-level problem descriptions
OpenSBLI is a Python-based modelling framework that is capable of expanding a set of differential equations written in Einstein notation and automatically generating C code that performs the finite difference approximation to obtain a solution. This C code is then targetted with the OPS library towards specific hardware backends, such as MPI/OpenMP for execution on CPUs, and CUDA/OpenCL for execution on GPUs. The main focus of OpenSBLI is on the solution of the compressible Navier-Stokes equations with application to shock-boundary layer interactions (SBLI). However, in principle, any set of equations that can be written in Einstein notation may be solved.
OpenSBLI has been through three main phases of development. Version 1 (master branch, Jacobs et al 2017) was developed by Satya Jammy and Christian Jacobs as a proof of concept, with application to a triply-periodic test case. Version 2 (cpc-release branch, see Lusher et al, 2021) was developed by David Lusher and Satya Jammy and included shock-capturing, wall boundary condition and an expanded range of applications, including a transitional shock-wave/boundary-layer interaction problem in 3D. Version 3 (jaxa-release, Lusher et al, submitted for publication, 2024) was primarily developed by David Lusher and includes support for multiblock grids, additional filter treatments and a flagship application to an airfoil in transonic flow.
OpenSBLI has been tested on large CPU and GPU machines and has shown excellent per-node performance and scaling characteristics. Despite the automated code generation process it is generally able to achieve similar performance levels to hand-crafted code. In addition, by exploiting the automation process it was possible (see Jammy et al 2019) to optimise the use of small numbers of intermediate storage arrays leading to a factor of two improvement over traditional hand-written codes. Examples of applications and publications are listed in the linked pages here. The release code includes a documentation folder.