High performance computing is a specialist area of programming rife with edge cases, caveats and undocumented features. Using decades of collective expertise with a range of high performance computing equipment, working as users, developers and systems administrators, Warwick's research software engineering group can help you to design new software intended for HPC deployment, optimise existing software for newer, larger clusters and unusual architectures and help with deploying your software on a range of HPC systems from "desktop clusters" to cloud deployments.

How does it work?

Warwick's research software engineering group has access to a wide range of software tools, hardware platforms and experience and expertise. We can work with your developers to provide information about problems that have already been encountered as systems scale up, solutions that have been developed to mitigate these problems and help provide access to the huge body of knowledge that is present in academic computing but is never formally published.

We can help to find parallel and serial "hot-spots" where optimization effort should be applied. Finally, we can assist with deploying your software, both by providing expertise to help work around differences and incompatibilities, but also by advising on new solutions such as Shifter or Singularity.

Training

Warwick research software engineering provides general software development training for researchers and research students via this link. In addition we can offer specialised modules on high performance and parallel computing in several languages, ranging from introductory parallelism with MPI and/or OpenMP to scaling codes to 1000s of cpus and many users. The former is available both as part of our regular training courses (not all training courses occur every year), or by special arrangement; more specialised offerings are by arrangement. Please contact us for the full range of training available.

Contact:

Claire Gerard, 07385 145064 /

Research Software Engineers,