XClose

Advanced Research Computing

Home
Menu

High Performance Computing Collaborations

Work with our experts in parallel programming and large-scale computing to accelerate your research.

Many of ARC’s Research Software Engineers and Data Scientists specialise in tackling cutting-edge challenges that go beyond the capabilities of a single workstation, using UCL’s research computing platforms and the largest national and international supercomputers. We have expertise in parallel programming on both CPUs and GPUs using a range of paradigms – from MPI, OpenMP, and CUDA, to modern abstraction layers such as SYCL and tool chains such as OneAPI. All programming languages common to high-performance computing (HPC) applications are supported, including C++, Fortran, and Julia. We have significant experience in benchmarking, performance profiling and optimisation, and in a wide range of computational modelling and statistical methodology. Our staff can provide assistance in tasks such as efficiently fitting models to data or quantifying uncertainty at scale.

The team also contributes to ARC’s education activities, delivering the Research Computing with C++ MSc module and parts of “Machine Learning and Big Data”, and introductory training sessions for UCL’s supercomputers.

ARC staff play key roles in national supercomputing initiatives. We form part of the distributed RSE team for DiRAC, the national HPC facility for various branches of theoretical physics, helping to evaluate project proposals and to ensure codes are optimised to get the most out of the DiRAC machines. In the ExCALIBUR programme, aiming to deliver the next generation of high-performance simulation software for the UK research community, ARC is leading the “benchmarking for exascale” core activity and continues to maintain the benchmark framework. Our staff are members of both the Steering Committee and H&ES Technical Working Group for ExCALIBUR.

Exploring next-generation technology is not just a national endeavour. ARC’s Project Contender provides various one-off systems with pre-release or esoteric hardware for UCL researchers to experiment on. Working closely with Research Infrastructure Developers we assess likely future developments in the field, and can advise on the best approaches to take for your project. Contact us to find out more.