Computing Infrastructure

At the scale of radio astronomy experiments for the next generation facilities discussed in this book, the computing infrastructure plays a crucial role, in particular with respect to High Performance Computing (HPC) systems. This chapter discusses the most recent advances in this area. Firstly, we address a new type of computing facilities, based on combining both CPUs and accelerators as GPUs and FPGAs into heterogeneous HPC systems. Secondly, we introduce a new architecture for enabling a more dynamical use of such heterogeneous hardware resources. Thirdly, we take a deep dive in software containerisation, a key technology for effectively scaling complex codes and fully exploiting computing resources while greatly improving scientific reproducibility. Finally, we conclude by providing an overview on how some recent yet widely adopted cloud-native technologies, as last-generation container orchestrators, can converge together with HPC-specific ones towards higher-level, unified frameworks.