Programming models and technologies
Modern supercomputers expose vast amounts of potential parallelism to programmers. However it is widely acknowledged that writing parallel codes to exploit these machines is very complex and in the domain of a few experts. This difficulty is compounded by the fact that modern HPC machines often contain not just one processing technology, but instead heterogeneous architectures comprising numerous distinct types of processors. This collection is a repository of EPCC outputs relating to parallel programming models and technologies which aim to address some of these challenges.
Items in this Collection
ePython is a 24Kb implementation of Python that we have developed, specifically targetting highly parallel micro-core architectures. The main version here is optimised for the Epiphany architecture, but can be easily ported ...