Software for exascale
We are investigating new and developing programming models for the next generation of HPC architectures.
The next generation of HPC architectures are expected to perform at the exascale level - performing a million million million calculations per second.
Through our flagship CRESTA project (Collaborative Research into Exascale Software Technologies and Applications), we are investigating new and developing programming models such as: the directive-based GPU model OpenACC; Co-array Fortran; NUMA based OpenMP issues and hybrid programming models. Implementation issues with MPI, auto-tuning techniques and fault tolerant algorithms are also strategic areas for us in the exascale context. These form part of wider research programmes in these areas.
CRESTA is an EU FP7 project.