The HPC (High-Performance Computing) department sees as its mission to develop programming methodologies and software tools that allow researchers to ride on the new tide of accelerator-enhanced HPC.
Together with key collaborators in US, China and Europe, members of the HPC department are currently engaged in three research activities:
- Study of programming methodologies that can effectively use supercomputers that have a cluster architecture while the main computing power being provided by accelerators, such as graphics processing units (GPUs) and many-integrated-core coprocessors. The research subjects include multi-hierarchical mesh and data partitioning, efficient data exchange schemes for intra-node and inter-node communication, effective masking of communication overhead by over-decomposition and pipe-lining, hybrid CPU-accelerator programming.
- Development of user-friendly software tools, which extensively use automated code generation to realize the above programming methodologies, so that non-expert programmers can easily enjoy the extreme computing power of accelerator-based supercomputers.
- Porting real-world applications of computational science to cutting-edge flagship supercomputers, such as Tianhe-2 (the world's No.1 supercomputer as of November 2014).
The current main source of external funding for the HPC department is the Research Council of Norway, from which the HPC department has obtained a 4-year FRINATEK grant entitled "User-friendly programming of GPU-enhanced clusters via automated code translation and optimization". The HPC department is also participating in another 3-year FRINATEK project entitled "PRoductivity and Energy-efficiency through Abstraction-based Parallel Programming (PREAPP)". In addition, the HPC department also coordinates Simula's participation in an EU Artemis project named "EMC2 - Embedded Multi-Core systems for Mixed Criticality applications in dynamic and changeable real-time environments".
Aiming to embrace the future of exascale computing, which is believed to extract its main computing power from cutting-edge accelerators, the HPC department warmly welcomes suggestions of joint national/European/international project proposals.