Skip to main content Skip to page footer

Hardware Aware Numerics and High Performance Computing

This research area at EMCL aims to develop hardware-aware numerical schemes, platform-specific implementations and simulation software for research on emerging processor technologies such as multicore CPUs, GPUs, manycore processors, FPGAs and new co-processor boards. It integrates advanced numerical concepts with vendor-specific and cross-platform programming environments to provide application developers with the best performance, optimal resource utilisation and productivity in a high performance computing environment.

Hardware-aware numerics is a multi-disciplinary approach to identifying the best combination of applications, physical models, numerical schemes, parallel algorithms, and platform-specific implementations that will deliver the fastest and most accurate results on a given platform. Hardware-aware numerics involves highly optimised platform-specific implementations, the design of communication-optimised data structures, and the data reuse through temporal and spatial blocking techniques, and the selection and development of appropriate mathematical models that express multi-level parallelism and scalability up to large numbers of cores.