Virtual Institute — High Productivity Supercomputing

READEX

EU (2015 - 2018)

Runtime Exploitation of Application Dynamism for Energy-efficient eXascale computing (READEX)

The importance of energy efficiency is constantly increasing in High Performance Computing (HPC). While systems can be adapted to individual applications in order to reduce energy consumption, manual tuning of platform parameters is a tedious and often neglected task. The READEX projects automates this by developing a tools-aided methodology for dynamic auto-tuning: It combines technologies from two ends of the computing spectrum: system scenario methodology from the embedded world and auto-tuning from the field of HPC.

The READEX methodology has been designed for exploiting the dynamic behaviour of software. At design time different functions are detected and optimized system configurations are determined. Functions with the same configuration are grouped into scenarios. This analysis is carried out with the Periscope Tuning Framework. It uses a multi-agent based approach to identify significant functions and to determine optimized system configurations. It also provides means for the specification of domain knowledge to improve the automatic tuning results. Part of this is the specification of application tuning parameters. These allow the developer to offload decisions about optimal algorithms the tool suite. The result of the analysis step, the tuning model, guides runtime tuning. During production runs of the user’s application, the READEX Runtime Library takes control. It is designed to apply the different configurations in a lightweight manor. Moreover, the READEX Runtime Library, which is implemented as a Score-P substrate plugin, will be able to adapt to a changing application behaviour. The latter is implemented by state-of-the-art machine learning mechanisms.

Further Information

Partners

Sponsor

European Union‘s Horizon 2020 research and innovation programme