Skip to main content

EPSRC ALEA Project

EPSRC ALEA project accords energy first-class status for many-core software developers

Parallel programmers are becoming increasingly aware of performance and correctness during the software development cycle. Unfortunately, they ignore the most precious resource that computing systems use today: energy. The EPSRC ALEA (Abstraction-Level Energy Accounting and Optimisation for Many-core Programming Languages) project accords energy first-class status in parallel programming languages. ALEA takes a radically new approach to energy characterisation and optimisation, by measuring the energy consumed by language-defined code and data abstractions, across the hardware components that each abstraction activates. Fine-grain energy accounting in ALEA breaks the current dependence of software on hardware-specific power models and opens up new pathways for energy optimisation in compilers and runtime systems. Dimitris Nikolopoulos, Director of Research in the HPDC Cluster, Bronis R. de Supinski, a visiting WLR Professor at Queen's and Zheng Wang, Lecturer in HPDC, will lead ALEA. They are joined by Co-Investigators from the Institute for Computing Systems Architecture at the University of Edinburgh. The project partners with ARM, IBM, Freescale Semiconductor, BluWireless, and Herta Security. The ALEA proposal was ranked first by the evaluation panel of the EPSRC SADEA programme, a part of the MACDES priority area.