Dr Karakonstantis awarded HiPEAC Paper Award
Dr Georgios Karakonstantis, a Senior Lecturer at the Institute of Electronics, Communications and Information Technology (ECIT) at Queen’s University Belfast (QUB) was awarded a HiPEAC Paper Award by the steering committee of the HiPEAC Network of Excellence for the recent publication of his team at the top International ACM ASPLOS conference.
HiPEAC is a network, the biggest of its kind in Europe, consisting of 2,000 world-class computing systems researchers and industry representatives working on High-Performance, Embedded, Architecture, and Compilation research areas. Every year HiPEAC awards members, who have accomplished to publish a paper in one of the top, highly competitive international computing systems conferences, where European research is under-represented.
This year only ECIT's team with 2 other teams from the UK (Cambridge, Edinburgh) and 7 more European research groups accomplished to publish in ACM ASPLOS, the International Conference on Architectural Support for Programming Languages and Operating Systems.
ASPLOS is the premier forum for interdisciplinary systems research, intersecting computer architecture, hardware and emerging technologies, programming languages and compilers, operating systems, and networking and is sponsored by ACM and world leading industry like ARM, Intel, AMD, IBM, Google, Facebook, Huawei. This year ASPLOS received a record number of 486 submissions from groups around the world and it was very competitive with an acceptance rate of 17%.
Dr Karakonstantis' paper, which was awarded, is entitled HaRMony: Heterogeneous-Reliability Memory and QoS-Aware Energy Management on Virtualized Servers and was co-authored by his PhD student: Konstantinos Tovletoglou, Research Fellow: Lev Mukhanov, and Prof. Dimitrios Nikolopoulos, now at Virginia Tech.
The team developed a framework that enables intelligent trade-offs between memory power and reliability through selective allocation of data on different memory domains and dynamic adjustment of DRAM operating settings. The developed QoS and temperature aware power management scheme was implemented on a real ARM based server, where it was shown that it can lead to ~8% system energy savings. An overview of the work can be found in the conference presentation.