Computer Systems and Networks


Energy efficient algorithms and systems

This topic deals with a novel area of algorithms and systems that are designed to balance energy consumption with the actual achievable performance. The design of energy efficient systems is considered as a key prerequisite to the construction of super-large scale computer systems (the so called exa-scale and beyond supercomputers). The energy efficient algorithms, while being a part of the planning for this super-large scale systems do have even more immediate use in mobile devices (the energy consumption may be more important aspect than the speed) and also in contemporary computer centers where energy bill is becoming a limiting factor for their grows. Student will get deeper understanding of the motivations and the solution currently employed in the design of both the energy efficient algorithms and systems.

Literature:

As there are currently no books covering this area, student is expected to use the journal and conference papers and related material to get the proper perspective and knowledge of the area. The primary bibliographical sources are:
Papers and presentations (co-)authored by Susanne Albers, e.g. Energy-Efficient Algorithms, Communications of the ACM, Vol. 53(5), Pages 86–96, 2010, related slides
On multi-processor speed scaling with migration, Journal of Computer and System Sciences, Vol. 81(7), November 2015, Pages 1194–1209
Papers from the Performance and Algorithms group at Berkley Lab and specifically the EDGAR project there on Energy Efficient Data and Graph algorithms
Further reading is also
Donofrie et al: Energy efficient computing for extreme-scale science, Computer, November 2009, pages 62–71
Dreslinski et al: Near-Threshold Computing: Reclaiming Moore’s Law Through Energy Efficient Integrated Circuits, Proceedings of the IEEE, Vol. 98(2), pages 253–266, 2010