PARSA researchers win the Best Paper Runner-Up award at MICRO'13
© 2014 EPFL
Onur Koçberber, Boris Grot, Javier Picorel and Prof. Babak Falsafi affiliated with the Parallel Systems Architecture Laboratory (PARSA) along with co-authors, were honored with the Best Paper Runner-Up award at the 46th International Symposium on Microarchitecture (MICRO-46) which is the premier forum for presenting, discussing, and debating innovative microarchitecture ideas and techniques for advanced computing and communication systems. This year MICRO was particularly competitive because it had an acceptance rate of only 16%.
Businesses, governments and societies are increasingly relying on collecting, analyzing and exchanging data to improve their products, services and ultimately to enhance our lives. The fundamental operation of data processing is to locate and extract the vital pieces of knowledge from a dataset. Doing so efficiently mandates the use of indexing data structures and algorithms.
In the awarded paper, the researchers identify indexing operations as a critical efficiency bottleneck in modern processors executing data analytics workloads on contemporary database management systems (DBMSs). They propose efficient hardware units, called Walkers, to improve both performance and energy efficiency of indexing operations. The programmable nature of Walkers guarantees broad applicability to both existing and future analytics applications, while the simple programming model facilitates adoption and integration into existing DBMSs. The Walkers make indexing operations fast, energy-efficient, and accessible. Read here to learn more about the paper.
Onur Koçberber and Javier Picorel are both pursuing a PhD in the PARSA lab directed by Prof. Babak Falsafi. Boris Grot was a post-doctoral researcher in the lab and he recently joined University of Edinburgh as an assistant professor.
The PARSA lab is a part of the EcoCloud research consortium, which pioneers technologies to make cloud computing scalable, cost-effective and sustainable.
