PDL Abstract

FAWNSort: Energy-efficient Sorting of 10GB

Winner of 2010 10GB Joulesort, Daytona and Indy categories.

Vijay Vasudevan, Lawrence Tan, Michael Kaminsky*, Michael A. Kozuch*, David Andersen,
Padmanabhan Pillai

School of Computer Science
Dept. Electrical and Computer Engineering
Carnegie Mellon University
Pittsburgh, PA 15213

*Intel Research Pittsburgh

In this document, we describe our submission for the 2010 10GB JouleSort competition. Our system consists of a machine with a low-power server processor and five flash drives, sorting the 10GB dataset in 21.2 seconds (±0.227s) seconds with an average power of 104.9W (±0.8W). This system sorts the 10GB dataset using only 2228 Joules (±12 J), providing 44884 (±248) sorted records per Joule.

Our entry for the 10GB competition tried to use the most energy-efficient platform we could find that could hold the dataset in memory to enable a one-pass sort. We decided to use a one-pass sort on this hardware over a two-pass sort on more energy efficient hardware (such as Intel Atombased boards) after experimenting with several energy efficient hardware platforms that were unable to address enough memory to hold the 10GB dataset in memory. The lowpower platforms we tested suffered from either a lack of I/O capability or high, relative fixed power costs, both stemming from design decisions made by hardware vendors rather than being informed by fundamental properties of energy and computing.