Beowulf Clusters: From Research Curiosity to Exascale

Dan Reed and Early Beowulf Cluster

Dan Reed, Vice President for Research and Economic Development, will present on Beowulf Clusters: From Research Curiosity to Exascale. The presentation will be on January 22nd, 2015 from 1:00-2:15PM in 2520D UCC.

Presentation Overview: Twenty years ago, Sterling and Becker’s seminal work on Beowulf clusters and UC-Berkeley’s Network of Workstations (NOW) projects both stimulated widespread experiments with commodity hardware and open source software for scientific computing.  At that time, the second wave of the “attack of the killer micros”[1] was well underway, as the rising performance of inexpensive x86 processors made it possible to assemble clusters with aggregate performance substantially greater than that achievable with vendor-packaged systems and at a much lower cost.  Simply put, a new price-performance standard was being created for advanced computing, and the long-held dreams of terascale performance were now within reach. From creation of performance measurement and analysis toolkits through development of tiled display walls and development of parallel file systems to porting and tuning of numerical libraries and creation of standard Linux distributions and associated tools for cluster management, it was also a time of software excitement.  There was growing recognition that Linux was now sufficiently mature and reliable, with a vibrant and growing international base of software developers, to support production research computing at scale. Today, that possibility is not only a reality, it is the global standard for scientific computing. High-performance x86 processors, augmented with functional accelerators (e.g., Intel Xeon Phi or NVidia Tesla), and open source Linux software and associated toolkits now dominate advanced computing.  Petascale clusters are increasingly common and dominate the Top500 list of the world’s most powerful computers.

Biosketch: Daniel A. Reed is Vice President for Research and Economic Development, as well as University Chair in Computational Science and Bioinformatics and Professor of Computer Science, Electrical and Computer Engineering and Medicine, at the University of Iowa. Previously, he was Microsoft’s Corporate Vice President for Technology Policy and Extreme Computing, where he helped shape Microsoft's longterm vision for technology innovations in cloud computing and the company's associated policy engagement with governments and institutions around the world. 

Before joining Microsoft, he was the Chancellor’s Eminent Professor at UNC Chapel Hill, as well as the Director of the Renaissance Computing Institute (RENCI) and the Chancellor’s Senior Advisor for Strategy and Innovation for UNC Chapel Hill. Prior to that, he was Gutgsell Professor and Head of the Department of Computer Science at the University of Illinois at Urbana Champaign (UIUC) and Director of the National Center for Supercomputing Applications (NCSA). He was also one of the principal investigators and chief architect for the NSF TeraGrid. He received his PhD in computer science in 1983 from Purdue University. Dr. Reed served as a member of the President’s Council of Advisors on Science and Technology (PCAST) and the the President’s Information Technology Advisory Committee (PITAC). 

A recording of this presentation is available here: