Announcements - Written by on May 19, 2017

OLCF Users to Elect New OUG Executive Board Members

Candidates will speak during OUG business meeting

Five candidates are running for three open positions on the Oak Ridge Leadership Computing Facility’s User Group (OUG) Executive Board. The OUG Executive Board is responsible for representing the needs of users and providing feedback on services and resources to OLCF leadership.

OLCF users are eligible to vote for the candidates between Monday, May 22 and Thursday, May 25. Ballots will be distributed to users via email Monday morning. Voting will close at 11 a.m. Thursday. Users are instructed to vote for their top three candidates. (Read statements from the candidates below.)

An event to learn more about the candidates will take place 4 p.m. EDT Wednesday, May 24 during the OUG business meeting. Each candidate will have three minutes to give a “lightning campaign talk,” explaining how he will represent OLCF users. Audience members will have an opportunity to ask questions after each talk.

Users who are at Oak Ridge National Laboratory for the 2017 OLCF User Meeting can attend the meeting—to be held in the Tennessee ABC room—in person. The meeting will also be livestreamed via BlueJeans. View the livestream here: https://bluejeans.com/916171904/.

Statements provided by the candidates are printed below.

James McClure, Virginia Tech

“As an INCITE PI, I have developed GPU-based simulation libraries, a framework for in situ analysis, and tools to connect experimental data sources with HPC simulation. I hope to serve OLCF users by working to ensure that scientific workflows benefit fully from the fundamental advancements that are taking place in computing.  As we move toward exascale, the efficiency of OLCF user workflows will be key to fostering a productive environment for science on future systems.”

Abhishek Singharoy, Arizona State University

“As one of the first to perform molecular dynamics simulations with 100 million atoms, I will facilitate deployment of GPU-friendly algorithms for biological and energy computations; user feedback from my work has already improved the performance of NAMD by 30 percent on Titan. Second, I will advocate remote visualization tools, e.g. using EGL, so users can skip time-consuming data transfer steps prior to analysis. Finally, I will train users in ‘good practices’ with biological simulations on Summit.”

Travis Johnston, Oak Ridge National Laboratory

“My experience with data analytics at scale will provide a unique perspective to the Executive Board. I will promote the use of resources required for leadership-class data analytics to accompany modeling and simulation. My HPC experience includes a recent success building and configuring Apache Spark on Titan (without MPI), running it at scale—18,000 nodes—and setting a new world record for the size of a Spark job. I use Spark to design deep neural networks for applications to scientific datasets.”

Michael Zingale, Stony Brook University

“I am a computational astrophysicist, user of ORNL computers for around 10 years, current board member, and a professor. This past term, I gave two presentations on the OLCF monthly conference call—on OpenACC and remote visualization using yt. Education is important to me, and I will continue to make presentations to demonstrate the techniques we’ve been using. I am also a strong believer in open codes—all of our code is on GitHub, allowing us to share the HPC techniques we learn with the community.”

Shantenu Jha, Rutgers University

Shantenu Jha, Rutgers University

Shantenu Jha is an associate professor of computer engineering at Rutgers University. He is a recipient of the National Science Foundation CAREER Award (2013) in cyberinfrastructure, and several prizes at SC’xy and ISC’xy. He was the recipient of the inaugural Rutgers Chancellor’s Excellence in Research (2016) for his cyberinfrastructure contributions to computational science. His research interests are at the intersection of high-performance distributed computing as applied to computational science. He collaborates extensively with scientists from multiple domains—including but not limited to molecular science, earth science, and high-energy physics. In each of these three domains he collaborates with OLCF users—designing and deploying tools and libraries to enable them to use OLCF resources in novel and diverse ways. These range from “iterative simulation and analysis” workflows to integrating OLCF resources with the distributed Worldwide LHC Computing Grid (WLCG). Jha’s objective is to represent and advocate the requirements of “non-traditional” high-performance computing, ranging from middleware deployment to policies supporting workflows experimentation.