CUG Program Chair Nicholas Cardo (far left) standing with the conference’s best paper award recipients. The best paper went to OLCF staffers (left to right) Veronica Vergara Larrea, Chris Fuson, and Wayne Joubert.
Staff members win best paper award, serve in leadership roles
The US Department of Energy’s (DOE’s) Oak Ridge National Laboratory (ORNL) has long been partnered with Cray for state-of-the-art supercomputing technology.
That partnership was on full display during the 2015 Cray User Group (CUG) conference April 26–30 in Chicago. Staff from the Oak Ridge Leadership Computing Facility (OLCF), a DOE Office of Science User Facility, served in leadership positions during the conference, and the team of OLCF staff members Veronica Vergara Larrea, Chris Fuson, and Wayne Joubert won an award for the conference’s best paper.
The paper, “Use of Continuous Integration Tools for Application Performance Monitoring,” reviewed a variety of different tools to more effectively monitor application performance on the OLCF’s high-performance computing (HPC) systems. “Our goal was to identify an existing open-source software tool that could be used to track environment stability and application performance over time,” Vergara Larrea said. “We evaluated a variety of continuous integration tools and determined that the Jenkins software ultimately met the majority of our requirements and was the best candidate to be used as the basis for the monitoring system.”
Vergara Larrera also was an author on another paper, “A More Realistic Way of Stressing the End-to-End I/O System,” which was a finalist for the conference’s best paper. That paper’s other authors included ORNL staff members Sarp Oral, Dustin Leverman, Feiyi Wang, and James A. Simmons, as well as former OLCF computational scientist Hai Ah Nam.
Fuson also participated in a “birds of a feather” session (for conference participants with specific shared interests) titled “System Testing and Resiliency in HPC.”
National Center for Computational Sciences Computing and Facilities Director Jim Rogers was reelected as CUG treasurer. He also presented a detailed study of GPU reliability and performance on the OLCF’s Cray XK7 Titan supercomputer. The presentation, “Experience with GPUs on the Titan Supercomputer from a Reliability, Performance, and Power Perspective,” also included research from OLCF staff members Devesh Tiwari, Saurabh Gupta, and Don Maxwell.
OLCF User Assistance and Outreach Group leader Ashley Barker chaired a session about system testing and resiliency in HPC and served as a deputy chair for the Programming Environments, Applications, and Documentation special interest group along with OLCF user support specialist Suzanne Parete-Koon.
Parete-Koon also worked with OLCF intern Jake Wynne to present a paper on archiving big data on the OLCF’s High-Performance Storage System. The presentation, Wynne’s first technical talk, highlighted a factor of 2 increase in data transfer speed after the group optimized the Hierarchical Storage Interface transfer agent.
In addition, Parete-Koon organized a “birds of a feather” session, “Getting the Most Out of Your User Group,” where user support staff began drafting a best practices document about organizing working groups for specific scientific domains or code bases.
Jason Hill, Storage Team lead at the OLCF, chaired the systems support special interest group, was a member of the programming committee, and was chair of the papers committee, which selected the conference’s best paper.
ORNL research scientist and OLCF user Sreenivas Sukumar spoke at the conference about ORNL computing resources for data analysis. Sukumar presented on the Urika-XA and Urika-GD Big Data Cray architectures, procured as part of the Compute and Data Environment for Science (CADES) initiative.Sukumar demonstrated the “app store” of scalable machine learning algorithms developed at ORNL and highlighted how these applications act as building blocks for scientists to develop domain-specific variations of algorithms.
User support specialist Fernanda Foertter gave a talk highlighting recent hackathon activities at ORNL and the National Center for Supercomputing Applications. This closed out a strong showing for the OLCF at CUG.
Oak Ridge National Laboratory is supported by the US Department of Energy’s Office of Science. The single largest supporter of basic research in the physical sciences in the United States, the Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit science.energy.gov.
