ORNL Delivers Knowledge, Expertise at SC16
OLCF staff share research and accomplishments with the HPC community
The US Department of Energy’s (DOE’s) Oak Ridge National Laboratory (ORNL) and Oak Ridge Leadership Computing Facility (OLCF), a DOE Office of Science User Facility located at ORNL, wrapped up a strong showing at SC16, the annual international conference of high-performance computing (HPC), networking, storage, and analysis.
Held November 13–18 in Salt Lake City, Utah, this premiere HPC event gave OLCF staff members the opportunity to share their work and knowledge with the HPC community and learn about the latest technologies and accomplishments from the world’s leading vendors, research organizations, and universities. The theme of this year’s conference was “HPC Matters.”
At the conference the OLCF was recognized by the HPCwire Readers’ Choice Awards for the Best Use of HPC Application in the Energy Industry. The award was for a gas turbine combustion simulation project conducted by General Electric on the Titan supercomputer.
OLCF users and partners also received recognition in Salt Lake City. OLCF user and Imperial College Professor Peter Vincent and his team were finalists for the prestigious Gordon Bell Prize for innovative, high-fidelity simulation of lightweight jet turbines. Axel Huebl, an OLCF user and doctoral researcher at Helmholtz-Zentrum Dresden-Rossendorf, received the George Michael Memorial HPC Fellowship, and Sunita Chandrasekaran, an assistant professor at the University of Delaware and GPU hackathon organizer, garnered the Award for Excellence for Early Career Researchers in HPC.
During SC16 OLCF staff members participated in a wide variety of activities, including tutorials, workshops, panel discussions, invited talks, poster sessions, technical paper presentations, and birds-of-a-feather sessions.
At the DOE booth OLCF Director of Science Jack Wells gave a presentation on Summit, the OLCF’s next leadership-class supercomputer, sharing details on the hybrid system’s planned architecture.
OLCF Computational Scientist Matt Norman delivered the keynote address at the third Workshop on Accelerator Programming Using Directives (WACCPD), which was held in conjunction with SC16. WACCPD is one of the major forums devoted to sharing knowledge and experience related to using directives and programming complex systems. The workshop also featured a paper presentation by ORNL researcher Graham Lopez, who authored “Towards Achieving Performance Portability using Directives for Accelerators” along with coauthors and OLCF staff members Wayne Joubert, Oscar Hernandez, and Veronica Vergara Larrea.
Other SC16 workshops that featured OLCF staff included a forum focused on the development of scientific algorithms for multipetascale and exaflop systems. Entitled the Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, the event included OLCF Chief Technology Officer Al Geist as an organizer. OLCF User Assistance Support Specialist Fernanda Foertter and National Center for Computational Sciences Director of Computing and Facilities Jim Rogers helped stage workshops addressing HPC training and supercomputing center operations, respectively.
Several papers coauthored by OLCF staff members were presented at SC16, including a paper analyzing how job placement affects performance on large-scale systems. The paper, entitled “A Multi-Faceted Approach to Job Placement for Improved Performance on Extreme-Scale Systems,” listed OLCF Technology Integration Group Leader Sudharshan Vazhkudai, OLCF HPC Systems Engineer Scott Atchley, and OLCF HPC Systems Engineer Chris Zimmer as coauthors.
Other OLCF staff members acknowledged as contributors to SC16 papers included Technology Integration Specialist Raghul Gunasekaran, Computer Visualization Specialist David Pugmire, and Computer Scientist Devesh Tiwari.
OLCF staff members also shared their expertise with attendees during multiple tutorials. OLCF Computational Scientist Judy Hill and Hernandez presented on GPU-accelerated POWER nodes at the tutorial entitled “Application Porting and Optimization on GPU-Accelerated POWER Architectures.” Additionally, Foertter and Pugmire contributed to tutorials on parallel programming in Fortran and scalable HPC visualization, respectively.
Birds-of-a-feather sessions, forums that allow participants to gather and discuss topics of common interest, featured OLCF session moderators including Computer Science Research Group Leader Dave Bernholdt, Rogers, Atchley, and Foertter.
Among panel participants, OLCF HPC Systems Administrator Matt Ezell contributed to “Data Analytics Support for HPC System Management,” while Vergara Larrea provided perspective for “Life on the Other Side: Early Career Panel.”
In other conference happenings, the SC16 poster session gave OLCF Storage System Evaluation Lead Sarp Oral, File System Administrator Feiyi Wang, and Vazhkudai an opportunity to showcase research findings related to large-scale HPC storage systems in a poster entitled “An I/O Load Balancing Framework for Large-Scale Applications (BPIO 2.0).”
Leading up to SC16, OLCF staff members helped shape this year’s conference by participating in organizing committees. ORNL staff members who chaired committees for SC16 included Vergara Larrea, who served as the Students @ SC Student Development Chair and Student Job Fair Chair; OLCF Networking Administrator Benny Sparks, who was the Wireless Team Cochair; and Rogers, who was the Logistics/Equipment Team Chair.
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.