Computation Contributes to Top HPC Conference
The world’s supercomputing experts gathered in New Orleans at the Supercomputing Conference (SC14) to discuss their work, see the latest equipment, and glimpse the future of computing, technology, and computational research.
Computation Deputy Associate Director Trish Damkroger served as conference chair for the event, which took place from November 16–21. Forty-six other LLNL staff members also had a hand in this year’s conference, by contributing to its technical program and participating in a full year of planning and coordination.
LLNL’s Computation directorate also had a strong showing in the conference’s technical content and among the award winners.
No. 1 Rank in Graph 500
In their search for new ways to solve difficult national security problems with the Sequoia supercomputer, LLNL scientists helped achieve the world’s best performance on the Graph 500 data analytics benchmark. The Graph 500 offers performance metrics for data-intensive computing or “big data,” an area of growing importance to the HPC community.
“To fulfill our missions in national security and basic science, we explore different ways to solve large, complex problems, most of which include the need to advance data analytics,” said Dona Crawford, associate director for Computation. “These Graph 500 achievements are a product of that work performed in collaboration with our industry partners. Furthermore, these innovations are likely to benefit the larger scientific computing community.”
LLNL and IBM computer scientists attained the No. 1 ranking by completing the largest problem scale ever attempted—scale 41—with a performance of 23.751 teraTEPS (trillions of traversed edges per second). The team employed a technique developed by IBM.
In addition to achieving the top Graph 500 ranking, LLNL computer scientists also demonstrated a scalable new technique for solving large graph problems on small clusters and even a single node using the Catalyst system. To achieve these results, Livermore computational researchers have combined innovative research in graph algorithms and data-intensive runtime systems.
HPCWire Editor’s Choice Award
The partnership that produced the first-of-a-kind Catalyst supercomputer was selected for an HPCWire “Best HPC Collaboration Between Government and Industry” award by readers and editors of the publication.
HPCWire publisher Tom Tabor presented the award to representatives from LLNL, Intel, and Cray in the DOE booth at SC14. HPCWire is a leading online news service covering the HPC industry.
A Cray CS300 high performance computing cluster, Catalyst was developed to test big data technologies, architectures, and applications.
Young Achievers in Scalable Computing Award
Computer scientist Abhinav Bhatele was recognized at SC14 for his HPC research by the IEEE Technical Committee on Scalable Computing. The Young Achievers in Scalable Computing Award acknowledges individuals within five years of receiving their Ph.D. degree who have made outstanding, influential, and potentially long-lasting contributions in the field of scalable computing. Abhinav’s work focuses on how researchers can use present and future HPC architectures with the highest possible efficiency.
Poster Awards, Fellowship for Former LLNL Interns
Three outstanding former LLNL interns were recognized during the SC14 awards ceremony.
University of California, Davis, student Dylan Wang won second prize in the ACM SRC Best Poster award in the undergraduate category. His poster was based on work he did at LLNL this summer on performance variability in supercomputers in collaboration with Abhinav Bhatele.
University of Illinois at Urbana-Champaign graduate student Amanda Bienz won first prize in the ACM SRC Best Poster award in the graduate category for her poster “Reducing Network Contention Associated with Parallel Algebraic Multigrid.” Amanda completed a graduate internship with Jacob Schroder this summer.
Harshitha Menon, also a graduate student at the University of Illinois at Urbana–Champaign, won the George Michael Memorial HPC PhD Fellowship, which honors exceptional PhD students around the world whose research focuses on HPC, networking, storage, and large-scale data analysis. Harshitha completed a graduate internship with Abhinav Bhatele in 2013.
In other supercomputing news, the latest edition of the TOP500 list of the world’s most powerful computers was released in New Orleans. The ranking of the five most powerful systems remains unchanged from the previous list announced in June of this year. LLNL’s Sequoia remains No. 3 and Vulcan No. 9. China’s Tianhe-2 (Milky Way - 2) retained the No. 1 ranking, and Oak Ridge National Laboratory’s Titan No. 2. Four of the top 10 supercomputers are at Department of Energy laboratories.