UPTON, N.Y., April 20, 2017 — The U.S. Department of Energy’s (DOE) Brookhaven National Laboratory has joined the OpenMP (for Multi-Processing) Architecture Review Board (ARB). This nonprofit technology consortium manages the OpenMP application programming interface specification for parallel programming on shared-memory machines, in which any processor can access data stored in any part of the memory.
As part of this consortium of leading hardware and software vendors and research organizations, Brookhaven Lab will help shape one of the most widely used programming standards for high-performance computing—the combination of computing power from multiple processors working simultaneously to solve large and complex problems. Brookhaven’s participation in the OpenMP ARB is critical to ensuring the OpenMP standard supports scientific requirements for data analysis, modeling and simulation, and visualization.
“Advancing the frontiers of high-performance and data-intensive computing is central to Brookhaven’s mission in scientific discovery. Our membership in the OpenMP ARB recognizes the importance we place upon OpenMP for our science portfolio, both now and in the future,” said Robert Harrison, chief scientist of the Computational Science Initiative (CSI) at Brookhaven Lab and director of the Institute for Advanced Computational Science at Stony Brook University, which joined OpenMP ARB at the end of 2016.
Barbara Chapman, director of CSI’s Computer Science and Mathematics research team at Brookhaven and a professor of applied mathematics and statistics and of computer science at Stony Brook, led the initiative to join the OpenMP ARB. Chapman, whose research focuses on programming models for large-scale computing, has been involved with the evolution of OpenMP since 2001.
Abid Malik, a senior technology engineer on Chapman’s team, and research assistant Verinder Rana will represent Brookhaven during monthly meetings with the ARB. They plan to join several of the subgroups that focus on evolving specific aspects of the OpenMP programming model, including those for computational accelerators (such as graphics processing units, or GPUs), the C++ programming language, and memory management.
Each ARB member organization makes suggestions on how OpenMP should be evolved to meet their specific requirements. In turn, the vendors decide which suggestions to implement, depending on how relevant they are to a wide range of applications.
According to Malik, OpenMP will benefit from Brookhaven’s expertise in tackling big data challenges, especially those posed by its DOE Office of Science User Facilities—the Center for Functional Nanomaterials, National Synchrotron Light Source II, and Relativistic Heavy Ion Collider. Using this expertise, Brookhaven will help advance the OpenMP standard for next-generation supercomputers, which will help scientists tackle increasingly complex problems by performing calculations at unprecedented speed and accuracy.
“The OpenMP language subgroup is actively working with the scientific community to prepare OpenMP for exascale computing,” said Malik. “Brookhaven’s big data experience will help expand OpenMP to include features useful for porting big data programs on multicore CPUs [central processing units] and GPUs.”
About Brookhaven National Laboratory
Brookhaven National Laboratory is supported by the Office of Science of the U.S. Department of Energy. The Office of Science is the single largest supporter of basic research in the physical sciences in the United States, and is working to address some of the most pressing challenges of our time. For more information, please visit science.energy.gov.
Source: Ariana Tantillo, Brookhaven National Laboratory