Powering Scientific Discovery Since 1974
Contact: Jon Bashor, email@example.com, +1 510 486 5849
The oil crisis of 1973 did more than create long lines at the gas pumps — it jumpstarted a supercomputing revolution.
The quest for alternative energy sources led to increased funding for the Department of Energy's Magnetic Fusion Energy program, and simulating the behavior of plasma in a fusion reactor required a computer center dedicated to this purpose. Founded in 1974 at Lawrence Livermore National Laboratory, the Controlled Thermonuclear Research Computer Center was the first unclassified supercomputer center and was the model for those that followed.
Over the years the center's name was changed to the National Magnetic Fusion Energy Computer Center and later the National Energy Research Supercomputer Center (NERSC). In 1983 NERSC's role was expanded beyond the fusion program, and it began providing general computing services to all of the programs funded by the DOE Office of Energy Research (now the Office of Science). The current name was adopted in 1996 when NERSC relocated to Lawrence Berkeley National Laboratory and merged with Berkeley Lab's Computing Sciences program. The name change — from "Supercomputer Center" to "Scientific Computing Center" — signaled a new philosophy, one of making scientific computing more productive, not just providing supercomputer cycles.
A Borrowed Supercomputer
Back in 1974, the center began operation with a borrowed Control Data Corporation 6600 computer. Initially most users had to connect modems to their terminals and dial up this computer via telephone, though some early users were able to use the ARPA network for connectivity. In 1975, a CDC 7600 replaced the 6600. In 1976, dedicated, leased phone lines connected the center with the major fusion energy research sites, allowing users to log into the 7600 from their local DEC PDP-10 minicomputers. The machine was rapidly filled to capacity, and for a while we had to purchase additional 7600 time from Berkeley Lab.
The center acquired a Cray 1 in 1978 and soon became known as an innovator in the management and operation of supercomputers. We converted our 7600 operating system, utilities, and libraries to the new machine, creating the Cray Time Sharing System (CTSS) — the first timesharing system for a Cray — and demonstrating that the machine could be used interactively. CTSS was subsequently adopted by nine other computer centers.
With each new system, data storage capabilities had to evolve as well. When NERSC first opened its doors in 1974, files were typically measured in megabytes. In 1976, the center could store a whopping 19,200 megabytes of data, primarily on online disks and nine-track tapes. But back then, users’ computing needs, and the hardware and software available to meet them, were quite different too. For example, managing data required staffers to move around behind the scenes. When a user filed a request, an operator would retrieve the tape from a rack and load it, then notify the user that the data were available. In some cases, the tapes were stored in a separate building at LLNL and were often picked up by a staffer riding a bicycle! Fortunately, delivery of the Automated Tape Library in 1979 changed this practice by allowing hands-off access.
The world's first Cray 2, a four-processor system, was installed at the center in 1985. We had already spent two years preparing the CTSS operating system for multitasking. This preparation paid off when the Cray 2 was available to users only one month after delivery.
Over the next decade, NERSC continued to enhance the productivity of cutting-edge computer architectures. In August 1997, NERSC achieved a milestone: successfully stopping and restarting a number of scientific computing jobs on a Cray T3E without any data processing loss or discontinuity. Called "checkpointing," the stop/restart procedure is believed to be the first time such a procedure has been accomplished on a massively parallel processing (MPP) system. This accomplishment opened a new era of robust, reliable, production-mode MPP computing.
Breaking the Petaflop Barrier
In 2010, the center broke the quadrillions-of-calculations-per-second mark with Hopper, a Cray XE6 named for pioneering computing scientist Grace Murray Hopper. With more than 150,000 processor cores, the system is capable of peak theoretical performance of 1.05 petaflops (quadrilliions of "floating point operations" per second). At the time, Hopper operated alongside a second supercomputing system (Franklin, a Cray XT4 accepted in 2009), an IBM iDataplex cluster (Carver) and various special-purpose and testbed systems, including PDSF (dedicated to high-energy physics) and the Joint Genome Institute's system (dedicated to genomic research). Carver was retired in 2010, Franklin in 2012 and Hopper in 2015.
Today NERSC provides some of the largest open computing and storage systems available to the global scientific community and continually evolves its systems to ensure that users are never presented with an entirely new system at any one time. The center is currently home to two world-class supercomputers: Edison, a Cray XC30 installed in 2012-2013, and Cori, a Cray XC40 installed in 2015-2016. And as the amount of scientific data continues to grow, NERSC has kept pace with the High-Performance Storage System (HPSS), an archival storage system that currently holds about 90 petabytes of data (more than 179 million files) and has a maximum capacity of 240 petabytes of data. DOE's Energy Sciences Network (ESnet)—the high-speed computer network managed by Berkeley Lab—has also grown and evolved along with NERSC, moving beyond computer access to provide a full range of communication services for DOE scientists.
The Next Generation
In 2015, NERSC moved from the Oakland Scientific Facility in downtown Oakland, where it had been since 2000, back to the Lab's main campus in the hills above UC Berkeley. Today NERSC shares the state-of-the-art Shyh Wang Hall with the rest of the Computing Sciences organization: ESnet and the Computational Research Division. Today NERSC serves nearly 7,000 researchers working on 700 projects annually, with users in 2016 reporting more than 2,200 scholarly papers made possible by NERSC resources.
In fact, NERSC has become an active partner in scientific programs that both shape and benefit from high-performance computing. Our high-performance computing systems and excellent client services are now complemented by expanded expertise in computational and computing science. In addition, with the continued growth of experimental data coming from facilities all over the U.S. and other countries, NERSC has become a net data importer, taking in a petabyte of data each month for storage, analysis and sharing in fields ranging from bioscience and environmental studies to cosmology and high-energy physics.
As a result, the center's focus on data science has expanded as well. The Cori system was purposely designed to support these efforts, with a number of unique features that enhance data movement, analysis and job scheduling. In addition, NERSC has implemented several programS aimed at improving users' code and application performance on the Cori architecture, thus further enhancing their scientific productivity. Finally, taking advantage of the center's proximity to the University of California-Berkeley, the Mathematical Sciences Research Institute, major computer and data communications companies in the San Francisco Bay Area and the diverse scientific research programs at Berkeley Lab, NERSC continues to achieve a level of scientific and technological collaboration unequaled by any other high performance computer center.
About NERSC and Berkeley Lab
The National Energy Research Scientific Computing Center (NERSC) is a U.S. Department of Energy Office of Science User Facility that serves as the primary high-performance computing center for scientific research sponsored by the Office of Science. Located at Lawrence Berkeley National Laboratory, the NERSC Center serves more than 7,000 scientists at national laboratories and universities researching a wide range of problems in combustion, climate modeling, fusion energy, materials science, physics, chemistry, computational biology, and other disciplines. Berkeley Lab is a DOE national laboratory located in Berkeley, California. It conducts unclassified scientific research and is managed by the University of California for the U.S. DOE Office of Science. »Learn more about computing sciences at Berkeley Lab.