This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| Edison (supercomputer) | |
|---|---|
| Name | Edison |
| Manufacturer | Cray |
| Location | Lawrence Berkeley National Laboratory |
| Country | United States |
| Date | 2012–2017 |
| Architecture | Cray XC30, Intel Xeon |
| Peak | 2.57 petaflops |
| Memory | 357 TB |
| Storage | Lustre filesystem |
Edison (supercomputer) was a petascale high-performance computing system deployed at the Lawrence Berkeley National Laboratory and operated by the National Energy Research Scientific Computing Center to support scientific research across multiple domains. Commissioned in 2012 and delivered by Cray Inc. using Intel Xeon processors, Edison served users from National Laboratories, universities, and industrial collaborations until decommissioning in 2017. The system contributed to projects involving climate modeling, astrophysics, materials science, and computational chemistry under programs such as the Department of Energy's Office of Science initiatives.
Edison was installed at the Lawrence Berkeley National Laboratory under the auspices of the U.S. Department of Energy and managed by the National Energy Research Scientific Computing Center, joining contemporaries like Titan (supercomputer) and Sequoia (supercomputer). Built by Cray Inc. as a Cray XC30 system using Intel Xeon processors, Edison provided a multi-petaflop capability to support flagship computational campaigns from Berkeley Lab researchers, collaborators at the University of California, Berkeley, and external teams funded through Office of Science allocations. The system's procurement and operation were influenced by policies from the Office of Management and Budget and requirements articulated in Advanced Scientific Computing Research programs.
Edison's hardware architecture was based on the Cray XC30 cabinet design integrating thousands of Intel Xeon E5 cores, high-speed interconnects from Cray Aries, and a global shared storage implemented via the Lustre filesystem. The machine comprised compute blades populated with multi-socket nodes linked by a dragonfly network topology and cooled within the Lawrence Berkeley National Laboratory data center infrastructure that included ASHRAE-guided environmental controls and Chiller systems. Memory capacity aggregated across nodes reached hundreds of terabytes, while the parallel filesystem provided multi-petabyte capacity and high I/O bandwidth for workflows related to NERSC projects, High Performance Computing kickoffs, and collaborative campaigns with the National Nuclear Security Administration and academic partners.
Edison achieved a theoretical peak performance on the order of 2.57 petaflops, with LINPACK benchmarks reporting figures comparable to contemporaneous systems such as Mira and earlier Jaguar (supercomputer). Application benchmarks using codes from NERSC workloads, including GROMACS, LAMMPS, and CESM, demonstrated scaling behavior influenced by the system's Cray Aries interconnect latency and node memory layout; performance studies were compared against metrics from Top500 listings and analyses by High Performance Computing centers. Operational performance monitoring used telemetry tied to Cray software stacks and system health dashboards aligned with best practices from XSEDE and OpenStack-era workflows.
The software environment on Edison included the Cray Linux Environment and a suite of compilers and libraries from vendors such as Intel and GNU toolchains, alongside message-passing implementations like MPI and parallel I/O libraries compatible with Lustre. Scientific packages and toolchains provided by NERSC encompassed performance analysis tools such as TAU and Statsite integrations, build systems leveraging CMake and ecosystem management practices aligned with GitHub-hosted repositories used by researchers from Lawrence Berkeley National Laboratory and partner institutions. Edison supported batch scheduling via SLURM and workflow integration with community portals including Science Gateway initiatives and data management policies interoperable with the DOE Data Management Plan expectations.
Operational responsibilities for Edison were handled by the National Energy Research Scientific Computing Center staff located at Lawrence Berkeley National Laboratory and coordinated with Cray service teams and facilities engineers. The data center facility adhered to standards referenced by ASHRAE and collaborated with Pacific Gas and Electric Company for electrical provisioning and power distribution units, while environmental monitoring interfaced with Department of Energy oversight. User support, allocations, and training were managed through mechanisms established by NERSC, including user accounts for personnel from universities, national laboratories, and industrial partners under peer-reviewed allocation schemes.
Edison enabled research across a wide array of domains including climate science campaigns using the Community Earth System Model and coupled simulations for atmospheric sciences, large-scale cosmology simulations for dark matter structure formation, high-resolution computational fluid dynamics for aerodynamics and combustion studies, and molecular dynamics investigations in materials science and biochemistry using packages like NAMD and GROMACS. Collaborative projects involved teams from Lawrence Livermore National Laboratory, Los Alamos National Laboratory, University of California, San Diego, and international partners, producing datasets archived in repositories aligned with DOE data stewardship and informing publications in journals such as Science and Nature.
Edison was formally decommissioned in 2017 as part of lifecycle planning at NERSC to make way for next-generation systems like Cori (supercomputer) and subsequent Perlmutter (supercomputer) procurements. Hardware retirement followed protocols coordinated with Cray, DOE asset management, and recycling programs consistent with federal property disposition guidance. The system's operational history contributed to lessons captured by NERSC about scaling on Cray XC30 architectures, influenced procurement and software support practices at Lawrence Berkeley National Laboratory, and supported a corpus of scientific publications and datasets that continue to inform research across computational science communities.