Mission and History

In 2002, Argonne National Laboratory established the Laboratory Computing Project to enable and promote the use of high-performance computing (HPC) across the Laboratory in support of its varied research missions.

The Laboratory Computing Resource Center was established, and in April 2003 LCRC began full operations with Argonne’s first teraflop computing cluster, Jazz. In 2010, Jazz was replaced by Fusion, with a peak performance of 30 teraflops. In late 2012, the Blues cluster replaced Fusion as LCRC’s primary computing resource, with an initial peak performance of 108 teraflops (60 Haswell nodes were added later). In early 2017, Fusion was retired to make way for Bebop, with a peak performance of 1 Petaflop. Our newest cluster, Improv, with a peak performance of 3.5 Petaflops, has been in production since October 2023.

LCRC’s initial objectives were to operate the computing facility, educate staff and students about HPC, and support application use and development. That approach was very successful, and today every Argonne research division uses LCRC.  LCRC resources are available not only to the entire laboratory community, but also to their academic and industrial collaborators.  LCRC staff continues to provide guidance to its users on application usage and evaluating the performance of their application codes.  As an added benefit, experience using the LCRC cluster enables users to transition some or all of their workload to leadership-class machines like ALCF.

Services

  • Provide mid-range supercomputing resources for Laboratory research projects, expanding our research horizons, supporting laboratory initiatives, and making us more competitive.
  • Provide user assistance and training, making science and engineering projects more successful and productive.
  • Provide a spectrum of scalable applications and tools, and train people to use them, enabling larger and more complex studies.
  • Promote datacenter efficiency and consolidation, e.g., via programmatic additions to the production clusters dedicated to specific projects.