BAE scores $699 million contract to support US Army supercomputers

Defense giant BAE Systems has secured a five-year $699 million contract with the US Army in support of the force's high-performance computing (HPC) programs.

According to BAE, it was awarded to cover Defense Supercomputing Resource Center (DSRC) operations and management services in support of the Army's High Performance Computing Modernization Program (HPCMP), an ongoing project to provide the necessary supercomputing resources to support research and provide timely information for decision makers on the battlefield.

BAE is perhaps better known for aircraft, armored vehicles, and warships than IT systems, but the company says it will provide the Army with "management and technical support" for the DSRC and its HPCMP operations as part of this $699 million contract.

We asked BAE what its exact role in the program would be and were told it would involve providing technical support services at four of the five US Defense Supercomputing Resource Centers (DSRC).

"These sites deliver the Department of Defense's High Performing Computing Modernization Program's (HPCMP) capability to provide DoD's scientists and engineers with the resources necessary to solve the most demanding problems through the strategic application of HPC, networking, and computational expertise," it said.

"In this role, BAE Systems will provide DSRC center operations, maintenance, and management of the capability, including program-wide services for HPC users. This effort includes all operational support required for DSRC resources and networks to over 5,000 users, data analysis and assessment services for scientific visualization and graphics support; and cybersecurity risk management support for all systems."

From this, it would appear that BAE engineers will be either providing a great deal of hand-holding support for the DSRC sites or may actually even be in charge of operating the supercomputer infrastructure.

• Supercomputer pinpoints exact origin of 'Black Beauty' meteorite from Mars
• What do you mean your exaflop is better than mine?
• US seeks exascale systems 10 times faster than current state-of-the-art computers
• US weather forecasters triple supercomputing oomph with latest machines

As part of its fiscal 2022 investment, the HPCMP acquired two extra supercomputing systems, with corresponding hardware and software maintenance services. These were set to be installed at the Army Combat Capabilities Developmental Command Army Research Laboratory (ARL) in Aberdeen, Maryland, and expected to enter production service in calendar 2023. Adding a total capacity of 10PB, this procurement will increase the HPCMP's aggregate supercomputing capability to 118 petaflops.

Both are based on HPE Cray EX hardware [PDF] running AMD "Genoa" Epyc processors, with one comprising 57,600 compute cores and 135TB of memory, backed by 13PB of storage, 2PB of which is NVMe-based SSDs. The second has 127,488 compute cores and 335TB of memory, plus 64 of AMD's MI250 GPUs, supported by 17PB of storage, 2PB of which is NVMe-based SSDs.

These systems will underpin the Department of Defense's most demanding computational challenges, including for artificial intelligence, data analytics, and machine learning, the HPCMP said.

HPE and AMD would seem to be doing well from supercomputers at the moment; the first official exascale computer, the Frontier system at Oak Ridge National Laboratory, uses Cray EX hardware running Epyc processors, as does Europe's most powerful supercomputer, the sub-exascale LUMI system at the at the IT Center for Science in Finland. ®