Hewlett Packard Enterprise (HPE) today revealed details of its massive $57 million HPC contract with the U.S. Department of Defense (DoD). The deal calls for HPE to provide the DoD High Performance Computing Modernization Program (HPCMP) with supercomputing capability and support services to “accelerate the development and acquisition of advanced national security capabilities.”

The DoD has ordered a total of seven HPE SGI 8600 systems: four for the Air Force Research Laboratory (AFRL) DoD Supercomputing Resource Center (DSRC) near Dayton, Ohio, and three for the Navy DSRC Air Force Research Laboratory DSRC at Stennis Space Center, Mississippi. The AFRL machines will be housed at the Wright-Patterson Air Force Base and will support hypersonics research and computational modeling of air, naval, and ground weapon systems and platforms. The Navy DSRC machines will be used for advanced weapons development and global weather modeling requirements. Combined the systems represent 14.1 petaflops of peak computational capacity and more than 24 petabytes of usable storage, leveraging DDN EXAScaler Lustre-based technology.

“In our data-driven world, supercomputing is increasingly becoming a key to stay ahead of competition – this applies to national defense just as to commercial enterprises” said Bill Mannel, vice president and general manager, HPC and AI, Hewlett Packard Enterprise in a statement. “The DoD’s continuous investment in supercomputing innovation is a clear testament to this development and an important contribution to U.S. national security. HPE has been a strategic partner with the HPCMP for two decades, and we are proud that the DoD now significantly extends this partnership, acknowledging HPE’s sustained leadership in high performance computing.”

When HPE purchased SGI for $275 million in 2016, SGI’s strength in the government vertical was one of the motivating factors. Introduced in 2017 and based on the legacy SGI ICE XA architecture, the HPE SGI 8600 offers petaflops speed for challenging problems in life, earth, and space sciences, to engineering, manufacturing and national security. HPE says the sixth generation server line provides scale and efficiency for complex, largest environments – “up to thousands of nodes with leading power efficiency” achieved with system direct liquid cooling of high-wattage components. Last June, HPE received the Green500 award for the HPE SGI 8600-based TSUBAME cluster, heralded as Japan’s fastest artificial intelligence supercomputer.

All seven DoD systems employ 24-core Intel Xeon Platinum 8168 (Skylake) processors on an Intel Omni-Path Architecture fabric. Four of the seven systems have been outfitted with Nvidia P100 GPUs.

The AFRL DSRC side of the contract consists of:

• A single system of 2,352 Intel Skylake CPUs plus 24 Nvidia Tesla P100 GPUs, 244 terabytes of memory, and 9.2 petabytes of usable storage.
• A single system of 576 Intel Skylake CPUs, 58 terabytes of memory, and 1.6 petabytes of usable storage.
• Two systems, each with 288 Skylake CPUs, 30 terabytes of memory, and 1.0 petabytes of usable storage.

The Navy DSRC will receive:

• Two systems, each consisting of 1,472 Skylake CPUs, 16 Nvidia Tesla P100 GPUs, 154 terabytes of memory, and 5.6 petabytes of usable storage.
• A single system consisting of 296 Skylake CPUs, four Nvidia Tesla P100 GPUs, 32 terabytes of memory, and 1.0 petabytes of usable storage.

The U.S. Army Corps of Engineers, Engineering and Support Center, Huntsville, Alabama, awarded the supercomputing contract, which includes five years of 24/7 system support with on-site system administration and applications support personnel from HPE. The machines are expected to enter production status in the later half of 2018.