HPE has been selected by the U.S. Department of Defense to provide seven new supercomputers to the Air Force Research Laboratory at Wright-Patterson Air Force Base, to support research into hypersonics and computational modelling of current and advanced air, naval, and ground weapon systems.  Claire Apthorp spoke to HPE to find out how supercomputers like these can revolutionise military innovation. 

Image: Marinette M arine

Earlier in the year Hewlett Packard Enterprise (HPE) announced that it had been awarded a large $57m contract from the US Department of Defense (DoD) to provide supercomputers. As supercomputing has become an ever bigger part of the toolset of the department’s scientists and engineers innovating around the most complex technological challenges, the thirst for ever more high performance computing (HPC) has only increased.

Under the contract, HPE is 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,” according to a statement.

The HPCMP was originally initiated by the DoD in 1992 in response to a congressional directive to modernise its laboratories' HPC capabilities. The resulting mandate allowed the DoD to consolidate a number of smaller high performance computing departments, each with its own history of supercomputing experience, that had independently evolved within the three services laboratories and test centres.

HPC tools are utilised by the military to solve some of the most complicated and time-consuming problems thrown up by technological development. Researchers can expand their toolkit to solve modern military and security problems using HPC hardware and software. So far HPC has been thrown at a wide variety of problems including assessing technical and management risks, such as performance, time, available resources, cost, and schedule.

According to the DoD, HPCMP supports its objectives through research, development, test, and evaluation. It allows scientists and engineers to focus on science and technology to solve complex defence challenges that could benefit from HPC innovation. To date, under HPCMP, the department has awarded a large variety of contracts both big and small that total an investment in HPC in the hundreds of millions of dollars.

This latest HPCMP contract includes the build and delivery of a total of seven HPE SGI 8600 systems: four for the US Air Force Research Laboratory (AFRL) DoD Supercomputing Resource Center (DSRC) near Dayton, Ohio, and three for the US Navy DSRC Air Force Research Laboratory DSRC at Stennis Space Center, Mississippi.

“Last year saw a significant increase in computing capacity at the DoD’s Supercomputing Resource Centers leading to a growing demand for highly classified HPC resources.”

“The DoD was focused on making several enhancements to its HPC ecosystem under then HPCMP in order to improve utility and effectiveness of its various key research and engineering investments,” a HPE spokesperson told us. “Last year saw a significant increase in computing capacity at the DoD’s Supercomputing Resource Centers leading to a growing demand for highly classified HPC resources.”

While the HPCMP has traditionally supported requirements with smaller, special purpose procurements, the maturity of certain programme initiatives has driven the need for larger machines, with more computational power, that could easily scale to meet demands as different projects in development progress.

In a statement at the time of the award in February, Bill Mannel, vice president and general manager, HPC and AI, HPE said: “In our data-driven world, supercomputing is increasingly becoming a key to stay ahead of competition – this applies to national defence just as to commercial enterprises. The DoD’s continuous investment in supercomputing innovation is a clear testament to this development and an important contribution to US 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.”

“The HPE SGI 8600 was chosen as it represents the best price/performance value to the DoD,” the spokesperson told GDT. “The HPCMP has one of the very best evaluation practices due to their many years of tracking emerging technology, acquiring and then operating world class HPC production systems.” According to officials, when considering purchase, support and operational cost, the HPE SGI 8600 represented the best value in general purpose HPC production systems.

“In our data-driven world, supercomputing is increasingly becoming a key to stay ahead of competition”

HPE says that it also brings the necessary level of specialist and professional services staff, as well as the skills and knowledge in the implementation and support of complex systems. “These are key differentiators that contribute to strategic partnership where clients place the trust in HPE to fulfil their mission requirements,” the spokesperson commented. 

The HPC machines are part of HPE’s SGI range. HPE bought SGI in 2016 at a cost of $275m with a specific eye towards capturing more government business. The HPE SGI 8600 was introduced last year and builds on the proven SGI ICE XA architecture. It leverages a collection of engineering innovations that were made by SGI to enable delivery of the best possible performance for an industry-standard clustered HPC system, according to HPE. These innovations include powerful CPUs, fast memory, the latest GPU technologies, high speed interconnect topologies and tuning of the Message Passing Interface. The HPE SGI 8600 is designed to scale to thousands of nodes as well as being optimised for deep learning through NVIDIA SXM2 compute trays.

All of the above makes the computers highly suited to the tasks they will undertake for the DoD. The AFRL machines will be housed at Wright-Patterson Air Force Base and will support hypersonics research and computational modelling of air, naval, and ground weapon systems and platforms. The US Navy DSRC machines will be used for advanced weapons development and global weather modelling requirements. The combined peak computational capacity of the seven machines is some 14.1 petaflops and more than 24 petabytes of usable storage, leveraging DDN EXAScaler Lustre-based technology.

“Should the DoD decide upon follow-on orders in the near future, HPE will be fully prepared to support the nation’s different missions”

Delivery for the AFRL DSRC and US Navy DSRC systems began in April and were set to continue through May 2018 at the time of writing. The spokesperson said that HPE’s Professional Services and HPE Pointnext teams had been working with the AFRL DSRC and Navy DSRC teams to plan and prepare for the delivery, installation and implementation of the TI-17 systems. 

“Should the DoD decide upon follow-on orders in the near future, HPE will be fully prepared to support the nation’s different missions,” the spokesperson concluded.