High-profile US national labs team to build 200 petaflop supercomputers


Three principal US national labs today affirmed they will team-up to build supercomputers that operate about 10 times faster than today’s most powerful high performance computing (HPC) systems.

The project, known as the Collaboration of Oak Ridge, Argonne and Livermore (CORAL) national labs will build 200 peak petaflops (quadrillions of floating point operations per second) systems for each of the labs, at a cost of about $125 million each, in the 2017-2018 timeframe, the group stated.

The collaboration sprung from the fact that the labs will all likely be replacing their current supercomputers – Argonne’s Mira, Livermore’s Sequoia and Oak Ridge’s Titan, at almost the same time.

A joint Request for Proposals for the CORAL procurement was issued Jan. 6 and responses were submitted Feb. 18.  Responses to that request are now being evaluated and the plan is that CORAL partners will select two different vendors and procure a total of three systems, two from one vendor and one from the other. Livermore is leading the procurement process, the group stated.

According to a statement, Livermore’s system, to be called Sierra, will be best suited to support the applications critical to nuclear stockpile stewardship. Oak Ridge and Argonne will employ systems that meet the needs of their DOE Office of Science missions which includes all manner of applications from climate change and energy development to advanced manufacturing and national security.

In the draft of technical requirements of CORAL, written last August, the group wrote:…scientific computation cannot yet do all that we would like. Much of its potential remains untapped-in areas such as materials science, earth science, energy assurance, fundamental science, biology and medicine, engineering design, and national security-because the scientific challenges are too enormous and complex for the computational resources at hand. Many of these challenges have immediate and global importance.

These challenges can be overcome by a revolution in computing that promises real advancement at a greatly accelerated pace.

Planned pre-exascale systems (capable of 1017 floating point operations per second) in the next four years and exascale systems (capable of an exaflop, or 1018 floating point operations per second) by the end of the decade provide an unprecedented opportunity to attack these global challenges through modeling and simulation. Data movement in the scientific codes is becoming a critical bottleneck in their performance. Thus memory hierarchy and its latencies and bandwidths between all its levels are expected to be the most important system characteristic for effective pre-exascale system