Mike Nartker
NS&D Monitor
7/11/2014
The National Nuclear Security Administration has entered into a contract arrangement with Cray to develop a new supercomputer to aid the NNSA’s stockpile stewardship activities. ‘Trinity,’ as the new supercomputer will be called, is scheduled for delivery beginning mid-2015 and will have “at least eight times greater applications performance” than Cielo, the current NNSA supercomputer located at Los Alamos National Laboratory, according to an NNSA release issued this week. “Trinity will serve the needs of the men and women who play an important role in solving extremely complex calculations that underpin the success of our nation’s Stockpile Stewardship Program.” Bob Meisner, NNSA Advanced Simulation and Computing program director, said in the release. “A very powerful mission-computing system, Trinity begins the transition to new exascale architectures. How well we make that transition has huge impacts on the future of stockpile stewardship.”
Trinity will be located at Los Alamos, and will also be used by the Lawrence Livermore and Sandia national laboratories. “Key drivers” for the new supercomputer, according to NNSA, include application performance improvements and larger memory for running more detailed weapons simulations. “The needs of the mission drive the need for increased memory rather than computing speed alone,” Bill Archer, Los Alamos ASC program director, said in the release. “Trinity will be a very fast machine, but the real key is having enough memory to solve extremely complex calculations for stockpile stewardship.”
Trinity will be the first “Advanced Technology” system for the NNSA’s Advanced Simulation and Computing Program, and will implementing the agency’s new computing strategy that “requires all AT systems to service NNSA mission workload while preparing the ASC applications for transition onto future advanced architectures,” the NNSA release says. It also notes, “Trinity will introduce the ‘Burst Buffer’ concept and ‘Advanced Power Management’ as part of the platform. These technologies will be provided as part of a fully integrated system consisting of compute nodes, memory, high speed interconnect and parallel file system.”
