Los Alamos Brings Complementary Strengths to the Exascale Computing Project

Founded during World War II as part of the Manhattan Project to develop nuclear weapons, Los Alamos National Laboratory (LANL) in New Mexico focuses on national security and the nation’s nuclear stockpile as well as the pursuit of research in science, technology, and engineering across multiple disciplines. LANL led the creation and use of applied computing in its earliest days and continues as a world leader in the field today, currently operating the Trinity supercomputer in support of the US Department of Energy (DOE) National Nuclear Security Administration (NNSA) Stockpile Stewardship Program.

LANL’s experience involves both the NNSA Advanced Simulation and Computing (ASC) program and the DOE Office of Science (DOE-SC) Advanced Scientific Computing Research (ASCR) program.

“The ASC activities are focused on the mission of delivering weapons code that predicts performance, whereas ASCR research activities span applied math, computer science, and co-design,” said John Sarrao, associate laboratory director for theory, simulation, and computation at LANL. “We’re at our best when we combine ASCR’s horizontal breadth and ASC’s vertical integration. That blend is beneficial in the pursuit of exascale, the next frontier in computing. Exascale is about delivering impact for mission. It is also about the merger and synergy of traditional numerical simulation, data science, and machine learning.”

NNSA has various drivers that are pointing LANL toward exascale. “We know we need to improve predictive modeling of materials as they age, as we change manufacturing processes, or as we put new types of materials into the system,” said Bill Archer, ASC program director at LANL. “I think everybody in the research community is interested in materials. It’s a hot topic across industry, academia, and the national labs.”

LANL is leading several efforts in DOE’s Exascale Computing Project (ECP) that bring together DOE-SC and NNSA’s Advanced Technology Development and Mitigation subprogram with the goal of developing new applications that enable new material models and other research areas that are designed to function with exascale machines. The lab is also working to advance programming models, compilers, and various debugging tools through the ECP.

“Because the ECP is such a big undertaking, communication is the toughest challenge,” Archer said. “We have to effectively exchange information to avoid duplication of effort and to find critical gaps in what we’re developing. The ECP leadership structure has been designed to address that concern by facilitating communication among the labs at multiple levels.”

Each of the six ECP partner labs is represented on the project’s Leadership Team.

LANL’s Stephen Lee, who has 30 years of experience in computational physics, code development, and high-performance computing application project leadership, is ECP’s deputy project director.

“Few things are done at a national laboratory that don’t involve other laboratories, vendors, universities, or the private sector in some way,” Lee said.  As an example, he cited the multiparty interactions involved in the Consortium for the Advanced Simulation of Light Water Reactors, of which LANL was a founding partner.

“The weapons program is collaborative,” Lee said. “We are responsible for the nuclear stockpile, but we work with Lawrence Livermore and Sandia National Laboratories on that activity in different ways, for different reasons. The same is true for the ECP. All of the partner labs bring different perspectives and different ways of thinking about and solving problems. That’s diversity in action and exactly what’s needed to solve the most difficult scientific and technological challenges that society faces.”