Lawrence Livermore National Laboratory



Breakthrough computer simulations that lend new insight or reveal new physics require the most powerful machines and advanced codes available.

simulation

High-order Arbitrary Lagrangian–Eulerian (ALE) framework developed by Livermore scientists can produce accurate and robust simulations of complex multi-material shock hydrodynamics on massively parallel computers.

WCI’s Weapon Simulation and Computing (WSC) program is an integral partner in the Department of Energy/National Nuclear Security Administration’s Advanced Simulation and Computing (ASC) program. By combining state-of-the-art simulation tools with world-class facilities and computers, WSC delivers the capabilities needed to support the nation’s nuclear weapons program as well as the development of inertial confinement fusion and numerous other national security missions. In carrying out these responsibilities, WSC helps advance LLNL’s high-performance computing power.

Livermore Computing Simulation Facility

WSC meets the challenges of stockpile stewardship at the first-class Livermore Computing (LC) simulation facility. LC creates a functional problem-solving environment for high-performance computing by providing the systems, tools, and expertise that make these simulations run smoothly and efficiently. Together, WSC and LC ensure that the Laboratory can fulfill its critical national security research and development responsibilities.

In partnership with LC, WSC develops software packages to improve operations and processing tasks. Our software catalog includes well-known high-performance libraries such as hypre (scalable linear solvers and multigrid methods) and SUNDIALS (nonlinear and differential equation solvers), while a robust tool suite supports a range of needs such as debugging, profiling, tracing, and performance analysis. WSC and LC computer scientists are also active in the open-source community. Other key research areas include extreme computing and hardware architecture.


A grenade modeled with ALE3D, a multi-physics simulation tool, using new embedded-mesh coupling technology. The grenade casing is modeled in a foreground mesh, which is able to fracture into pieces, while the high explosive is modeled in the background mesh. In this simulation, the impact of a fragment comes in from the lower left, detonating the grenade. Such simulation capabilities are fundamental to WSC's mission support and lead to advances in other national security areas.


An ALE3D simulation of laser powder melt, showcasing next-generation materials solutions. With additive manufacturing technology, WSC scientists explore new, versatile materials for replacing critical stockpile components.


 

Next-Generation Computational Capabilities

To simulate the multiscale physics necessary for complete, science-based predictive capability, stockpile stewardship efforts will continue to require enormous computational power. As next-generation computing technology evolves to process and refine data of this magnitude, LLNL aims to be on the forefront of exascale computing—capable of a thousand petaFLOPs, or a quintillion calculations (1018), per second. This threshold means achieving a 25-fold increase in current computational performance.

WSC takes a multifaceted approach to exascale strategy:

  • Application development: improving parallelism, reliability, and other performance capabilities
  • Software technology: expanding and sustaining vertically integrated applications
  • Hardware technology: leveraging research and development of next-generation systems
  • Exascale systems: supporting cost-effective system expansion, site preparation and maintenance, and test environments

Installation of our newest and most powerful supercomputing system, Sierra, in 2017 will advance our progress toward exascale capabilities and includes planning for sustainable exascale-generation facilities. Along with our other ASC and LC resources, WSC will be positioned to meet these complex architectural and processing challenges associated with advancing our national security mission as well as benefit the broader computing community.

Livermore’s 20-petaFLOP Sequoia machine was deployed in 2012. Above is a video tour of Sequoia.