Sequoia (supercomputer)

Sequoia (supercomputer). Sequoia was a Blue Gene/Q supercomputer installed at the Lawrence Livermore National Laboratory (LLNL) in California. Funded by the National Nuclear Security Administration (NNSA) under the Advanced Simulation and Computing Program, its primary mission was stockpile stewardship to ensure the safety and reliability of the United States nuclear arsenal without physical testing. The system also enabled groundbreaking research in fields like astrophysics, climate science, and human genome analysis.

Overview

Sequoia represented the third generation of the Blue Gene series from IBM, specifically designed for extreme-scale, power-efficient computing. It was housed within the Terascale Simulation Facility at Lawrence Livermore National Laboratory, a United States Department of Energy (DOE) site managed by Lawrence Livermore National Security, LLC. The system's deployment was a cornerstone of the National Nuclear Security Administration's efforts to maintain the nation's strategic deterrent through advanced computational modeling. Its architecture prioritized massive parallelism and reliability to run complex, multi-physics simulations for weeks at a time.

Hardware and architecture

The system was built using IBM's Blue Gene/Q architecture, comprising 98,304 compute nodes, each containing a 16-core PowerPC A2 processor. In total, Sequoia featured 1,572,864 processor cores and 1.6 petabytes of DDR3 memory. It utilized a highly scalable five-dimensional torus network for low-latency internode communication and incorporated specialized hardware for reliability, such as ECC memory and redundant components. The entire installation occupied 96 racks and required a sophisticated liquid cooling system to manage its thermal load, with a total power consumption of approximately 7.9 megawatts.

Software and applications

Sequoia ran a lightweight, custom Linux kernel known as Compute Node Kernel (CNK). Its primary workload consisted of classified simulation codes like Ares and Kull for stockpile stewardship, which modeled the complex physics of nuclear weapons. Unclassified scientific applications included Cardioid, which simulated the human heart, and the Lattice QCD framework for particle physics. The system also ran large-scale climate models for the Intergovernmental Panel on Climate Change and supported research in turbulence and combustion for the Department of Energy.

Performance and rankings

Upon its unveiling, Sequoia achieved 16.32 petaFLOPS on the LINPACK benchmark, claiming the number one position on the TOP500 list in June 2012 and surpassing the previous leader, Fujitsu's K computer. Its theoretical peak performance was 20.1 petaFLOPS. It also topped the Graph500 list for data-intensive computing and ranked highly on the HPCG benchmark. Sequoia maintained a position within the top ten of the TOP500 for several years, demonstrating sustained capability before being surpassed by newer systems like Tianhe-2 and Titan (supercomputer).

History and development

The development of Sequoia was initiated by IBM and the National Nuclear Security Administration in the late 2000s as part of the Advanced Simulation and Computing Program's push towards exascale computing. It was delivered to Lawrence Livermore National Laboratory in 2011 and entered full operational capability in 2012. The system was formally dedicated in a ceremony attended by officials from the Department of Energy and Congress. After eight years of service, Sequoia was decommissioned in 2020, with its computational missions transitioning to successor systems like Sierra (supercomputer) and the future El Capitan (supercomputer).

Category:Supercomputers Category:Lawrence Livermore National Laboratory Category:IBM supercomputers Category:Blue Gene Category:TOP500