Turning megawatts into petaflops and exaflops
Building with posters hanging on the outside that depict several LLNL mission areas

We house some of the world’s most powerful supercomputers

Livermore Computing (LC) offers a world-class high performance computing (HPC) environment with constantly evolving hardware and software resources, state-of-the-art infrastructure, and a wealth of HPC expertise in porting, running, and tuning real-world, large-scale applications. The HPC environment at LLNL is key to the stockpile stewardship mission—helping to ensure the safety, reliability, and effectiveness of the nation’s nuclear weapons. The environment also supports “grand challenge” science and technology initiatives with topics such as NIF target design, climate research, and other national security missions. Grand challenges encourage scientists to push the limits both scientifically and computationally.

LC is home to more than 185 petaflops of compute power serving over 3,000 scientists and engineers, massive shared parallel file systems, powerful data analysis platforms, testbeds for evaluating next-generation hardware and software, advanced visualization resources, and archival storage capable of storing hundreds of petabytes of data. Sierra, a 125-petaflop system focused on stockpile stewardship applications to sustain the nuclear deterrent, was accepted in September 2018. Other major systems include the 20-petaflop Sequoia system; 19-petaflop Lassen system; 5-petaflop Vulcan system; Jade and Quartz systems at 3 petaflops each; and additional large multi-core, multi-socket Linux clusters with various processor types. These systems and their environments are specifically engineered to support the computational workloads and data storage/manipulation specifications needed by our customers.

LLNL’s computing and simulation environment spans several buildings with very large computer rooms that are remotely monitored and controlled. The largest facility provides 48,000 ft2 and 37.5 MW of power for systems and peripherals, and additional power for the associated machine-cooling system—more than one acre of floor space devoted to computing. One of the most modern HPC facilities in the world, a world-class engineering and facilities staff maintains this LEED-certified facility in collaboration with a 24/7 operations staff.

As we plan the next generation of innovative computing and simulation facilities, our focus is on emphasizing function over form, reducing energy intensity, and optimizing the efficiencies of HPC systems. For example, we recently stood up a modular, sustainable new facility for unclassified machines.