Dawn (supercomputer)
| Company type | Supercomputer |
|---|---|
| Industry | High-performance computing |
| Founded | 2011 |
| Defunct | 2020 |
| Fate | Decommissioned |
| Headquarters | Lawrence Livermore National Laboratory , United States |
| Products | Blue Gene/Q supercomputer |
IBM Sequoia was a petascale Blue Gene/Q supercomputer engineered by IBM for the National Nuclear Security Administration as part of the Advanced Simulation and Computing Program (ASC). Delivered to the Lawrence Livermore National Laboratory (LLNL) in 2011, it reached full operational status by June 2012 and was decommissioned in 2020. Its final position on the TOP500 list was #22 in November 2019.
On June 14, 2012, Sequoia attained the title of the world’s fastest supercomputer, outperforming the K computer with a LINPACK benchmark of 17.17 petaflops—63% higher than K’s 10.51 petaflops—and incorporating 123% more cores (1,572,864 compared to 705,024). Additionally, it demonstrated superior energy efficiency, consuming 7.9 MW versus K’s 12.6 MW.
By November 2017, Sequoia had dropped to sixth place on the TOP500 ranking; earlier, on June 17, 2013, it was positioned third behind Tianhe-2 and Titan.
Sequoia enabled groundbreaking scientific computations, becoming the first system to sustain performance exceeding 10 petaflops. The cosmological simulation platform HACC achieved close to 14 petaflops on a benchmark involving 3.6 trillion particles, while the Cardioid cardiac electrophysiology model reached nearly 12 petaflops simulating the human heart in near real-time.
The supercomputer operated on Linux: over 98,000 nodes ran the CNK operating system, with an additional 768 I/O nodes utilizing Red Hat Enterprise Linux. These were interconnected through the Lustre parallel filesystem.
Dawn prototype
IBM created the "Dawn" prototype delivering 500 teraflops based on the Blue Gene/P architecture to assess Sequoia’s capabilities. This system was delivered in April 2009 and ranked ninth on the Top500 list by June of that year.
Purpose
Sequoia’s main role was to support nuclear weapons simulation, succeeding the Blue Gene/L and ASC Purple supercomputers at Lawrence Livermore National Laboratory. It also contributed to scientific investigations in fields such as astronomy, energy, lattice QCD, genomics, and studies related to climate change.
Node architecture
As a Blue Gene/Q platform, Sequoia consisted of 96 racks containing a total of 98,304 compute nodes (with 1,024 nodes per rack). Each node housed a 16-core A2 processor alongside 16 GB of DDR3 memory. This configuration resulted in a combined total of 1,572,864 processor cores and approximately 1.5 PiB of memory distributed across the system. Nodes were linked through a five-dimensional torus network topology.
Job scheduler
The computing resources of Sequoia were managed by Lawrence Livermore National Laboratory using the SLURM scheduler—also employed by Dawn and China’s Tianhe-IA—to optimize workload distribution efficiently.
Filesystem
LLNL implemented the Lustre parallel filesystem and adapted ZFS for Linux to serve as Lustre object storage devices (OSDs), enhancing performance and adding advanced features. In September 2011, NetApp was contracted by the DoE to supply 55 PB of storage capacity supporting Sequoia’s filesystem infrastructure.
Power usage
The entire system consumed roughly 7.8 MW of electrical power, delivering remarkable energy efficiency at approximately 2068 Mflops/watt—making it nearly six times more efficient than Dawn and over two-and-a-half times more efficient than the June 2011 TOP500 leader, the K computer.
Application
In January 2013, Sequoia achieved a milestone as the first supercomputer to operate more than one million processing cores concurrently within a single application. The Center for Turbulence Research (CTR) at Stanford Engineering utilized Sequoia’s capabilities to address intricate fluid dynamics problems, focusing on the prediction of noise produced by supersonic jet engines.