Yellowstone (supercomputer)

Yellowstone

Company type	Supercomputer
Industry	High-performance computing
Founded	2012 (2012)
Defunct	December 31, 2017 (2017-12-31)
Fate	Retired and replaced by Cheyenne system
Headquarters	Cheyenne , United States
Products	IBM iDataPlex supercomputer
Owner	National Center for Atmospheric Research

Yellowstone was the first supercomputer installed at the NCAR-Wyoming Supercomputing Center (NWSC) located in Cheyenne, Wyoming, becoming operational by mid-2012. It was retired on December 31, 2017, and replaced by the Cheyenne system.

As a petascale computing platform, Yellowstone facilitated cutting-edge scientific investigations in Earth system science, enabling sophisticated modeling and examination of intricate atmospheric, oceanic, and geophysical phenomena. Its performance supported research into climate dynamics, extreme weather events, geomagnetic disturbances, carbon capture, aviation risk mitigation, and wildfire behavior. Funded through contributions from the National Science Foundation, the State of Wyoming, and the University of Wyoming, and managed by the National Center for Atmospheric Research, Yellowstone’s mission was to improve predictive simulations that aid societal planning and decision-making.

System description

Yellowstone consisted of an IBM iDataPlex cluster delivering 1.5-petaflops of performance, comprising 4,536 dual-socket compute nodes equipped with 9,072 Intel Xeon E5-2670 processors (each with 8 cores), totaling 72,576 cores running at 2.6 GHz. The system featured a combined memory capacity of 145 terabytes. Connectivity between nodes was provided by a Mellanox FDR InfiniBand fat tree network. Its software stack included Red Hat Enterprise Linux tailored for scientific workloads, the LSF Batch Subsystem and Resource Manager, along with the IBM General Parallel File System (GPFS).

Yellowstone operated within NWSC’s unified high-performance computing ecosystem built around a shared file system designed to optimize workflows by supplying common computational, analytical, and visualization environments. This shared storage infrastructure—termed the GLobally Accessible Data Environment (GLADE)—offered 36.4 petabytes of online disk space accessible not only by Yellowstone but also by two data analysis and visualization clusters named Geyser and Caldera. Additionally, it supported data servers for both local and remote users and included an archival system capable of preserving up to 320 petabytes of research datasets. High-bandwidth networks linked this environment to science gateways, data transfer services, remote visualization tools, Extreme Science and Engineering Discovery Environment (XSEDE) hubs, and international collaborators.

By integrating compute resources with file systems, storage solutions, and high-speed networking within a single facility, scientists could perform high-resolution geophysical simulations alongside analysis and visualization tasks without transferring large data volumes between separate infrastructures. This arrangement enhanced research efficiency by minimizing delays associated with data movement and reducing the amount of data researchers needed to download to their home institutions. Annually, Yellowstone delivered over 600 million processor-hours dedicated to Earth system science investigations.

See also

• Supercomputer architecture
• Supercomputer operating systems

External links

• "Yellowstone"
• "Wyoming supercomputer moves in"
• "Supercomputer will help researchers map climate change down to the local level"
• "Yellowstone Super First to Crunch Local Climate Models"
• "NCAR's Data-Centric Supercomputing Environment: Yellowstone"
• "IBM Yellowstone Supercomputer To Study Climate Change"
• "IBM Installs Sandy Bridge EP Supercomputer for NCAR"
• "IBM working on NCAR supercomputer."
• "U.S. weather boffins tap IBM for 1.6 petaflops super"
• "NCAR-Wyoming Supercomputing Center website"
• "NCAR-Wyoming Supercomputing Center Fact Sheet"
• "NCAR-Wyoming Supercomputing Center - UW website"