AUSTIN, Texas — The Texas Advanced Computing Center (TACC) at The University of Texas at Austin has launched a powerful new supercomputer, Stampede3, which is set to enable dynamic open science research projects across the U.S. For over a decade, the Stampede systems have been key components in the National Science Foundation’s scientific supercomputing ecosystem. The latest addition, Stampede3, made possible by a $10 million award from the NSF, will facilitate computational and data-driven science and engineering research and education.
Tommy Minyard, TACC’s director of Advanced Computing Systems and principal investigator of the Stampede3 project said, “During our pre-production period, users experienced capabilities such as an increase in speed-up for scientific applications due to better memory bandwidth per core provided by the Intel Xeon CPU Max processors. And for the first time, we are using a storage system with no spinning disk — we are expecting a significant improvement for users in their I/O performance and reliability.”
TACC continues its partnerships with Dell Technologies and Intel on Stampede3. This nearly 10 petaflop system offers substantial computing capability for diverse scientific applications while also re-purposing hardware from previous NSF investments to support high-throughput users.
Katie Antypas, office director for the NSF’s Office of Advanced Cyberinfrastructure commented on this development: “Stampede3 brings a significant increase in computational and data capabilities to the science and engineering research community. The new high-bandwidth memory node architecture as well as the all-flash filesystem will accelerate a wide range of applications, and I expect it will be in high demand by the user community.”
During its pre-production period alone, more than 450 distinct users ran half a million jobs on Stampede3. It is expected that this system will enable thousands of researchers nationwide to investigate questions requiring advanced computing power across various fields ranging from biology to fluid dynamics.
Among the early users and projects, Stampede3 will support the Galaxy project in biology, which aims to improve its data analysis capabilities. In fluid dynamics, it will aid Texas A&M University's research into understanding turbulent flows at various speeds. In industrial chemistry and materials science, the University of Michigan will use Stampede3 to develop a kinetic Monte Carlo simulation method for studying solute atoms' long-time evolution into solute clusters.
In addition to these features, TACC has added an experimental GPU hardware subsystem for artificial intelligence and machine learning to Stampede3. This move advances the University’s Year of AI initiative and highlights the unique AI data processing capabilities available at UT.
Stampede3 is equipped with a new 4 petaflop capability for high-end simulation, a new graphics processing unit/AI subsystem, reintegrated nodes for higher memory applications, legacy hardware to support throughput computing, VAST Data storage system, and a highly scalable network interconnect from Cornelis Networks. It comprises 2,044 compute nodes with almost 200,000 cores, more than 350 terabytes of RAM, 10 petabytes of new storage, and almost 10 petaflops of peak capability.
Stampede3 is set to serve the open science community from 2024 through 2029.