AUSTIN, TX — Stampede3, a powerful new supercomputer that will enable groundbreaking open science research projects in the United States and leverage existing high-performance computing investment funds, is coming to the Texas Advanced Computing Center (TACC) at the University of Texas at Austin.
For more than a decade, the Stampede system (Stampede (2012) and Stampede2 (2017)) has been a flagship product of the National Science Foundation’s (NSF) scientific supercomputing ecosystem. The Stampede system continues to serve important functions for researchers in all areas of science.
Enabled by a $10 million grant from the NSF for computer hardware, Stampede3 will become the latest strategic resource for the country’s open science community when it becomes fully operational in early 2024. This will enable thousands of researchers across the country to investigate problems that require advanced computing power.
“We will continue our long-standing partnership with Dell and Intel to leverage NSF’s investment in Stampede2 for this new scientific resource that will make Intel’s graphics processing units widely available to the NSF open science community, using state-of-the-art processors with high-bandwidth memory,” said Dan Stanzione, Executive Director of TACC.
Stampede3 offers:
- New 4 petaflops capability for high-end simulation: 560 new Intel® Xeon® CPU Max series processors with high-bandwidth memory-capable nodes add nearly 63,000 cores for the largest and most performance-intensive computing jobs.
- The new graphics processing unit/AI subsystem includes 10 Dell PowerEdge XE9640 servers with 40 new Intel® Data Center GPU Max series processors for AI/ML and other GPU-accelerated applications.
- 224 3 social reintegrationrd Gen Intel Xeon Scalable Processor Node for high memory applications (added to Stampede2 in 2021).
- Legacy Hardware Supporting Throughput Computing — Over 1,000 Existing Stampede2 2nd 2nd Generation Intel Xeon Scalable Processor nodes will be included in new systems to support high-throughput computing, interactive workloads, and other small workloads.
- New Omni-Path Fabric 400 Gb/s technology delivers scalable performance through a network interconnect with 24 TB/s of backplane bandwidth, enabling low latency, superior application scalability, and high connectivity to I/O subsystems.
- 1,858 compute nodes with over 140,000 cores, over 330 terabytes of RAM, 13 petabytes of new storage, and nearly 10 petaflops of peak capacity.
“Stampede3 provides user community access to CPU nodes with high-bandwidth memory to accelerate application performance,” said Katie Antypas, office director of Advanced Cyber Infrastructure Division at NSF. “Furthermore, the transition from Stampede2 to Her Stampede3 is transparent to users, easing the transition to the new system. We are confident that this will become a popular platform for the broader scientific and engineering community.”
The Stampede3 project, like the previous Stampede system, encompasses more than just technology. Stampede3 also includes first-class operations, user support and training, education, outreach, documentation, data management, visualization, analytics-driven application support, and research collaboration.
The new system will be delivered in the fall of 2023 and is expected to be fully operational in early 2024 with no service interruptions from Stampede2 to Stampede3. Serving the open science community from 2024 to 2029.
“The best way to argue the need and future of Stampede3 science and engineering is to look at the current success of Stampede2 nearing the end of its production life,” said Stanzione. “Individual jobs on Stampede2 have been successful even at extreme scales of 500,000 cores and come from nearly every scientific field.”
Since its introduction, more than 11,000 users have worked on more than 3,000 funded projects and executed more than 11 million simulation and data analysis jobs on Stampede2 during its operational life so far.
