TACC Frontera's 2022: Academic supercomputer to run intriguing experiments
Plus: Director reveals 10 million node hours, 50-70 million core hours went into COVID-19 research
The largest academic supercomputer in the world has a busy year ahead of it, with researchers from 45 institutions across 22 states being awarded time for its coming operational run.
Frontera, which resides at the University of Texas at Austin's Texas Advanced Computing Center (TACC), said it has allocated time for 58 experiments through its Large Resource Allocation Committee (LRAC), which handles the largest proposals. To qualify for an LRAC grant, proposals must be able to justify effective use of a minimum of 250,000 node hours and show that they wouldn't be able to do the research otherwise.
Two additional grant types are available for smaller projects as well, but LRAC projects utilize the majority of Frontera's nodes: An estimated 83% of Frontera's 2022-23 workload will be LRAC projects.
Frontera was built in 2019, and premiered on the Top 500 supercomputers list at number five. Since then, the CPU-based, 8,008-node supercomputer has slipped to 13th, but TACC still maintains that it's "the most powerful supercomputer in academia."
In addition to its 38.7 petaflop main system, Frontera also includes the Longhorn GPU-accelerated workload systems, the Frontera GPU system consisting of 90 immersion-cooled Nvidia GPU nodes and its Ranch tape storage system.
Getting TACC back on track
2020 was a bad year for the world, and a particularly busy one for Frontera and the TACC, which was immediately diverted toward COVID-19 research that TACC director Dan Stanzione described as a "huge effort for us."
"Ten million node hours, and approximately 50-70 million core hours went into COVID-19," Stanzione said. TACC and Frontera worked on molecular dynamics, drug discovery, epidemiology, and genomics work related to the coronavirus.
LRAC experiments are some of the biggest, most complicated tasks Frontera is capable of running, and the list of experiments scheduled for the coming year reflect that.
As one example, researchers from Carnegie Mellon University and University of California Riverside plan to use Frontera to develop super-resolution simulations of galaxies and quasars that the team said are necessary to keep up with advancements in telescope technology. Specifically, the team wants to find a way to overcome simulation limitations that force researchers to maximize resolution or volume, but not both.
- Microsoft claims it has spun up a top-five AI supercomputer for its pals at OpenAI – but won't reveal the full specs
- Researchers trick Tesla into massively breaking the speed limit by sticking a 2-inch piece of electrical tape on a sign
- Fujitsu claims 'major technical milestone' in quantum simulation
- Google helps develop AI-driven lab machine to diagnose Parkinson's
Another experiment hopes to gather data on simulated flows of Mach-6 speeds against a surface at a 35-degree angle. This experiment aims to improve understanding of vehicles moving at hypersonic velocities.
Supercomputing advances are being tested as well, with researchers from Binghamton University in New York planning to use Frontera to screen a large batch of potential chemicals for use as superconductors, a la drug discovery.
Many of the remaining experiments [PDF] deal with astrophysics modeling, while others are looking in to supercell thunderstorms, elevation modeling, and quantum computing. ®