Frontera brings massive computing power to science’s toughest challenges

The most powerful supercomputer in academia has its next slate of research challenges.

The Texas Advanced Computing Center (TACC) at The University of Texas at Austin and the National Science Foundation (NSF) today announced the allocations of supercomputing resources from the Frontera project for 58 science investigations for 2022-2023. The projects range from high-resolution geospatial mapping to the design of new superconductors. Led by researchers at 45 institutions across 22 states, the allocations began on April 1, 2022.

Frontera—currently the largest academic supercomputer in the world and 13th fastest anywhere—is an NSF-funded system designed to enable discovery at the largest-scales of science and engineering (S&E). Researchers are awarded time on Frontera based on their need for very large-scale computing, and their ability to effectively use a leadership-class supercomputer. Deployed in 2019, Frontera has enabled many 'first in the world' science discoveries, including the first all-atom simulations of the COVID-19 protein spike; the largest, most realistic simulations of supercell thunderstorms; and the largest cosmological simulation performed at very high resolution.

"In addition to providing NSF's Leadership Computing System, the Frontera project also provides unique expertise and experience to the nation to help U.S. S&E researchers be effective in using computation and data analytics at the largest-scale in the scientific discovery process," said Manish Parashar, Director of the NSF Office for Advanced Cyberinfrastructure.  "NSF looks forward to the transformative discoveries that the project will enable with this new cohort of Frontera users."

The allocations awarded this month represent the third cohort of Frontera users selected by the Large Resource Allocation (LRAC) committee—a peer-review panel of computational science experts who convene annually to assess the readiness and appropriateness of projects for time on Frontera. To be considered for an LRAC allocation, researchers need to justify the scientific need for the request and be able to effectively use at least 250,000 node hours  annually, with a maximum award of five million node hours per project.

Among this most recent cohort of awardees is Claire Porter, director of the Polar Geospatial Center at the University of Minnesota, who received an allocation of 2.2 million node hours on Frontera to develop high-resolution earth surface elevation data products for the scientific community.

"The power and size of Frontera allow us to extract high-resolution topography from satellite imagery on a global scale," Porter said. "These fundamental datasets are critical to understanding our changing Earth, from coastal erosion and vegetation change to glacier movement and permafrost collapse. Because of Frontera, we can observe landscapes over time in exquisite detail."

Another awardee, Roxana Margine, associate professor of Physics at Binghamton University, will use her time on Frontera to identify new, superconducting materials.

"Prediction of superconductors in silico made possible with recently developed advanced modeling methods is a transformational shift from the traditional serendipitous experimentation," she said. "We will use the Frontera LRAC allocation to screen a large chemical space of materials to identify practical superconductors with potential to revolutionize energy, transportation, communication, and other emerging technologies."

The LRAC process allows researchers to request resources from a handful of subsystems that are part of the Frontera project, in addition to the 38.7 petaFLOPS main system. These include Longhorn, an IBM system designed for GPU-accelerated workloads; the Frontera GPU system, 90 NVIDIA GPU nodes contained in 4 Green Revolution Cooling tanks; and Ranch, TACC's long-term tape storage system. Frontera users are also frequently among the first to gain access to new hardware prototypes at the center.  Beyond the LRAC awards, Frontera supports several hundred Pathways, Large Scale Community Partnership, Startup and other projects each year. LRAC project will utilize approximately 83% of the 73 million total node hours on the system this year.

The Frontera project team also hosts four Texascale Days events each year—opportunities for researchers to run at least half-scale up to full-scale runs, with several researchers having run jobs with 8,000 nodes—and supports five Frontera Fellows annually, PhD or post-doctoral students who are supported to use Frontera to elevate their research.

"Frontera is a premier scientific resource for the U.S., intended to serve as a research accelerator for the investigators who are selected through the LRAC process," said John West, TACC Director of Strategic Initiatives. "The impact Frontera has had on science so far has been incredible and we look forward to the many breakthroughs that this new cohort will achieve."

More information:
John West et al, Large Scale Science on NSF's Frontera System, Computing in Science & Engineering (2021). DOI: 10.1109/MCSE.2021.3070962

Provided by Texas Advanced Computing Center