- Frontera is the fastest supercomputer on a university campus and the 5th fastest in the world overall
- More powerful computing will help researchers solve large-scale problems faster than ever before
- Frontera will also serve as test-bed for the design of a 10x faster next-generation system
In 2017, the National Science Foundation (NSF) offered a $60 million prize to the US college or university that could propose the best high-performance computing system specifically for academic research. In 2018, the Texas Advanced Computing Center (TACC) at the University of Texas at Austin won that competition. The Frontera supercomputer is the result.
Beginning operations this spring, Frontera is now the fifth fastest supercomputer in the world overall and the fastest on a university campus. With this kind of power, scientists hope to be able to discover new materials for hypersonic flight, better predict major weather events, and explore the mysteries surrounding the origins of the universe.
“Computation is arguably the most critical tool we possess to reach more deeply into the endless frontier of science,” says Dan Stanzione, principal investigator on Frontera and lead researcher on TACC’s previous supercomputers, Stampede1 and Stampede2.
As computers become more powerful, the questions scientists can ask get bigger. Here are just a few topics that Frontera intends to tackle:
Physics and the universe
The Large Hadron Collider (LHC) in Switzerland detects the tiniest particles on earth and helps researchers better understand the fundamental composition of the universe. The computation needed to interpret results and develop explanatory models is supplied by a global network that includes TACC. But LHC is now gearing up for a 10x increase in collisions, and Frontera will help meet the demands of a corresponding leap in data collection and reconstruction complexity.
Sustainable fusion energy
Fusion could be the next big advance in sustainable energy but right now development is held back by an inability to reliably predict and manage disruptions that might damage a fusion reactor. Recently, deep learning has been used to accelerate prediction of disruption events, and scientists are excited to experiment with Frontera’s hybrid design that enables both traditional HPC simulations and machine learning.
Global food production
Computational modeling of crop genomes and their interactions with different environments is helping scientists increase global food supply. Drones collect crop data which then requires massive computing power to integrate with gene sequencing information. By being available to researchers across the US, Frontera will help train the next generation of biologists and life scientists working to sustainably feed the world.
Exceedingly fast and environmentally friendly
As of June 2019, Frontera ranked number five in terms of overall performance on the Top500 list of the fastest supercomputers on the planet. This ranking is based on LINPACK benchmarks, which measure a machine’s computing power through its ability to handle floating point operations.
Although Top500 clocked Frontera’s performance at 23.5 PetaFLOPS, the system has a theoretical peak of 38.7 PetaFLOPS — twice that of Stampede2, which remains in operation through 2022 and beyond. For a human to do what Frontera can, they’d have to perform a single calculation every second for the next billion years.
Additionally, Frontera has a subsystem of graphics processing units (GPUs) that are meant to make solving deep learning problems easier. “For certain application classes that can make effective use of GPUs, the subsystem will provide a cost-efficient path to high-performance,” says Stanzione.
Frontera will need around 5.5 megawatts of energy to function, which is enough to power more than 3,500 homes. To meet these needs, the team behind Frontera’s construction worked toward a reliance on renewable energy.
On-site solar panels, as well as the purchase of energy credits from wind-power generators in West Texas, will help reduce the need for fossil fuels while a 1 million-gallon thermal energy storage tank will help cut back on peak power consumption. In fact, TACC expects a third of Frontera’s power to come from renewable energy sources.
Computing power to the people
Despite Frontera’s many strengths, it won’t be the end of computing innovation. In fact, TACC and the NSF both see this system as a testbed for technologies that will allow them to build a system 10 times more powerful than Frontera within the next five years.
TACC is currently working with Argonne National Laboratory to explore the architecture and performance the Aurora21 machine that should enter production in 2021. This system is meant to be the first exascale computer in the world, and it will greatly alter what scientists can do.
“We’ll be planning for the next generation of computational science and what that means in terms of hardware, architecture, and applications,” Stanzione said. “We’ll start with science drivers – the applications, workflows, and codes that will be used – and use those factors to determine the architecture and balance between storage, networks, and compute needed in the future.”
It’s impossible to know what that future holds, especially in a fast-moving area like HPC. However, all signs from the Frontera team indicate that supercomputing has an exciting road ahead. From reducing environmental impact to ensuring wider access to computational power, we’re about to see some exciting developments.
Read more:
- The 5 fastest supercomputers in the world
- The race to exascale
- US Department of Energy recommits to exascale race
