The Texas Advanced Computing Center (TACC) at The University of Texas at Austin has announced the launch of Frontera, funded by a $60 million award from the National Science Foundation (NSF). Frontera is the fastest supercomputer at any university and the 5th most powerful system in the world. TACC is also home to Stampede2, the second-fastest supercomputer at any American university.
The system was designed with technologies from Dell, Intel, Mellanox, DataDirect Networks, Nvidia, IBM, CoolIT and Green Revolution Cooling, TACC inaugurated a new era of academic supercomputing with a resource that will help the nation's top researchers explore science at the largest scale and make the next generation of discoveries.
‘Scientific challenges demand computing and data at the largest and most complex scales possible. That's what Frontera is all about,’ said Jim Kurose, assistant director for Computer and Information Science and Engineering at NSF. ‘Frontera's leadership-class computing capability will support the most computationally challenging science applications that U.S. scientists are working on today.’
Frontera combines Dell EMC PowerEdge servers with 8,008 compute nodes, each of which contains two 2nd generation Intel Xeon scalable ‘Cascade Lake’ processors, totalling more than 16,000 processors and nearly half a million cores, connected by a 200 gigabit per second HDR Mellanox InfiniBand high-speed network. The system incorporates innovative flash storage from DataDirect Networks and novel cooling systems from CoolIT, Cooltera, and Green Revolution Cooling (GRC), and employs several emerging technologies at unprecedented scale, including high-powered, high-clock rate versions of the latest Intel Xeon processors, Intel Deep Learning Boost, Intel Optane memory, and several kinds of liquid cooling.
Frontera has been supporting science applications since June and has already enabled more than three dozen teams to conduct research on a range of topics from black hole physics to climate modelling to drug design, employing simulation, data analysis, and artificial intelligence at a scale not previously possible.
First announced in August 2018, Frontera was built in early 2019 and passed formal acceptance recently. The systems earned the number 5 spot on the TOP500 list in June, achieving 23.5 PetaFLOPS (23.5 thousand million million floating-point operations per second) on the high-performance LINPACK benchmark, a measure of the system's computing power.
Olexandr Isayev, a chemist from the University of North Carolina, used Frontera to run more than three million atomic force field calculations in less than 24 hours — a major achievement in high-speed quantum computation. The calculations are part of an effort to train an artificial intelligence system that can predict the likely characteristics of new drug compounds and identify compounds with the ability to target specific cells.
‘It's a great machine, especially for quantum mechanics applications,’ Isayev said. ‘We're really looking forward to running large-scale calculations that were not possible before.’
Ganesh Balasubramanian, an assistant professor of Mechanical Engineering and Mechanics at Lehigh University, has been using Frontera to study the dynamics of organic photovoltaic materials and model manufacturing conditions.
‘The lightning speed at which Frontera performs computations is very beneficial,’ said Balasubramanian, who during the early user period experienced a five-time speed-up in his simulations of solar material manufacturing. ‘Overall, the entire pace of computational research will be increased by the arrival of Frontera.’
Manuela Campanelli, an astrophysicist at the Rochester Institute of Technology, has been using Frontera to perform the longest simulations ever of the merger of neutron stars, including for the 2017 event detected by the Laser Interferometer Gravitational-Wave Observatory (LIGO), the Europe-based Virgo detector, and 70 ground- and space-based observatories.
‘Frontera is an amazing system because it gives us a very large number of computer nodes that we can use to solve very complex problems,’ Campanelli said. ‘These types of resources are unavailable on most university campuses, so we really need to have Frontera in order to be able to do the simulation we do.’
