September 5, 2019
Dell EMC and Intel introduced Frontera, an academic supercomputer that replaces Stampede2 at the University of Texas at Austin’s Texas Advanced Computing Center (TACC). The companies announced plans to build the computer in August 2018 and were funded by a $60 million National Science Foundation grant. According to Intel, Frontera’s peak performance can reach 38.7 quadrillion floating point operations per second (petaflops), making it one of the fastest such computers for modeling, simulation, big data and machine learning.
VentureBeat reports that Frontera’s 38.7 petaflops compares to Stampede2’s peak performance of 18 petaflops. Frontera also “earned the fifth spot on the twice-annual Top500 list with 23.5 petaflops on the LINPACK benchmark, which ranks the world’s most powerful non-distributed computer systems.”
“The Frontera system will provide researchers computational and artificial intelligence capabilities that have not existed before for academic research,” said Intel vice president/general manager of the company’s extreme computing organization Trish Damkroger. “With Intel technology, this new supercomputer opens up new possibilities in science and engineering to advance research, including cosmic understanding, medical cures, and energy needs.”
Frontera is made up of “hundreds of 28-core 2nd Gen Xeon Scalable (Cascade Lake) processors slotted within Dell EMC PowerEdge servers” as well as “Nvidia nodes for single-precision computing,” with chip architecture “based on Intel’s Advanced Vector Extensions 512 (AVX-512).” For the majority of its nodes, Dell EMC supplies water and oil cooling from system integration firm CoolIT and Green Revolution Cooling.
Frontera also uses “Mellanox HDR and HDR-100 interconnects to transmit data at speeds of up to 200Gbps per link between the switches that connect its 8,008 nodes.” Each node is “anticipated to draw around 65 kilowatts of power,” of which TACC supplies one-third from wind power credits, wind power production and solar power.
For storage, DataDirect Networks architected four different environments that total 50+ petabytes “paired with 3 petabytes of NAND flash capable.” Three of them are general-purpose but the fourth offers connectivity of 1.5 terabytes per second.
Frontera is already being used by a variety of scientists, including Rochester Institute of Technology astrophysics professor Manuela Campanelli who is developing a simulation to learn more about the origin of energy bursts from neutron star mergers; University of Texas Austin professor George Biros who is building bio-physical models of brain tumor development to more effectively diagnose and treat gliomas brain tumors; and University of North Carolina at Chapel Hill assistant professor Olexandr Isayev, who is training an AI model that describes the force fields and potential energy of molecules based on their 3D structure.
Frontera, which is expected to operate for five years, will next “involve application-specific accelerators, including quantum simulators and tensor core systems that together deliver a factor of 10 times faster overall compute.” TACC reported that “up to 80 percent of the available hours on Frontera will be accessible through the NSF Petascale Computing Resource Allocation program.”
For more information, including a short video, visit the Intel Newsroom.