HP Frontier Becomes World's Fastest Supercomputer, Beating Japan's Fugaku

Today, Hewlett Packard Enterprise announced its latest creation, “Frontier.” It is a supercomputer built by HPE for the U.S. Department of Energy’s Oak Ridge National Laboratory (ORNL), which recorded around 1.1 exaflops, making it the first supercomputer globally to break the exascale speed limitation.

As per the Top 500 list of the world’s most powerful supercomputers, it’s the quickest supercomputer globally. HP Frontier is also the number one in the mixed-precision computing category that rates the performance of the computers in AI formats with 6.88 exaflops.

Furthermore, the new supercomputer ranks as the number one choice on the Green500 list of the world’s most energy-efficient supercomputers with 52.23 gigaflops performance per watt, making it 32% more energy-efficient compared to the previous choice.

ORNL’s Frontier can solve problems at 10x faster speed and 8X more complexity

HP Frontier is the most powerful supercomputer globally, capable of delivering extraordinary performance and innovative capabilities; hence it will speed up the discoveries, make breakthroughs and address tough challenges.

The supercomputer, which is more powerful than the next seven choices on the list, will equip the scientists with a better ability to solve problems that are 8X more complex at a 10X faster. Frontier can also reach even higher speed levels with a theoretical peak performance of around two exaflops.

This powerful device is expected to impact important areas such as disease diagnoses, cancer, prognoses, renewable energy, drug discovery, and more.

Justin Hotard, the executive vice president, and GM, HPC & AI, at HPE, said, “Today’s debut of the Frontier exascale supercomputer delivers a breakthrough of speed and performance and will allow us to answer questions we never knew to ask.”

Inside HP Frontier: A technological marvel

HPE created the Frontier with innovative technologies provided through HPE Cray EX supercomputers to improve performance. It also aims to target new AI and machine learning apps to enhance accuracy quicker and more resourcefully.

Here are some of its most significant parts:

74 HPE Cray EX cabinets as every weight around 8,000 lbs or more. Every node comprises one optimized 3rd Gen AMD EPYC™ processor and around four AMD Instinct™ MI250x accelerators that provide 9,408 CPUs and 37,632 GPUs in the complete system.
HPE Slingshot networking cables wroth of 90 miles. It will provide extremely high-performance Ethernet fabric created for the next-gen AI and HPC solutions.
The Cray Clustersor E1000 Storage system lets the Frontier’s Orion storage system provide up to 700 Petabytes of storage capacity, a peak writer speed of 35 terabytes per second, and more than 15 billion random input/output operations each second.
It has refined liquid-cooling capabilities that promote a quiet data center.