Frontier Holds Top Supercomputer Spot, Aurora Closes in on AI Power

See who else made the biannual Top500 world’s supercomputer list

Ben Wodecki, Jr. Editor

May 15, 2024

4 Min Read

Frontier has once again been named the world's most powerful supercomputer but is no longer the only exascale system as HPE’s Aurora looks to snatch its crown.

The biannual Top500 list ranks the world’s supercomputers. The latest list saw Frontier, also known as OLCF-5, hold onto the top spot it’s held since June 2022.

Made up of a total of 8.7 million combined CPU and GPU cores, Frontier was until recently the only exascale supercomputer in the world — until now.

Enter Aurora, the latest supercomputer to break the exascale barrier with a High-Performance Linpack (HPL) score of 1.012 exaflops per second. This means Aurora can perform calculations in one second that would take the entire Earth's population billions of years to complete.

Having finished second overall on the previous Top500 list despite only being half-finished, Aurora was crowned the fastest AI-dedicated supercomputer this time around.

The supercomputer was specifically designed for running high-intensity AI workloads including next-generative generative AI model training.

Aurora achieves 10.6 exaflops per second on 89% of the system when running AI workloads. Frontier’s AI score was just 2.35 exaflops per second.

Housed at the Argonne Leadership Computing Facility in Illinois, Aurora is based on HPE’s Cray EX and uses Intel’s Max Series GPU architecture.

Related:AI's New Frontier: Training Trillion-Parameter Models with Much Fewer GPUs

The supercomputer still isn’t finished either — having been fully installed last Just, the supercomputer is slowly being brought online and optimized and expected to break the 2 exaflop barrier, knocking Frontier off its long-standing perch.

“We are honored to celebrate another significant milestone in exascale with Aurora, which delivers massive compute capabilities to make breakthrough scientific discoveries and help solve the world’s toughest problems,” said Trish Damkroger, HPE’s general manager for high-performance computing and AI infrastructure solutions.

“The Aurora supercomputer surpassing exascale will allow it to pave the road to tomorrow’s discoveries,” said Ogi Brkic, Intel’s general manager of data center AI solutions. “From understanding climate patterns to unraveling the mysteries of the universe, supercomputers serve as a compass guiding us toward solving truly difficult scientific challenges that may improve humanity.”

Eagle, Fugaku: More Non Movers

The top spots for the latest Top500 largely remain the same on the previous list. With Frontier and Aurora again taking up the top two spots, Eagle and Fugaku also retained their previous third and fourth places, respectively.

Related:Fastest AI Supercomputer in the UK Comes Online

Third place Eagle, built by Microsoft, contains 1.12 million computing cores including H100 chips from Nvidia and Xeon Platinum 8480C processors from Intel.

Launched last August, the Microsoft-owned supercomputer is another AI-focused unit, designed to offer access to AI models hosted on Azure cloud.

The Microsoft supercomputer retained its crowns for being the highest-ranked cloud system on the list and the fastest H100-equipped system among the top 10.

Once the world’s most powerful supercomputers, Fukagu again finds itself in the No. 4 spot, having been pushed down the pecking order by Aurora and Eagle last time out.

Fukagu is still the most powerful supercomputer outside of the U.S. Half of the top 10 were from the U.S.

The top 10 largely remained unchanged, with only one new entrant: Alps.

Based in Switzerland’s National Supercomputing Center (CSCS), the Alps supercomputer is another HPE Cray system. It achieved an HPL score of 270 petaflops per second, earning Alps a sixth-place spot, surpassing Leanardo and Summit but not enough to pip LUMI into the top five.

Intel, AMD Remain Supercomputing Hardware Leaders

The latest Top500 list was dominated by supercomputing units housing processors from Intel, IBM and AMD.

Of the top 10, five used Intel Xeons (Aurora, Eagle, Leonardo, MareNostrum 5 ACC and Eos) while two used AMD processors (Frontier, LUMI) and one system used IBM processors (Summit).

Fugaku uses custom hardware from Fujitsu — the A64FX microprocessor, an Arm-based CPU. While the newcomer Alps was the only unit in the top 10 powered by an NVIDIA Grace 72C processor.

Newcomers: Yep1, Baskerville, Dhabi

Some new supercomputing faces joined the Top500 this time out. Among those was Baskerville, a Lenovo-based system from the U.K.’s University of Birmingham.

Yep1, an Nvidia-powered supercomputer owned by SEO software company Ahrefs also appeared on the list, coming in at position 34 — above Dammam-7, the supercomputer of Saudi oil giant Aramco.

The Dhabi Supercomputer also appeared on the latest Top500 list. Dhabi is designed for running AI workloads and is operated by Technology Innovation Institute (TII), the Abu Dhabi-based research lab behind the Falcon AI model.

Also new on the Top500 is Isambard-AI, the U.K.’s newest and fastest supercomputer. Phase one of the project entered the list at number number 128.

Green List: All Newcomers

The gong for the greenest supercomputer in the world went to JEDI, the new Jupiter supercomputer procured by EuroHPC, the EU’s supercomputing initiative.

JEDI, the 190th most powerful supercomputer, is a BullSequana XH3000 machine that runs on Nvidia’s Grace Hopper superchips, boasting a total of 19,584 cores.

Isambard-AI was the second-greenest supercomputer in the world. While the third was Poland-based Cyfronet.

The top three greenest supercomputers were all newcomers on the Top500 list.

Frontier, the overall list leader, ranked 11th, though the Top500 team said the system “deserves an honorable mention when discussing energy efficiency.”

About the Author(s)

Ben Wodecki

Jr. Editor

Ben Wodecki is the Jr. Editor of AI Business, covering a wide range of AI content. Ben joined the team in March 2021 as assistant editor and was promoted to Jr. Editor. He has written for The New Statesman, Intellectual Property Magazine, and The Telegraph India, among others. He holds an MSc in Digital Journalism from Middlesex University.

Keep up with the ever-evolving AI landscape
Unlock exclusive AI content by subscribing to our newsletter!!

You May Also Like