The Race for Exascale: A Recent History of the World’s Fastest Supercomputers

Nations jostle for bragging rights to have the world's fastest supercomputer.

6 Min Read
A photo of the utility-scale IBM Quantum System One, which was unveiled in April 2024.
The utility-scale IBM Quantum System One was unveiled in April 2024.IBM

For many years, exascale was the supercomputing industry’s Holy Grail. Since the first petascale (1015 FLOPS) computer entered operation in 2008, the US, China, Japan, and Europe raced to build exascale supercomputers. 

Over the past 15 years, various supercomputers have jostled for the top spot, as supercomputer speeds have risen from single petaflop speeds to two exaflops. In recent years, US investment has given its latest supercomputers the title of fastest supercomputer. 

Data Center Knowledge has been there every step of the way, tracking changes in the Top 500 list, investment announcements, and the construction of new supercomputers, along with other sector-related news.  

Check out our highlights of the past 15 years of supercomputing: 

2024: IBM Quantum Computer Goes Online 

In April 2024, Rensselaer Polytechnic Institute (RPI) and IBM unveiled the world’s first-ever IBM quantum computer on a university campus, six months after beginning installation. Faculty, researchers, students and collaborators will be able to use the system to advance quantum computing research.  

Read more: IBM Quantum Computer Goes Online at University 

2024: Accelerating Quantum Supercomputing Research 

In January 2024, Quantinuum said it would install its H1 Series ion-trap quantum computing technology on-site at the campus of the Japanese national research and high-performance computing (HPC) center Riken. The quantum-supercomputer hybrid platform will support a research program that Softbank, the University of Tokyo, and Osaka University are undertaking to develop tools and applications to integrate quantum computers with classical supercomputers effectively. 

Related:Assessing the State of Quantum Data Centers: Promises vs. Reality

Read more: Riken Accelerates Quantum Hybrid Research With Quantinuum Partnership 

2024: Training Trillion-Parameter Models 

Scientists from the famed Oak Ridge National Laboratory trained a one trillion parameter model using just a few thousand GPUs in their Frontier supercomputer. They used just 3,072 GPUs to train the giant large language model out of 37,888 AMD GPUs housed in Frontier. 

Read more: AI’s New Frontier: Training Trillion-Parameter Models with Much Fewer GPUs 

2024: Europe Launches Exascale Supercomputer Bid 

Europe unveiled plans to launch its first exascale computer in 2024 to rival the most powerful supercomputers in the world. Jupiter will be housed at the Jülich Supercomputing Centre in Germany and is expected to be capable of one exaflop, or one billion-billion calculations per second. The cost of building and operating the supercomputer for six years is projected at €500 million ($545 million). 

Related:A History of Microsoft Azure Outages

Read more: Europe to Launch its First Exascale Supercomputer in 2024 

2023: Breaking the 1,000-Qubit Barrier 

In November 2023, Quantum startup Atom Computing created a 1,225-site atomic array in its next-generation quantum computing platform, making it the first quantum computer to exceed 1,000 qubits. 

Read more: Atom Computing Quantum Computer Breaks 1,000-Qubit Barrier 

2023: LLNL’s ‘El Capitan’ Exascale Supercomputer 

The 2-exaflop El Capitan supercomputer was poised to become the world’s fastest supercomputer when it began construction in 2023. The supercomputer is powered by Hewlett Packard Enterprise’s Cray hardware and the latest generation AMD processors. 

Read more: LLNL's El Capitan Could Be World's Fastest Supercomputer in 2024 

2022: Oak Ridge’s ‘Frontier’ Supercomputer 

The HPE-made Frontier supercomputer was named the world’s fastest supercomputer in 2022, putting the US back in first place. Frontier aimed for a theoretical 2 exaflops and was based on the HPE Cray EX235a architecture, equipped with AMD EPYC 64C 2GHz processors with more than 8.7 million cores. 

Read more: US Retakes Lead With World’s Fastest Supercomputer 

2020-2022: Fujitsu’s ‘Fugaku’ Supercomputer 

In 2020, Fujitsu’s ARM-powered Fugaku supercomputer took the lead as the world’s fastest supercomputer, beating out US and Chinese rivals. The system, built by Fujitsu for Japan’s RIKEN Center for Computational Science remained in the top spot until 2022. 

2019: Argonne’s ‘Aurora’ Supercomputer 

In 2019, Intel and the US Department of Energy announced plans to build the country’s first exascale supercomputer at Argonne National Laboratory by 2021. The supercomputer would use Intel Xeon processors and Cray hardware and become the first high-performance computing system in the US that reaches exaflop speed. 

Read more: Intel and US Energy Department Plan Exascale Aurora Supercomputer by 2021 

2018: HPE’s Arm-Based ‘Astra’ Supercomputer 

In 2018, HPE, the US Department of Energy, and Sandia National Laboratories announced a partnership to build Astra, at the time the world’s largest ARM-based supercomputer. 

Read more: HPE and DOE Partner to Build Largest ARM-Based Supercomputer 

2018: Oak Ridge’s ‘Summit’ Supercomputer 

In 2018, the US unveiled Summit, a supercomputer designed by IBM and Nvidia. Summit is powered by the former’s Power 9 processors and the latter’s Tensor Core GPUs. Summit features a unique architecture combining HPC and AI computing capabilities. 

Read more: IBM, Nvidia Build ‘World’s Fastest Supercomputer’ for US Government 

2017: DoE Reinvests in Exascale 

In 2017, the US Department of Energy awarded $258 million in funding to six US tech companies in an effort to build the country’s first exascale supercomputer

Read more: Energy Department Awards $258 Million to Develop Exascale Supercomputers 

2016-2018: Wuxi’s ‘Sunway TaihuLight’ Supercomputer 

Wuxi’s Sunway TaihuLight supercomputer was the top performer on the Top 500 Supercomputers list in 2016. Sunway uses simple CPUs without memory caches but with its 10.6 million-plus processor cores divided into clusters of 65. 

Read more: Nvidia Makes Its Mark in Top 500 Supercomputers List for Power Efficiency 

2013-2015: China’s ‘MilkyWay-2’ (Tianhe-2) Supercomputer 

Between 2013 and 2015, China’s Milkyway-2 (Tianhe-2) had a long run as the world’s fastest computer, ranking in the top spot five times. Clocking in at 33.86 petaflop/s Milkyway-2 has almost doubled the performance of Titan, the previous fastest supercomputer. 

2013: A New Supercomputer Performance Benchmark 

In 2013, it was reported that HPC thought leaders were looking to develop a new benchmark for supercomputing, reflecting the need to better correlate to computation and data access patterns found in many applications today.

Read more: Looking Beyond Linpack: New Supercomputing Benchmark in the Works 

2012: DoE Invests in Exascale Supercomputing 

In 2012, the US Department of Energy announced funding for exascale supercomputing, awarding $62 million in contracts to advance "extreme scale" supercomputing. Meanwhile, Lawrence Berkeley National Laboratory broke ground on a new facility to house its supercomputer vision. 

Read more: DOE Announces Funding for ‘Extreme Scale’ Supercomputers 

2012: Oak Ridge’s ‘Titan’ Supercomputer 

In 2012, Oak Ridge’s Titan supercomputer became the new fastest supercomputer, unseating IBM’s Sequoia. Titan has 20 petaflops of computing power, beating Sequoia’s 16.3 petaflops. 

Read more: Titan is New Champion of Top500 Supercomputing List 

2012: LLNL’s ‘Sequoia’ Supercomputer 

LLNL’s Sequoia supercomputer became the fastest supercomputer in mid-2012, putting the US at pole position on the Top 500 list for the first time since 2009. 

Read more: Sequoia Supercomputer Puts US Back Atop the Top 500 

2011: Fujitsu’s ‘K’ Supercomputer 

In 2011, Fujitsu’s K supercomputer took the position of fastest supercomputer in the world, reaching a benchmark of 10 petaflops. 

Read more: New Top 500 Champ: The K Supercomputer 

2010: China’s ‘Tianhe-1A’ Supercomputer 

China’s Tianhe-1A supercomputer achieved 2.507 petaflops in 2010 and took the top spot on the Top 500 list, unseating the US Jaguar supercomputer. 

Read more: China’s Tianhe-1A Achieves 2.507 Petaflops 

2009: Oak Ridge’s ‘Jaguar’ Supercomputer 

In 2009, Oak Ridge’s Jaguar supercomputer unseated Los Alamos’ RoadRunner supercomputer, claiming pole position on the Top 500 list. Jaguar posted speeds of 1.75 petaflop/s, beating RoadRunner’s 1.04 petaflops/s. 

Read more: Jaguar Bumps RoadRunner in Top 500 

Subscribe to the Data Center Knowledge Newsletter
Get analysis and expert insight on the latest in data center business and technology delivered to your inbox daily.

You May Also Like