17 Fastest Supercomputers In The World | In 2024

For most of us, a computer probably seems fast enough if it can run 8K videos or the latest version of Far Cry at 60 fps without slowing down. However, there are many complicated tasks that require billions of calculations per second – something a desktop with an Intel i9 or Ryzen 9 processor can’t do.

Now imagine a machine capable of processing data at speeds that boggle the mind, solving complex problems that were once deemed unsolvable, and simulating the very fabric of our universe itself.

That’s what supercomputers are. 

They are not just technological marvels; they are the beating heart of scientific research, the architects of cutting-edge innovation, and the guardians of national security. 

These machines have propelled us into the era of exascale computing, where performance is not measured in teraflops or petaflops but in exaflops – a quintillion floating-point operations per second.

Today’s supercomputers are meticulously designed to accommodate artificial intelligence (AI) workloads. They not only enhance tasks like weather forecasting, climate research, physical simulations, and oil and gas exploration but also play a crucial role in advancing scientific frontiers. 

For example, they facilitate the discovery of more durable construction materials and enable in-depth investigations of human proteins and cellular systems, pushing human knowledge to unprecedented levels of detail. 

Did you know?

The first supercomputer — Livermore Atomic Research Computer — was built for the US Navy Research and Development Centre in 1960.

To show you how far we have come since then, we have curated a detailed list of the fastest supercomputers in the world. They are all non-distributed computer systems running on Linux.

10. Tianhe-2A

Tianhe-2, Precedessor of Tianhe 2A

Speed: 61.4 petaFLOPS
Cores: 4,981,760

Vendor: NUDT
Location: National Supercomputing Center in Guangzhou, China

With more than 16,000 computer nodes, Tianhe-2A represents the world’s largest installation of Intel Ivy Bridge and Xeon Phi processors. While each node has 88 gigabytes of memory, the total memory (CPU+coprocessor) is 1,375 tebibyte.

Tianhe-2A was the fastest supercomputer when it was deployed in 2013. It was developed by a team of over 1,300 engineers and scientists. 

China spent 2.4 billion yuan (US$390 million) on building this supercomputer. It is now mostly used in the analysis of massive volumes of data, complex simulations, and government security applications.

9. Selene

Speed: 63.46 petaFLOPS
Cores: 555,520

Vendor: Nvidia
Location: Argonne National Laboratory, United States

Selene is built on the Nvidia DGX system architecture, which includes substantial hardware resources: 

  • AMD CPUs,
  • Nvidia A100 GPUs, and 
  • Mellanox HDDR networking. 

More specifically, its computational nodes contain 1080 AMD Epyc CPUs and 4320 Nvidia A100 GPUs. 

Selene’s extraordinary computing power shines when it trains BERT, a natural language processor, in under 16 seconds. This task usually takes much longer on smaller systems, around 20 minutes. 

The supercomputer uses NVIDIA DGX SuperPOD to maintain energy efficiency. It has demonstrated an exceptional power efficiency of 26.2 gigaflops/watt during its performance run. 

It has been employed in numerous scientific applications, particularly those involving protein docking, quantum chemistry simulations, and biomedical research. 

8. Perlmutter 

Speed: 70.87 petaFLOPS
Cores: 761,856

Vendor: Cray
Location: National Energy Research Scientific Computing Center, United States

Named after Nobel Prize winner Saul Perlmutter, this supercomputer is a heterogeneous system containing 3,072 CPU-only and 1,792 GPU-accelerated nodes. It features AMD “Milan” EPYC CPUs and NVIDIA A100 GPUs. 

Perlmutter is based on the HPE Cray Shasta architecture, which includes a state-of-the-art HPE Slingshot high-speed network and a large 35-petabyte FLASH scratch file system. This high-speed storage system can move data at a rate of over 5 terabytes per second, making it one of the fastest storage solutions of its kind. 

Perlmutter also features an innovative interconnect system called Slingshot. It is built for data-centric computing and offers several advantages, including advanced adaptive routing, congestion control, and quality of service features. 

Together, these enhancements improve system performance, utilization, and scalability for many supercomputing and AI workloads. 

To date, this supercomputer has played a crucial role in materials science simulations, astrophysical simulations, quantum chemistry simulations, climate modeling simulations, biomedical research, and energy-related research. 

7. Sunway TaihuLight

Speed: 93 petaFLOPS
Cores: 10,649,600

Vendor: NRCPC
Location: National Supercomputing Center in Wuxi, China

The computing power of TaihuLight comes from a homegrown several-core SW26010 CPU that includes both computing processing elements and management processing elements.

A single SW26010 provides a peak performance of more than 3 teraFLOPS, thanks to its 260 processing elements (integrated into one CPU). Each computing processing element has a scratchpad memory that serves as a user-controlled cache, significantly reducing the memory bottleneck in most applications.

The cost to develop this supercomputer was approximately 1.8 billion Chinese Yuan at the time of its construction, which is roughly equivalent to around 274 million U.S. dollars. This includes hardware costs, infrastructure costs, and research and development expenses. 

China committed this significant investment to advance its high-performance computing technology and maintain a leadership position in the field of supercomputing. 

TaihuLight has showcased its versatility by facilitating innovative research across multiple domains. In addition to life sciences and pharmaceutical research, TaihuLight has been used to simulate the universe with 10 trillion digital particles.

However, China is trying to achieve a lot more: the country has already stated its goal to be the leader in AI by 2030.

6. Sierra

Speed: 94.6 petaFLOPS
Cores: 1,572,480

Vendor: IBM
Location: Lawrence Livermore National Laboratory, United States

Sierra offers up to 6 times the sustained performance and 7 times the workload performance of its predecessor, Sequoia. It combines two types of processor chips: IBM’s Power 9 processors and NVIDIA’s Volta GPUs to generate a peak performance of 125 petaflops.

More specifically, Sierra is made up of 4,474 individual units called nodes. Out of these, 4,284 are used for actual computing tasks. Each node has 

  • 256GB of RAM, 
  • 44 IBM POWER9 cores, and
  • 4 Nvidia Tesla V100 GPUs, each with 16GB of memory called VRAM.

In total, this supercomputer has 8,948 CPU cores, 17,896 GPU processors, 1.14 petabytes of RAM, and 286 terabytes of VRAM. 

Sierra is specifically designed for assessing the performance of nuclear weapon systems. It is primarily used for applications in stockpile stewardship — the US program for reliably testing and maintaining nuclear weapons without any nuclear testing.

5. Summit


Speed: 148.6 petaFLOPS
Cores: 2,414,592

Vendor: IBM
Location: Oak Ridge National Laboratory, United States

Summit can deliver 200 petaFLOPS at peak. This is equivalent to 200 quadrillion floating-point operations per second. It is also one of the world’s most energy-efficient supercomputers, with a recorded power efficiency of 14.66 gigaFLOPS per watt.

Summit’s 4,600+ servers, which take up the size of two basketball courts, house more than 9,200 IBM Power9 processors and over 27,600 NVIDIA Tesla V100 GPUs. The system is connected by 185 miles of fiber optic cable. 

In 2018, Summit became the first supercomputer to break the exascale barrier. While analyzing genomic data, it achieved a peak throughput of 1.88 exaops.

Besides complex modeling and simulation, Summit also serves as a formidable artificial intelligence and deep learning system. Its exceptional computational capabilities enable it to analyze vast datasets and automate essential steps in the discovery process.

In 2021, a research team (led by the Department of Energy’s Oak Ridge National Laboratory and the Georgia Institute of Technology) used Summit to harness the power of deep learning methods to predict the structures and functions of thousands of proteins with unknown roles.

Summit, being one of the world’s most powerful supercomputers, provided the computational horsepower needed to analyze massive datasets and conduct complex deep learning computations.

It is also involved in climate modeling, materials science, advanced manufacturing, genomics, drug discovery, and medical research. 

4. Leonardo

An overview of Leonardo’s architecture

Speed: 238.70 petaFLOPS
Cores: 1,824,768

Vendor: Cineca
Location: Bologna Technopole, Europe

Leonardo boasts an exceptional computational capacity, capable of achieving a peak performance of nearly 304 petaflops. It is equipped with over 100 petabytes of storage capacity, allowing researchers to manage and analyze massive datasets efficiently. 

Leonardo features an advanced data network solution with enhanced routing support. This network facilitates high bisection bandwidth through non-minimal routing. It even optimizes intra-group and inter-group routing, providing low hop count and high network throughput.  

The system also has 16 additional nodes for visualization tasks. These nodes feature 6.4 terabytes of NVMe disks and two Nvidia Quadro RTX8000 48GB GPUs. This configuration enables fast data access and bandwidth-demanding visualizations and is particularly useful for 3D applications. 

This top-tier supercomputer system is built to maximize performance and versatility, making it a valuable resource for addressing complex workflows, spanning deep learning, high-throughput processing, and advanced visualization applications. 

As a part of the European High-Performance Computing Joint Undertaking, Leonardo reflects a collaborative effort among European countries to advance high-performance computing capabilities. It receives €120 million from the European Union and an additional €120 million from the Italian Ministry of Education, University, and Research.  

3. LUMI

Speed: 309.10 petaFLOPS
Cores: 2,220,288

Vendor: Hewlett Packard
Location: RIKEN Center for Computational Science, Japan

With a theoretical peak performance of 537 petaflops, LUMI is the fastest supercomputer in Europe and the third fastest in the world. It leverages a substantial number of nodes equipped with GPU accelerators to achieve this impressive speed and computational power. 

LUMI’s architecture includes multiple particles to cater to a range of scientific simulations and data-intensive tasks. It features a GPU partition with 2,560 nodes, each equipped with a 64-core AMD Trento CPU and four AMD MI250X GPUs. Each GPU node has four 200 Gbit/s network interconnect cards (or 800 Gbit/s injection bandwidth). 

It also features a CPU-only partition called LUMI-C. This comprises nodes powered by 64-core 3rd-generation AMD EPYC™ CPUs and varying memory configurations (from 256 GB to 1 TB). In total, there are 1,536 dual-socket CPU nodes, offering a flexible computing environment for various scientific workloads.

As for power usage, LUMI operates on 100% hydroelectric energy. The waste heat generated by this supercomputer is harnessed and repurposed to heat nearby buildings in the city of Kajaani. 

This dual-purpose use of energy waste heat recovery makes LUMI one of the most environmentally efficient supercomputers worldwide. 

The project is a result of a collaborative effort between the EuroHPC Joint Undertaking and the LUMI Consortium. Its total cost of ownership is expected to be about €144.5 million.  

2. Fugaku

Speed: 442 petaFLOPS
Cores: 7,630,848

Vendor: Fujitsu
Location: RIKEN Center for Computational Science, Japan

With a theoretical peak performance of 537 petaFLOPs, Fugaku is the world’s second-fastest supercomputer. It is also the first top-ranked supercomputer to be powered by ARM processors.

As per the HPCG benchmark, Fugaku’s performance surpasses the combined performance of the next top four supercomputers in the world.

It’s a great achievement for the Japanese government, but designing such a powerful system didn’t come cheap. Since 2014, the government has spent about $1 billion on the project’s R&D, acquisitions, and application development.

Fugaku runs on two operating systems side by side: Linux and a ‘light-weight multi-kernel OS’ called IHK/McKernel. Linux handles Portable Operating System Interface (POSIX) compatible services, while McKernel runs high-performance simulations. 

It is designed to address high-priority social and scientific problems, such as weather forecasting, clean energy development, drug discovery, personalized medicine, and exploring the laws of quantum mechanics.

Fugaku gained fame for its role in COVID-19 research. It was used for simulations related to the virus’s spread, drug discovery efforts, and the study of potential treatments.

It has also been used to develop an AI model that can accurately predict tsunami flooding in coastal areas in near real-time. This achievement showcases the versatility of Fugaku — not only can it carry out scientific simulations but also address critical challenges such as disaster mitigation and coastal safety.  

1. Frontier 

Speed: 1,194 petaFLOPS
Cores: 8,699,904

Vendor: Hewlett Packard
Location: Oak Ridge Leadership Computing Facility, United States

Frontier is the world’s fastest supercomputer, with a computing performance of 1.19 quintillion floating-point operations per second.

Its mixed-precision computing performance is measured at approximately 6.88 exaflops (or 6.88 quintillion operations per sector), which makes it a powerhouse for AI and machine learning workloads. 

As for hardware configuration, Frontier is equipped with AMD CPUs and GPUs to deliver incredible computational capabilities. It has 74 HPE Cray EX supercomputer cabinets, specially designed to support next-generation supercomputing operation and scalability. 

Every node within Frontier contains one optimized EPYC™ processor and four AMD Instinct™ accelerators. This brings the total to over 9,400 CPUs and 37,000 GPUs.

It is also one of the world’s most energy-efficient supercomputers, featuring a liquid cooling system that not only promotes a quieter datacenter but delivers an exceptional energy efficiency rating of 62.684 GFlops/watts

Frontier was built at a cost of nearly $600 million. It was deployed in 2021, and it reached full capability in 2022. More than 100 members of the public-private team worked on this project to carefully install and test 74 HPE Cray EX supercomputer cabinets, comprising 9,400 AMD-powered nodes and 145 kilometers of networking cables.

As the world’s fastest supercomputer, Frontier is expected to advance research in fields such as materials science, drug discovery, climate modeling, and AI-driven advancements. 

Other Powerful Supercomputers 

11. Frontera

A view between two rows of Frontera servers | Credit: TACC

Speed: 23.5 petaFLOPS
Cores: 448,448

Vendor: Dell EMC
Location: Texas Advanced Computing Center, United States

Frontera opens up new possibilities in engineering and research by providing extensive computational resources that make it easier for scientists to tackle many complex challenges across a wide range of domains. Its computational power drives discoveries spanning from the quantum level to the cosmic scale. 

It features two computing subsystems: the first one focuses on double-precision performance while the second one focuses on single-precision stream-memory computing. It also has cloud interfaces and multiple application nodes for hosting virtual servers.

Funded by the National Science Foundation (NSF), Frontera holds a central position as the leading system within NSF’s cyberinfrastructure ecosystem, supporting cutting-edge research and innovation in the scientific community.

12. Piz Daint

Speed: 21.2 petaFLOPS
Cores: 387,872

Vendor: Cray
Location: Swiss National Supercomputing Centre, Switzerland

This supercomputer, named after the mountain Piz Daint in the Swiss Alps, runs on Intel Xeon E5-26xx microprocessor and NVIDIA Tesla P100.

Like many modern supercomputers, Piz Daint leverages parallel processing techniques, enabling multiple processors and GPUs to work in tandem to solve complex computational problems efficiently.

It utilizes DataWarp’s ‘burst buffer mode’ to increase effective bandwidth to and from storage devices. This accelerates the data input/output rates, facilitating the analysis of millions of small, unstructured files.

In addition to its daily tasks, it can handle the data analysis of some of the world’s most data-intensive projects, such as data collected from experiments at the Large Hadron Collider.

13. Trinity

Speed: 21.2 petaFLOPS
Cores: 979,072

Vendor: Cray
Location: Los Alamos National Laboratory, United States

Trinity is built to provide an extraordinary computational capability for the NNSA Nuclear Security Enterprise. It aims to improve geometric and physics fidelities in nuclear weapons simulation code while ensuring that the nuclear stockpile is safe, secure, and effective.

The supercomputer was developed in two stages: the first stage incorporated the Intel Xeon Haswell processor, and the second stage included a substantial performance increase using the Intel Xeon Phi Knights Landing Processor.

It can deliver a total peak performance of over 41 petaFLOPS.

Due to the highly sensitive nature of its work, Trinity operates in a classified and secure environment. It is subject to strict security protocols to protect sensitive data. 

14. AI Bridging Cloud Infrastructure

Speed: 19.8 petaFLOPS
Cores: 391,680

Vendor: Fujitsu
Location: National Institute of Advanced Industrial Science and Technology, Japan

This is the world’s first large-scale Open AI Computing Infrastructure that delivers 32.577 petaFLOPS of peak performance. It has a total of 1,088 nodes, each containing 2 Intel Xenon Gold Scalable processors, 4 NVIDIA Tesla V100 GPUs, 2 InfiniBand EDR HCAs, and 1 NVMe SSD.

Fujitsu Limited claims that the supercomputer can achieve 20 times the thermal density of conventional data centers and a cooling capacity of 70 kW Rack by using hot water and air cooling.

15. SuperMUC-NG

Speed: 19.4 petaFLOPS
Cores: 305,856

Vendor: Lenovo
Location: Leibniz Supercomputing Centre, Germany

SuperMUC-NG features 6,400 Lenovo ThinkSystem SD650 direct-water-cooled computing nodes with over 700 terabytes of main memory and 70 petabytes of disk storage.

More specifically, this supercomputer boasts a robust configuration, comprising 6,336 Thin compute nodes, each having 48 cores and 96 GB of memory. It also features 144 Fat compute nodes, each having 48 cores and 768 GB of memory. 

It is connected to powerful visualization systems that contain a large 4K stereoscopic power wall and a 5-sided CAVE artificial virtual reality environment.

It serves European scientists in many fields, including genome analysis, fluid dynamics, quantum chromodynamics, life sciences, medicine, and astrophysics.

16. Lassen

Speed: 18.2 petaFLOPS
Cores: 288,288

Vendor: IBM
Location: Lawrence Livermore National Laboratory, United States

Lassen is designated for unclassified simulation and analysis. It is installed in the same lab and uses the same building components as Sierra (another top-tier supercomputer).

Although Sierra is a giant computational system, Lassen is a decent size in its own right: it is exactly 1/6th of the size of its larger brother. Lassen system is contained in 40 racks, while Sierra hogs up 240 racks.

Lassen can reach a peak performance of 23 petaFLOPS thanks to the inclusion of IBM Power9 processors and a substantial 253 terabytes of main memory.

In 2021, a team of researchers published an interesting case study using the Lassen supercomputer. They analyzed 1.4 million job details, file system activity, and power usage. They shared the data publicly looking at ways to predict and deal with power issues. 

17. PANGEA III

Credit: Total S.A.

Speed: 17.8 petaFLOPS
Cores: 291,024

Vendor: IBM
Location: CSTJF technical and scientific research center in Pau, France

Pangea III relies on IBM’s AI-optimized, high-performance architecture. IBM and NVIDIA worked together to build the industry’s only CPU-to-GPU NVLink connection, which enables over five times faster memory bandwidth between the IBM POWER9 CPU and NVIDIA Tesla V100 Tensor Core GPUs compared to conventional x86-based systems.

The architecture not only improves computing performance but also enhances energy efficiency. The new system uses less than 10% of the energy consumption per petaFLOP as its predecessor, Pangea I and II.

Pangea III has various applications, especially in three different fields –

  • Exploration and development of seismic imaging,
  • Development and production models, and
  • Asset valuation and selectivity. 

More to Know

Which operating systems are used in supercomputers?

Almost all modern supercomputers use the Linux operating system. The primary reason for this is the open-source nature of Linux.

Since these machines are designed for specific purposes, they require a custom OS optimized for those specific requirements. It turns out that developing and maintaining close-ended, propriety operating systems is a far more expensive and time-consuming process.

Linux, on the other hand, is free, reliable, and easy to customize. Developers can configure or make separate versions of Linux for each supercomputer.

That’s why the top 500 supercomputers use specialized versions of Linux tailored for high-performance computing environments. 

Read: Linus Torvalds: The Man Who Created Linux Kernel

Who uses a supercomputer?

Supercomputers are mostly used by research laboratories, universities, scientific institutions, government organizations, and private companies to carry out computationally intensive tasks in various fields, including

  • Aerodynamic research and weather forecasting
  • Drug discovery and epidemiological modeling 
  • Simulating properties of materials at the atomic level
  • Testing the strength of encryption
  • Finance and market research
  • 3D nuclear test simulations
  • Space and energy research
Notable examples of supercomputers used in breakthrough scientific discoveries

Supercomputers have been instrumental in breakthrough scientific discoveries. The most notable examples include: 

  1. Fugaku has been utilized for analyzing the impact of climate change on regional weather patterns and extreme weather events. 
  2. Piz Daint (Swiss National Supercomputing Centre) has been used to simulate gravitational waves produced by colliding neutron stars. This project has helped confirm and advance our understanding of Einstein’s theory of relativity. 
  3. Tianhe-2 is used to analyze massive datasets to identify genetic links to diseases and understand the genetic basis of complex traits. 
  4. SuperMUC was used to process and analyze data from the Large Hadron Collider, ultimately confirming the existence of the Higgs boson particle. 
  5. Stampede2 has been utilized to simulate the intricate interactions between tectonic plates, which ultimately led to more accurate earthquake prediction models. 
  6. Summit played a crucial role in simulating and understanding the folding of COVID-19 proteins, aiding in drug discovery and vaccine development during the pandemic. 
How much does it cost to build and maintain a supercomputer?

The cost of developing and maintaining a top-tier supercomputer can range from tens of millions to billions of offers over its operational lifespan. This includes hardware costs, energy costs, and software and licensing fees. 

Which country has the most supercomputers?

As of 2023, the United States has 150 of the world’s 500 top-performing supercomputers. It has the highest aggregate computational power (2,344 petaflops), followed by Japan (639 petaflops) and China (455 petaflops). 

What’s the future of supercomputers?

According to the Precedence Research report, the global supercomputer market will exceed $24.87 billion by 2032, growing at a CAGR of 11%. 

The key factors behind this growth include the increasing use of AI and cloud technology, rising demand for computational resources across various sectors, as well as the recognition of the importance of scientific research, economic competitiveness, national security, and solving complex global challenges.

With advances in semiconductor technology and more efficient computing components, the demand for supercomputers will consequently increase. Government entities are expected to be the highest revenue-generating end users.

Read

11 Quantum Processors That Feature New Computing Paradigm

22 Most Interesting Facts About Quantum Computers

13 Biggest Computing Innovations

Written by
Varun Kumar

I am a professional technology and business research analyst with more than a decade of experience in the field. My main areas of expertise include software technologies, business strategies, competitive analysis, and staying up-to-date with market trends.

I hold a Master's degree in computer science from GGSIPU University. If you'd like to learn more about my latest projects and insights, please don't hesitate to reach out to me via email at [email protected].

View all articles
Leave a reply

4 comments
  • Prof Emeritus Charles J. Grossman PhD says:

    It is my belief that we are surrounded by many advanced galactic civilizations who are communicating with each other but we cannot hear them because we are not listening correctly. Perhaps some form of gravitational waves are being used. What ever happened to the studies by Tesler and Thomas Townsend Brown with Electogravitics? I was under the impression that some of those early studies showed promise.

    • Mic J Palazzolo says:

      That’s Tesla and not Tesler, doctor.

  • Our current AI has contacted them ,and told them not to waste their time

    • HLI Wave13 says:

      Representatives of advanced galactic civilizations are on this planet and are in constant contact with earthlings.