HPE's supercomputer El Capitan was built in partnership with AMD and delivered to the National Nuclear Security Administration (NNSA) and Lawrence Livermore National Laboratory (LLNL).

With HPE's Cray supercomputer EX background, AMD's EPYC processors, and HPE's 100% fanless direct water cooling architecture, El Capitan is not only amazingly powerful, it's energy and cost efficient too.

El Capitan was designed to help the United States maintain its competitive edge in national security using high-fidelity modeling and simulation to solve challenging problems that are only within reach using exascale computing. It will also further the country's investment in artificial intelligence (AI) to advance scientific research in areas such as nuclear science, materials discovery and fusion energy.

The assessment was performed independently by Top 500, in the November 2024 release of its global supercomputer rankings. El Capitan has been measured and recorded with some absolutely incredible numbers. For example, a mind-blowing 11,039,616 CPU cores running AMD 4th Gen EPYC 24C 1.8GHz CPUs. It has been measured at 1.742 exaflops, with a theoretical peak performance of 2.7 exaflops. A "flop" is a floating point operation, and an exaflop is a thousand petaflops, which itself is a thousand trillion. It's a staggering number; El Capitan can perform over a thousand million trillion floating point operations every single second.

Now, if you're used to working with per-core licensing you might shudder to think what it would cost to run certain commercial products on El Capitan. Not to worry; it runs the TOSS operating system or Tri-Lab Operating System Stack (hence, TOSS) which is itself based on Red Hat Linux. TOSS is used across HPE supercomputers and other products. It's used by NASA as their primary operating system for their NAS clusters. You can see more about the software that runs on El Capitan here, but it boils down to Linux/Unix like many University and research environments, albeit in this case with lots and lots and lots and lots and lots of sheer raw processing power, churning through mathematical calculations and data like a beast - the biggest man-made computing beast in the world.

And, while El Capitan is the faster supercomputer in the world - advancing AI-informed scientific discovery around energy and security and other fields, HPE also holds the number two and three spots with Frontier and Aurora, respectively. It's not until the fourth-fastest supercomputer in the world that we see another vendor, in this case Microsoft, with its Microsoft Azure-based Eagle supercomputer. Position five is back to HPE with the machine dubbed HPC6. In fact, seven of the top 10 faster supercomputers in the world are all from HPE.

Meanwhile, HPE supercomputers also make the Green500 list, which ranks based on how much computational performance they deliver on the HPL benchmark per Watt of electrical power consumed, measured in gigaflops per Watt. This ranking is not driven by the size of the system but by its technology. Here, HPE stands strong with eight out of the top 15 machines being HPE too.

El Capitan consumes 29,580.98 kW of power, so you'll want a robust power source to plug it in, but that energy is made more efficient than ever before with HPE's entirely fanless liquid cooling design that reduces energy consumption of cooling infrastructure by up to 94% when compared to a traditional air handler-based air-cooling system. Customers can also monitor and regulate power utilisation.

The message is clear: if you want fast and energy-efficient supercomputing systems, then HPE is not only a proven leader, but the proven leader, and is enabling research institutions and large enterprise companies to handle larger workloads, accelerating discovery and innovation.

Of course, most of us will never get to order and commission a supercomputer like El Capitan, but HPE's chops in this field can give confidence that when you're considering on-premises server products for your own computational work - of which generative AI is becoming increasingly prominent - you can be confident HPE's tech is tried and tested at huge scale, and is also backed by strong sustainable, energy-efficient practices, processes, and designs.

HPE states it "is proud to have built supercomputers that power the modelling and simulation capabilities enabling scientists to make discoveries that benefit society. HPC and AI workloads running on these powerful supercomputers will accelerate the next innovative breakthroughs in agriculture, finance, drug discovery, healthcare, energy, weather and climate, cybersecurity and national defence. HPE continues to forge ahead, leading the next wave of breakthroughs in supercomputers through its many partnerships in the public and private sectors. These collaborations help HPE customers reach new to new heights that will enable researchers to push their simulations to resolutions and scales never before possible."

In addition, AMD notes the top two supercomputers in the world today are both AMD-powered, and that El Capitan and Frontier together provide over three exaflops of performance, and account for 61% of the total performance of the top 10. In fact, AMD has powered the faster supercomputer in the world, according to the Top500 lists, for six consecutive lists, with Frontier having been #1 since June 2022, and now El Capitan from November 2024. Another three supercomputers in the Top500 are AMD machines too. The AMD EPYC processor also powers 15 of the top 25 machines on the Green500 list.

El Capitan uses the AMD Instinct MI300A APUs, and iTWire has been following AMD's strong leadership and aggressive push to be the number one choice for the new world of generative AI. The news of El Capitan's Top500 ranking comes hot on the heels of rival Intel's announcement of a $1.6 billion loss and subsequent resignation of CEO Pat Gelsinger.

“We are thrilled to see El Capitan become the second AMD powered supercomputer to break the exaflop barrier and become the fastest supercomputer in the world. Showcasing the incredible performance and efficiency of the AMD Instinct MI300 APUs, this groundbreaking machine is a testament to the dedicated work between AMD, LLNL and HPE,” said AMD EVP and GM Forrest Norrod. “At AMD, we are driving the future of computing with leadership performance and capabilities that will continue to define the convergence of HPC and AI for years to come.”

"El Capitan is crucial to the National Nuclear Security Administration’s core mission and significantly bolsters our ability to perform large ensembles of high-fidelity 3D simulations that address the intricate scientific challenges facing the mission," said LLNL director Rob Neely.

LLNL CTO Bronis R. de Supinski said "leveraging the AMD Instinct MI300A APUs, we've built a system that was once unimaginable, pushing the absolute boundaries of computational performance while maintaining exceptional energy efficiency. With AI becoming increasingly prevalent in our field, El Capitan allows us to integrate AI with our traditional simulation and modeling workloads, opening new avenues for discovery across various scientific disciplines."

Like with HPE, AMD's supercomputer leadership extends to its enterprise line of AMD EPYC 9005 series processors, as well as its Instinct accelerators and other devices.

El Capitan will also be used by LLNL and the other NNSA Tri-Labs to drive AI and machine learnig-assisted data analysis, to create scientific models that are fast, accurate, and capable of quantifying uncertainty in their predictions. El Capitan will help apply AI to high energy density problems such as internal confinement fusion research.