AMD Turbocharges El Capitan AMD Exascale Computing fastest supercomputer placed sixth in Supercomputing 2024.
Exascale Meaning
Exascale computing is a subset of supercomputing that is extremely powerful. Systems that use an infrastructure of CPUs and GPUs to handle and analyze data may execute billions of calculations per second.
The newest Top500 list states that the El Capitan supercomputer at Lawrence Livermore National Laboratory (LLNL), powered by AMD Instinct MI300A APUs and constructed by HPE, is the fastest supercomputer in the world with a High-Performance Linpack (HPL) score of 1.742 exa Oak Ridge National Lab’s Frontier system and El Capitan, which placed 18 and 22 on the Green500 list, showed AMD EPYC CPUs and AMD Instinct GPUs’ HPC performance and energy efficiency.
AMD Powering HPC and AI
The most significant supercomputers are still powered by AMD compute engines, which also provide outstanding technical computing performance to businesses and government labs worldwide.
With up to 37% greater generational IPC performance for HPC and AI applications, the most recent AMD EPYC 9005 Series processors are the finest server CPUs for business, AI, and cloud computing. In comparison to the competition, these processors offer up to 3.9X quicker time to insights for scientific and HPC applications that address the most difficult issues in the world.
From AI solutions to AMD Exascale-class supercomputers, AMD Instinct accelerators offer the data center’s best performance at any scale. AMD Instinct MI300X and MI325X accelerators provide improved AI performance and memory, while the MI300A APU combines CPU and GPU cores with stacked memory for HPC and AI applications.
AMD EPYC CPUs and AMD Instinct accelerators power several innovative AI and supercomputing initiatives, including:
Italian energy giant Eni introduced the HPC 6 supercomputer featuring AMD EPYC CPUs and Instinct GPUs. Powerful industrial supercomputer HPC 6 is sixth fastest worldwide. A new supercomputer with 5th Gen AMD EPYC CPUs is being delivered and installed at Paderborn University.
An HPE Cray Supercomputing EX system with 5th Gen AMD EPYC CPUs will be used by Sigma2 AS to replace two of Norway’s three state-owned supercomputers. This supercomputer is anticipated to be Norway’s fastest system once it is completely built.
AMD and IBM have partnered to offer AMD Instinct MI300X accelerators on IBM Cloud as a service.
With the goal of improving performance and power efficiency for Gen AI models, such as high-performance computing applications for business clients, this solution is anticipated to become available in the first half of 2025. Additionally, the partnership will make it possible for Red Hat Enterprise Linux AI inferencing and AMD Instinct MI300X accelerators to be supported within IBM’s Watsonx AI and data platform.
Additionally, a next-generation supercomputer system for Japan’s National Institutes for Quantum Science and Technology (QST) will be powered by AMD Instinct MI300A APUs. The NEC Corporation-built system will power scientific and artificial intelligence research for the National Institutes for Fusion Science and Quantum Science and Technology using 280 AMD Instinct MI300A APUs.
Leading the AMD Exascale Era
AMD continues to lead HPC deployments globally in terms of performance and energy efficiency as the sole manufacturer powering many exascale supercomputers.
The NNSA Tri-Labs LLNL, Los Alamos, and Sandia National Laboratories rely on El Capitan, the most potent supercomputer in the world and the first exascale-class machine for the NNSA, as their primary computing resource. By offering the enormous processing capacity required to guarantee the safety, security, and dependability of the country’s nuclear deterrent without testing, it will be utilized to promote scientific research and national security.
This cutting-edge system represents a significant advancement in HPC, allowing for previously unheard-of modeling and simulation capabilities that are crucial for other vital nuclear security missions, including counterterrorism and nonproliferation, as well as NNSA’s Stockpile Stewardship Program, which certifies the aging nuclear stockpile.
In order to further advance LLNL’s AI-driven objectives of developing scientific models that are quick, precise, and able to quantify uncertainty in their predictions, LLNL and the other NNSA Tri-Labs are also utilizing El Capitan and its companion system, Tuolumne, to drive AI and machine learning-assisted data analysis. Tuolumne will be utilized for unclassified open scientific applications such as seismic modeling, biosecurity/drug discovery, and climate modeling, while El Capitan will apply AI to high energy density challenges like inertial confinement fusion research.
Beyond El Capitan, AMD and HPE power Frontier, the first AMD exascale supercomputer. Oak Ridge National Lab’s Frontier is the second-fastest computer in the world at 1.35 exaflops. AMD Instinct GPUs and EPYC CPUs power it. Frontier supports researchers in biomedical research, climate modeling, and large language model training, underlining its importance in scientific growth and AI advancements.
These cutting-edge systems offer enormous computing capacity that greatly advances a variety of fields of study, such as materials science, climate modeling, and the creation of AI models. By empowering researchers in many domains and supporting AI model development, El Capitan and Frontier are shaping science and technology and providing answers to global concerns. AMD is committed to offering high-performance computer resources to advance scientific research.
