AMD continues to lead the charge in high-performance computing (HPC), as it powers the world’s fastest supercomputer, El Capitan. This marks the second AMD-powered supercomputer to break the exaflop barrier, solidifying their dominance in the field of exascale computing.
El Capitan, housed at the Lawrence Livermore National Laboratory (LLNL), is equipped with AMD Instinct⢠MI300A APUs and built by Hewlett Packard Enterprise (HPE). It achieved an impressive High-Performance Linpack (HPL) score of 1.742 exaflops, according to the latest Top500 list. This achievement highlights AMD’s commitment to pushing the boundaries of computational power and energy efficiency.
Energy Efficiency and Performance Leadership
Not only does El Capitan top the performance rankings, but it also ranks highly in energy efficiency. The system, alongside the Frontier supercomputer at Oak Ridge National Lab, claimed spots on the Green500 list, proving the energy efficiency of AMD EPYC processors and AMD Instinct GPUs in HPC workloads.
âWe 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,â said Forrest Norrod, executive vice president and general manager at AMD. âThis is a testament to the close collaboration between AMD, LLNL, and HPE, as we continue driving innovation in HPC and AI performance.â
Impact on National Security
El Capitan is set to play a critical role in advancing the National Nuclear Security Administration’s (NNSA) mission, enabling a new level of precision in 3D simulations that are crucial for national security. It will allow scientists to tackle high-stakes scientific challenges, such as nuclear deterrent safety, without the need for physical testing.
Rob Neely, director of LLNLâs Advanced Simulation and Computing program, said, âEl Capitan significantly enhances our ability to perform large-scale, high-fidelity simulations that are vital to ensuring the safety and reliability of our nationâs nuclear stockpile.â
Integrating AI with HPC
One of the most exciting aspects of El Capitan is its ability to integrate AI into traditional simulation workloads. This opens up new possibilities for research in fields like climate modeling, biosecurity, and even drug discovery. The system offers unprecedented computational capabilities that will drive AI advancements in scientific research.
Bronis R. de Supinski, LLNLâs chief technology officer for Livermore Computing, highlighted the importance of AI integration: âWith the AMD Instinct MI300A APUs, weâve built a system that was once unimaginable. El Capitan will allow us to integrate AI with our traditional workloads, paving the way for groundbreaking discoveries across multiple disciplines.â
AMDâs Role in Global Supercomputing
Beyond El Capitan, AMD is powering other major supercomputing projects worldwide. For instance, the Frontier system at Oak Ridge National Lab, which is powered by AMD EPYC CPUs and AMD Instinct GPUs, is the second-fastest supercomputer globally. Additionally, AMD is involved in other projects, such as the HPC 6 supercomputer for Italian energy company Eni, and upcoming installations in Norway and Japan.
In a strategic collaboration, AMD and IBM are set to deploy AMD Instinct MI300X accelerators as a service on IBM Cloud, further expanding their reach in AI and HPC solutions. This partnership will enhance performance for AI models and scientific research, providing powerful tools for enterprises and research institutions alike.
Conclusion
As the only company powering multiple exascale supercomputers, AMD is proving to be a driving force in the future of HPC and AI. With El Capitan set to lead scientific innovation and national security initiatives, and with AMD’s continued investments in HPC infrastructure worldwide, the future of exascale computing looks brighter than ever.
