On the Supercomputing 2024 convention at present underway in Atlanta, AMD, in collaboration with HPE and the Lawrence Livermore Nationwide Laboratory (LLNL), introduced what has been formally declared because the world’s strongest supercomputer, El Capitan. El Capitan is the primary exascale-class machine for the Nationwide Nuclear Safety Administration (NNSA), however may even be highly effective useful resource for the NNSA Tri-Labs- LLNL, Los Alamos, and Sandia Nationwide Laboratories as effectively.
El Capitan’s primary space of focus is nationwide safety. The system might be tasked with the complicated modeling and simulation required for the NNSA’s Stockpile Stewardship Program that certifies the getting old nuclear weapons stockpile. It would, nonetheless, additionally deal with the computing necessities for different vital nuclear safety missions and counterterrorism analysis.
El Capitan was constructed by HPE on its direct liquid-cooled blade-based, high-density clustered Cray EX255a HPC system. It’s powered by AMD Intuition MI300A APUs – which pack each CPU and GPU cores — and leverages HPE’s high-bandwidth Slingshot-11 interconnect expertise.
“We’re thrilled to see El Capitan develop into the second AMD powered supercomputer to interrupt the exaflop barrier and develop into the quickest supercomputer on the planet. Showcasing the unbelievable efficiency and effectivity of the AMD Intuition MI300 APUs, this groundbreaking machine is a testomony to the devoted work between AMD, LLNL and HPE,” stated Forrest Norrod, govt vice chairman and common supervisor, AMD. “At AMD, we’re driving the way forward for computing with management efficiency and capabilities that may proceed to outline the convergence of HPC and AI for years to return.”
The AMD MI300A packs 24 ‘Zen 4’ cores throughout three chiplets, at the side of six accelerated compute dies (XCDs) with 38 compute models (CUs), every with 32 KB of L1 cache, 4 MB L2 cache shared throughout CUs, and 256 MB AMD Infinity Cache shared between XCDs and CPUs. There’s additionally 128GB of HBM3 reminiscence shared coherently between CPUs and GPUs, providing as much as 5.3TB/s of peak throughput. All instructed, El Capitan includes a whopping 11,039,616 complete cores, providing a measured 1,742 PFlop/s (Rmax), with a theoretical peak 2,746.38 PFlop/s. And the system consumes 29,580.98 kW of energy.
That’s a ton of computing energy, which landed El Capitan at #1 atop the Top500 supercomputer record, making it the quickest supercomputer on the planet at the moment. The system additionally lands at quantity 18 on the on the Green500 record, which ranks the techniques within the Top500 by way of power effectivity. AMD has had nice success in its supercomputing efforts. The corporate powers the no. 1 and no. 2 quickest supercomputers on the planet, in El Capitan and Frontier (Oakridge Nationwide Labs), but in addition 5 of the highest 10 on the Top500 and 40% of the ten most power environment friendly supercomputers on the planet.
