The list of The world’s most powerful supercomputers have a new leader: the supercomputer The Captainbuilt by HPE for Lawrence Livermore National Laboratory (LLNL) of the United States Nuclear National Security Administration (NNSA). The system features sustained computing power of 1.74 exaflops, with peak performance reaching 2.79 exaflops.
Thanks to its power, El Capitan has dethroned what until now was the most powerful supercomputer in the world: Frontier, which now becomes the second. It incorporates AMD chips, specifically the AMD instinct MI300A APUs, which combine CPU, GPU and memory cores with high bandwidth. Not only is it the fastest and most powerful computer, but it is also among the 20 most sustainable supercomputers in the world.
The system incorporates more than 11,000 computing nodes, and 5.4375 Petabytes of memory, achieving peak performance 20 times higher than that of LLNL’s previous supercomputer, Sierra, which reaches a maximum performance of 125 Petaflops. It is developed from the HPE Cray Supercomputing EX system, designed specifically for exascale computing. Additionally, it integrates HPE direct liquid cooling solutions.
El Capitan will allow the laboratory in which it is installed, in addition to advancing in different fields of science, to significantly improve its ability to predict and model the performance of nuclear weapons as these systems are renewed and updated. In this way, LLNL will be able to better understand its situation and capacity, and improve its ability to make informed decisions regarding security.
But this will not be the only task for which El Capitan will be used. Its computing power will also be used in various missions, focused on emergency aid. For example, in natural disasters and other human-caused crises.
It will be able to do so, among other things, because it is capable of offering higher resolution in 3D modeling than was possible until now with other supercomputers. Either because they were not able to obtain it or because it was too expensive in terms of computing to obtain it regularly.
In addition, El Capitan can also be able to more precisely replicate medications, allowing higher quality analyzes of its components and allowing scientists to incorporate more real-world factors, such as environmental conditions or imperfections, into their tests and analyses. fruit of manufacturing processes.
Until now, the calculations necessary to do this were impossible, or it took weeks or months for LLNL systems to obtain them. With this supercomputer, these calculations are solved in days and even a few hours.
Apart from this, it will also be used in the modeling of high energy density physics experiments, such as those of inertial confinement fusion, the study of the dynamics of ballistic reentry and the improvement of the understanding of the behavior of materials in conditions extreme.
Bronis R. de Supinski, LLNL Technology Lead for Livermore Computinghas commented that «Taking advantage of the AMD Instinct MI300A APUs we have developed a system that was until now unimaginable, surpassing the limits of computational performance while maintaining exceptional energy efficiency. With AI becoming more prevalent in our field, El Capitan allows us to integrate it with our traditional modeling and simulation workloads, opening new avenues for discovery across various scientific disciplines.«.
On the part of AMD, Forrest Norrod, Vice President and head of the companystressed that they are delighted to see El Capitan «become the second supercomputer powered by AMD to break the exaflop barrier and become the fastest supercomputer in the world. Showcasing the incredible performance and efficiency of AMD Instinct MI300 APUs, this revolutionary machine is a tribute to the dedicated work carried out between AMD, LLNL and HPE«.
