The Supercomputer Race: Frontier Dominates as Europe Prepares for Exascale Upgrade

In the realm of high-performance computing, a groundbreaking milestone is on the horizon as the world’s most powerful supercomputers are slated to perform quintillion calculations per second by 2024. As our global society continues to generate and rely on an immense wealth of data across all sectors, these technological leviathans play a crucial role, offering unparalleled support in scientific research and numerous fields that benefit human progress.

Monstrous in their calculation capabilities, supercomputers like Frontier at the Oak Ridge National Laboratory (ORNL) in Tennessee are redefining what's possible. Frontier has now cemented its position as the world’s foremost supercomputer by achieving an astonishing 1.194 exaflops per second on the High-Performance Linpack (HPL) benchmark.

The expansive use of supercomputers extends to an array of scientific domains, such as simulating nuclear fusion for cleaner energy, delving deep into quantum mechanics, modeling biological molecules to understand life processes, cracking complex cryptographic systems, and predicting climatic changes with higher accuracy than ever before.

Frontier, which is the brainchild of the combined innovation of HPE and AMD, is a technical masterpiece occupying 680 square meters at ORNL using Hewlett Packard Enterprise (HPE) Cray EX architecture. This beast of a machine integrates 9,472 3rd-generation AMD Epic processors and 37,880 AMD Radeon Instinct accelerators to achieve its record-breaking performance, all the while leading energy efficiency with its Green500 ranking.

Contemporaneously, the second-fastest supercomputer, Aurora, located at the Argonne Leadership Computing Facility (ALCF), stands as another exemplar of computational might. Although unfinished, it has already reached half its expected capacity with an impressive 585 Pflop/s. The full completion of Aurora is poised to reach exascale status by 2024, making it even faster than Frontier.

Aurora’s noteworthy specifications are a testament to Intel’s processing prowess. With Intel Xeon Sapphire Rapids CPUs and Ponte Vecchio GPUs at its core, Aurora's architecture is a harmonious blend of 21,248 CPUs and 63,744 GPUs. Its sheer computational strength will support workloads across multiple scientific frontiers from atomic to aerodynamic research.

Yet, the concentration of such power isn’t solely restricted to the United States. LUMI, Europe’s answer to supercomputing challenges, showcases a strong inclination towards sustainable energy, being fully powered by hydro energy. Housed by the European High-Performance Computing Joint Undertaking, LUMI not only dazzles with a capacity equivalent to 1.5 million modern laptops but also warms homes in Kajaani with its waste heat.

Interestingly, while Microsoft's cloud-based Eagle supercomputer offers unprecedented core counts through Intel and Nvidia partnerships, Asia’s crown jewel, Fugaku, developed by Riken Centre for Computational Science, boasts a towering core count following its upgrade, enhancing its already formidable computational prowess.

The ascent into the exascale era heralds a new chapter in computing, especially with EuroHPC’s ambitious project in the Jülich Supercomputing Centre anticipated to come online by the end of 2024. This pioneering European exascale computer is expected to contribute significantly to AI advancements, medical simulations, and climate projections that could reshape our planet's future.