What is the Graph500 and why is this benchmark relevant?
The Graph500 is an international benchmark that ranks supercomputers based on their performance in analyzing large-scale graphs. Unlike benchmarks like the TOP500, which focus on raw computing power through floating-point operations, the Graph500 assesses the effectiveness of systems in data-intensive tasks such as network and social analysis, cybersecurity, and bioinformatics. The key measurement within this benchmark is the Breadth-First Search (BFS), an algorithm that is essential for quickly searching through vast amounts of interconnected data.
For IT professionals and researchers in high-performance computing (HPC), the Graph500 is a valuable gauge for applications where data patterns, relationships, and structures are more important than numerical computing power. This includes AI training, genomics, fraud analysis, and weather simulations. It is therefore a benchmark that is becoming increasingly relevant in a world where data interpretation is at least as important as data speed.
Fugaku maintains top position in Graph500
During ISC High Performance 2025 in Hamburg, it was announced that the Japanese supercomputer Fugaku has achieved first place in the Graph500 ranking for the eleventh consecutive time. With an impressive score of 204,068 TeraTEPS (trillions of traversed edges per second), Fugaku once again demonstrates its strength in processing complex graph data.
Fugaku was developed by the RIKEN Center for Computational Science in collaboration with Fujitsu and consists of no less than 152,064 nodes. In addition to its victory in the Graph500, the system also claimed second place in the HPCG benchmark (High Performance Conjugate Gradients), seventh position in the TOP500, and sixth place in the HPL-MxP ranking.
The continuous top performance is the result of collaboration between RIKEN, the Institute of Science Tokyo, Fixstars Corporation, Nippon Telegraph & Telephone Corporation, and Fujitsu. The technical infrastructure behind Fugaku, based on the Arm architecture, excels in scalability and energy efficiency, two elements that are becoming increasingly crucial in the world of supercomputing.
Technological breakthroughs and societal impact
Fugaku distinguishes itself not only by brute computing power but also by its contribution to societal applications. The supercomputer was used in collaboration with Yokohama National University for the first real-time prediction of multiple tornadoes caused by typhoons. By combining advanced weather simulations with large-scale parallel processing, a new level of speed and accuracy in climate modeling was achieved.
The application of Fugaku extends beyond meteorology. In the life sciences, disaster response, energy research, production optimization, and socio-economic analyses, the supercomputer has proven its value. This broad applicability makes Fugaku a key player in applying HPC on a societal scale.
From simulations to energy-efficient CPUs: what’s next?
The technology behind Fugaku forms the basis for FUJITSU-MONAKA, a new generation of CPUs based on a many-core Arm architecture. This processor, developed with support from the Japanese NEDO program (New Energy and Industrial Technology Development Organization), is manufactured using 2nm technology and combines high performance with energy efficiency and broad software compatibility.
This step demonstrates that innovations from the world of supercomputing not only contribute to scientific progress but also find their way into broader IT environments. Think of data centers, cloud infrastructures, and edge computing, where the demand for powerful yet efficient hardware is growing.
Fugaku remains dominant
The continued dominance of Fugaku in the Graph500 underscores not only Japan's leading position in the world of supercomputing but also highlights the growing role of graph-based benchmarks in modern IT applications. At a time when data analysis, AI, and simulations are becoming increasingly complex, performances like those of Fugaku provide a clear perspective on where the next step in computing lies.
