The RIKEN Center for Computational Science (R-CCS) has completed the installation and initiated the operational launch of its new supercomputer, “ROQUO,” at its campus in Kobe, Japan. Developed under a project commissioned by the New Energy and Industrial Technology Development Organization (NEDO), the system serves as the high-performance computing (HPC) backbone for Japan’s national hybrid computing initiative. The platform is designed to establish a unified infrastructure that seamlessly bridges classical supercomputing acceleration with physical quantum processors, addressing advanced computational workloads that cannot be resolved by traditional architectures alone.
The system’s technical architecture consists of 135 compute nodes equipped with the NVIDIA GB200 NVL4 platform, pooling a total of 540 Blackwell graphics processing units (GPUs) and 270 Grace central processing units (CPUs). Inter-node communication is managed by NVIDIA Quantum-X800 InfiniBand networking fabric, which delivers high-speed data transfers up to 3.2 terabits per second with low signal latency. To maximize energy efficiency under heavy workloads, ROQUO utilizes a specialized warm-water cooling infrastructure that relies on natural outside air towers to cool the system with 32°C water, reducing total facility power consumption by approximately 20% or more compared to standard air-cooled baselines.
During operational validation preparations, the supercomputer was evaluated using the High Performance LINPACK benchmark across the entire 135-node cluster. ROQUO achieved a measured double-precision floating-point performance of 19.80 petaflops, matching its targeted operational thresholds for scientific and technical computing. This stable performance baseline is sustained by the system’s structural balance between raw processing capacity and high-throughput communication routing, which efficiently coordinates the collective data synchronizations and large-scale parallel matrices required by intensive classical simulations.
A primary operational objective for the platform is the demonstration of tightly coupled, multi-backend hybrid workflows using the software-based SQC Interface. ROQUO is physically interconnected with Japan’s flagship supercomputer, Fugaku, as well as on-premises quantum hardware including the superconducting IBM Quantum System Two, designated as ibm_kobe, and Quantinuum’s Reimei trapped-ion quantum computer. This consolidated network allows researchers to execute complex hybrid applications, run quantum circuit simulations at scale, and benchmark noisy intermediate-scale quantum outputs against classical references through a distributed cloud topology.
Moving forward, R-CCS will manage the system’s daily operations, providing computing resources to the scientific community and industrial partners through an open test user program. The operational experience accumulated from managing this hardware environment—particularly the coordination of massive GPU clusters and high-efficiency liquid cooling—will be directly applied to the development of Japan’s next-generation flagship supercomputer, codenamed FugakuNEXT. Initial application frameworks deployed on the machine will focus on pioneering converged domains like quantum machine learning, algorithm optimization, and post-5G secure communication protocols.
The official operational disclosures, full system architectures, and institutional project descriptions can be reviewed through the RIKEN Corporate Newsroom here and the detailed RIKEN R-CCS Technical Portal here.
June 23, 2026
Leave A Comment