ASUS and Taiwan’s National Center for High-performance Computing (NCHC) have launched a new AI supercomputer that has entered official operation and secured the No. 29 position on the TOP500 list. The system features a dual-architecture design combining the Nano4 NVIDIA HGX H200 cluster with the NVIDIA GB200 NVL72 platform—Taiwan’s first fully liquid-cooled deployment of this configuration. The Nano4 cluster delivers 81.55 PFLOPS of performance, enhancing Taiwan’s AI computing capacity for applications spanning generative AI, deep learning, big data, and HPC.
NCHC designed the Nano4 architecture and liquid-cooling system, while ASUS provided integration expertise and infrastructure deployment support. The supercomputer utilizes direct liquid-cooling technology achieving a PUE of 1.18, combining performance efficiency with reduced energy consumption. Its compute environment includes ASUS ESC N8-E11V servers with NVIDIA HGX H200, ESC8000-E12 systems with NVIDIA MGX H200 GPUs, and ESC NM2N721-E1 servers using the GB200 NVL72 architecture.
For storage, ASUS incorporated OJ340A-RS60 and RS501A-E12-RS12U systems supporting multiple data types and configurations. Deployment time was shortened significantly through the ASUS Infrastructure Deployment Center, reducing cluster setup from three weeks to three days. ASUS Professional Services further supported validation, integration, and optimization to ensure scalable and reliable AI operations.
Leave a comment