Supermicro Begins Mass Production of Nvidia Vera Rubin Liquid-Cooled Racks, Starts NVL72 Rack Shipments
TECHWORLD ·
✦ AI Summary
Supermicro said it has begun shipping racks for Nvidia's next-generation AI platform, the Vera Rubin NVL72.
Supermicro announced on the 29th the start of shipments for the Vera Rubin NVL72 rack, which integrates DCBBS and the direct liquid cooling technology stack (DLC-2).
Supermicro presented a management and support system that includes liquid cooling system integration, rack-level testing and validation, and support for data center deployment.
Supermicro said it has begun shipping racks for Nvidia's next-generation AI platform, the Vera Rubin NVL72. On the 29th, Supermicro announced the start of shipments for the Nvidia Vera Rubin NVL72 rack, which integrates DCBBS and the direct liquid cooling technology stack (DLC-2).
The Nvidia Vera Rubin platform is designed around direct liquid cooling. As AI computing performance improves, power consumption and heat generation are increasing, and it is becoming more difficult to address thermal management in high-density AI infrastructure with air cooling alone.
Accordingly, Supermicro integrated the entire coolant circulation system, including the cold plate, manifold and cooling tower connections. With its integrated liquid cooling system, Supermicro manages the heat generated by AI systems.
Supermicro conducted rack-level testing and validation with the liquid cooling system fully integrated. This follows Supermicro's strategy to shorten the time from on-site data center deployment to the start of operations.
Supermicro presented DLC-2, its direct liquid cooling technology stack optimized for the Vera Rubin platform. Supermicro's portfolio includes cold plates, manifolds, hose kits, rack power shelves, in-row cooling distribution units (CDU), in-rack CDU, liquid-to-air sidecar CDU, rear door heat exchangers and facility cooling towers. This portfolio includes the components needed for liquid cooling.
Supermicro provides DCBBS blueprints that address power scales from 5MW to gigawatt levels. The DCBBS scalable unit supports up to 1,152 GPUs and 331TB of HBM4 memory. The company says these large-scale expansion specifications support data center scale-out.
Supermicro manages the entire data center deployment process, from site surveys to ongoing support, with a single dedicated team. The scope of integrated management includes project design, systems integration, L11 and L12 testing, on-site deployment and ongoing support. The scope of support includes individual server system design, mass production of liquid-cooled racks and on-site deployment of clusters based on Nvidia reference architectures.
Charles Liang, Supermicro president and CEO, said the company has spent years building up its liquid cooling technology stack, manufacturing capabilities and deployment team. He said this accumulation enables customers to order scalable units.
He added that end-to-end integrated systems are immediately ready for commercial environments. He also said that the DLC-2 technology developed by Supermicro is at the core of its differentiated capabilities, and that the technology makes it possible to provide a fully integrated solution for the Nvidia Vera Rubin platform.
Source: TECHWORLD · Park Gyu-chan
Original: https://www.epnc.co.kr/news/articleView.html?idxno=407458
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.