Schneider Electric, a global digital transformation specialist in energy management and automation, announced a partnership with Nvidia in the first quarter of this year to work together to optimize data center infrastructure and advance innovation in artificial intelligence (AI) and digital twin technologies.
Recently, at the Computing Future Data Center Forum of the 2024 Schneider Electric Innovation Summit, Schneider Electric announced the results of the cooperation between the two sides and launched the first reference design for the intelligent computing center. The design is tailored to deploy Nvidia’s high-performance accelerated computing platform, which aims to provide a standardized benchmark for AI deployment and operations within the data center ecosystem. At the same time, the reference design can also be used as a general intelligent computing center solution for high-density cabinets. This innovative reference design not only lays a solid foundation for building future-ready solutions, but also marks a new milestone in the intelligent computing center industry for efficient deployment and efficient operation.
PCD3.T760
Schneider Electric publishes a reference design for the intelligent Computing Center
Under the guidance of the “artificial intelligence +” strategy, the data center industry is undergoing a far-reaching change, and the intelligent computing center, as the “backbone” of this change, not only promotes the development of the industry, but also leads the future development trend. According to the Action Plan for High-quality Development of Computing Infrastructure, by 2025, the scale of China’s computing power is expected to exceed 300 EFLOPS, and the proportion of intelligent computing power will reach more than 35%. In the face of the rapidly growing demand for intelligent computing power, the deployment of high-density AI clusters has become an inevitable choice for the development of the industry. AI clusters will bring a series of subversive challenges to the design and operation of the physical infrastructure of data centers, including the increasing pressure on the demand for power supply capacity, the design of highly elastic room to adapt to future IT technology, and the digital platform to achieve intelligent operation and maintenance. Developing a standardized reference design can provide evidence-based “guidance” for addressing these challenges, helping to ensure that operators build an efficient, sustainable and scalable infrastructure for future growth.
The No. 99 reference design released by Schneider Electric covers the electromechanical system design of the renovation of the stock data center and the new intelligent computing center, focusing on four technical areas of facility power supply, refrigeration, IT room, and full life cycle software, providing a complete reference design framework for the artificial intelligence cluster with an IT capacity of up to 3770kW to meet the needs of different application scenarios:• Facility power supply optimization: Provide distributed redundant power distribution system for IT room, so as to maximize the computing power under the premise of a certain power supply capacity, or reduce the demand for the power supply capacity under the premise of a PCD3.T760
certain power supply;
• Facility refrigeration optimization: The introduction of CDU cold plate refrigeration system, combined with near-end wind wall and other technologies, to achieve a perfect transition from air cooling to liquid cooling, improve the chip performance and reliability, but also improve the efficiency of the refrigeration system, thereby reducing the overall energy consumption;
• Stock renovation & new IT room: For liquid cooling applications and high-density hardware configuration, focus on the differentiated application scenarios of stock data center upgrading and new intelligent computing center, and provide feasible reference design to realize the deployment of high-density AI cluster;
• Full life cycle software: Advanced planning, design and operation tools provide basis for enterprise layout decisions, improve deployment efficiency, and realize data center visualization, reduce operational risks, and efficiently ensure the safe and reliable operation of the data center throughout the life cycle.