Inspur Information, a IT infrastructure solutions provider, announced it will be offering cold plate liquid-cooling technology in all of its server products.
Inspur Information has incorporated cold plate liquid-cooling technology into all of its products (general-purpose servers, high-density servers, rack servers, and AI servers), which can be fully customized for a diverse array of scenarios.
Inspur Information’s liquid-cooled general-purpose servers are adaptable to a range of scenarios due to their high flexibility. NF5280M6 is a 2U general-purpose flagship product powered by third-generation Intel Xeon Scalable processors.
Cold-plate liquid-cooling technology reduces its power consumption for cooling by 80 percent, greatly improving reliability, availability, and serviceability (RAS).
NF5280M6 allows the delivery of rack solutions while offering flexibility in decoupling from liquid-cooled racks. NF5260FM6 is an innovative general-purpose server designed for large data centers. Its hybrid air-liquid cooling system supports a liquid-cooling mode for high-power consumption modules, such as processors, and an air-cooling mode for low-power consumption modules, such as drives and I/O cards.
Inspur Information developed the liquid-cooled rack server ORS3000S based on its experience in open-computing rack servers. It increases cooling efficiency by 50 percent. It features a centralized power supply design with N+N redundancy, allowing power for multiple nodes, and ensuring the highest power supply efficiency at all times thanks to dynamic power balancing and optimization technology.
Compared to products that use a distributed power supply, it has a 10 percent higher overall efficiency. It also supports integrated deployment of computing, storage, and networking modules. It also enables efficient maintenance and deployment, shortening the delivery cycle by 5-10 times.
The product has been deployed on a large scale in JD Cloud data centers, providing computing power for online shopping festivals with increased performance of 34 percent–56 percent, lower PUE (Power Usage Effectiveness), and reduced carbon emissions.
Inspur Information launched several liquid-cooled high-density servers to meet the demands of applications such as simulation and R&D in universities and research institutes. The i24LM6 server supports 4 dual-socket nodes in a 2U space, and provides computing power of 447.5 TFLOPS in a single rack.
Inspur Information’s liquid-cooled AI servers cater to the diverse demands for computing power in the AI field. NF5688LM6 supports eight NVIDIA A100 Tensor Core GPUs with NVIDIA NVLink technology to deliver maximum GPU computing power. It comes with 13 PCIe expansion slots to provide PCIe expandability for the training of massive AI models. Moreover, the modular design of the CPU and GPU cold plates improves ease of installation and maintenance.
NF5498LA5 is the industry’s first liquid-cooled AI server designed for open accelerated computing. It supports OAM-compliant accelerators from multiple manufacturers and has a PUE as low as 1.1. The high-bandwidth fully-connected topological structure enables the direct interaction between any two OAMs and provides peer-to-peer bandwidth of up to 896 GB/s.
Inspur Information recently built Asia’s largest development and manufacturing facility for liquid-cooled data centers. The facility has an annual capacity of 100,000 servers. The facility is the first in the industry to achieve mass delivery of liquid-cooled rack servers with a total delivery cycle of 5-7 days, helping data centers reduce their PUE to below 1.1.