Inspur released two four-socket servers that support the latest  3rd Gen Intel Xeon scalable processors, aiming to provide customers with higher density, greater performance, and reliable computing power to support digital transformation.

San Jose, California, June 18, 2020 – Inspur, a global leading IT infrastructure provider, released two M6 series four-socket servers supporting the latest 3rd Gen Intel Xeon scalable processors, including NF8260M6, a 2U 4-socket server optimized for cloud scenarios, and NF8480M6, a 4U 4-socket server for mission critical application scenarios of traditional enterprise customers.

The 3rd Gen Intel Xeon Scalable Processor  supports four-socket and eight-socket servers, with a single CPU of up to 28 cores. Bfloat16 numeric format is supported for the first time in a mainstream CPU to enhance the AI deep learning performance and computing speed. Meanwhile, application optimization is carried out in cloud computing, in-memory database, deep learning and virtualization.

Inspur released a four-socket server supporting the platform, aiming to provide customers with higher density, greater performance, and reliable computing power to support digital transformation.

Mining data value with higher compute density

Inspur NF8260M6 can support 4 processors in a 2U space, up to 112 cores, offering a higher compute density which can cut down on 50% of space use, 40% of operation and maintenance cost, and 7% of power consumption, thus reducing the overall TCO as compared to 2U 2-socket servers.

The two four-socket servers support up to 48DIMM, DDR4 and the new Intel Optane persistent memory (PMem)200 Series, providing increased memory capacity. Each socket provides a maximum of 4.5TB memory (6*512GB Intel Optane PMem + 6*256GB DDR4 DRAM) to support application scenarios such as in-memory database, data warehousing, SAP HANA and OLAP, to assist customers in the analysis and mining of massive data sets, facilitate decision-making and marketing, and create greater value and benefits for customers. In addition, the non-volatile feature of Intel Optane PMem ensures the integrity of memory data in abnormal circumstances such as sudden power failure.

Load balancing to meet different business needs

Both servers adopt balanced topology design to ensure balanced distribution of business load and improve overall business performance, while reducing cross-CPU service processing delay, providing more efficient data processing solutions for load balancing, deep package inspection (DPI) and other scenarios.

For IO expansion, NF8480M6 and NF8260M6 onboard can support up to six standard PCIe slots to meet more than 90% of the customer's business needs without additional Riser expansion. Additionally, they can be expanded more through Riser to meet different customer needs such as balanced expansion, heterogeneous computing, and high cost performance. Moreover, Inspur four-socket servers provide different disk solutions for customers with different storage requirements to meet high IOPS, high cost performance and mainstream business needs respectively, so as to truly achieve on-demand distribution.

Stability and reliability to ensure key business continuity

It is important for four-socket servers not only to have strong single-machine processing capacity, but also to have good RAS characteristics to ensure the continuity of key businesses. The Inspur four-socket server adopts system fault-tolerant technology, such as thermal switching, intelligent detection and intelligent management, at the chip, component, machine and system levels, which can effectively isolate the impact of hardware and signal errors on the system, reduce the risk of unplanned downtime, and further improve product reliability, so as to ensure business continuity and better data stability.

William Chen, Deputy General Manager of Inspur’s server product line, said that the principle of "extreme scenario design" was adopted in Inspur servers that effectively combines platform and modularization strategies for specific application scenarios of users, so that each product can be efficiently and flexibly expanded according to application requirements, helping enterprises achieve digital transformation.