Average performance improvement of 1.65 times compared to the previous generation
Suitable for environments with increasing workloads, such as AI and cloud.

The era has arrived where everything from artificial intelligence data to data from numerous IoT devices is connected to the cloud. Intel, which has played a pivotal role in the cloud market moving toward a hybrid form beyond public and private, has unveiled its next-generation processor products.

Intel Korea held a launch event for the 'Xeon Scalable Processor' at the Conrad Hotel in Yeouido on the 17th.

Na Seung-ju, Senior Vice President and General Manager of Sales for Intel Data Center Group Korea, said, “With the new mesh architecture core, we have resolved the issue of reduced efficiency in the existing ring architecture as the number of cores increased and improved transmission speeds. In server platforms that use many cores, all cores utilize memory and IO simultaneously.”

Intel Xeon Scalable processors are specially designed for data centers and network infrastructure and demonstrate an average performance improvement of 1.65 times compared to the previous generation. It includes existing workloads, cloud computing, high-performance computing, and artificial intelligence.

By increasing FLOPS (Floating-point Operations per second) by up to 2x with Intel AVX512 and integrated Intel Omni-Path architecture ports, it provides improvements in computational power, I/O flexibility, and memory bandwidth.

Yoon Ha-ryong, a senior researcher at Samsung Electronics’ Memory Business Division who is collaborating with Intel on the memory sector prior to the official launch, stated, “As the number of memory channel slots increased from four to six, memory performance improved and speed increased,” adding, “In terms of memory utilization performance, it improved by 65%, and actual computational performance also improved by 63%.” He further explained that high capacity can be used without performance degradation, allowing the speed required by the industry to be met.

Regarding data security, Executive Director Na stated, “There is almost no overhead in encrypting data.” He explained, "With key protection technology, it is stored separately in hardware rather than in memory or storage accessible to everyone."

In addition, it was stated that with enhanced virtualization performance, it is ideal for data centers that dynamically self-provision resources in on-premises, network, and cloud forms.

Executive Director Na emphasized, “We can build the most agile and scalable artificial intelligence platform using VMD (Volume Management Device), a volume management technology. Inference requires processing a large amount of data in real time, and since Xeon is connected closest to the backbone, it improves training time by 113 times. This can shorten a process that would take days to just a few hours.”

Intel Xeon Scalable processors are receiving widespread support from various companies including Amazon, AT&T, BBVA, Google, Microsoft, Montefiore, Technicolor, and Telefónica.