As enterprises modernize their infrastructure to meet the demands of next-generation workloads such as AI, demand for high-performance, high-efficiency computing is increasing across all areas, from data centers to networks to edge to PCs.

Intel announced on the 25th that it is launching the Xeon 6 processor with P-cores to solve these challenges.

“The Xeon 6 family delivers industry-leading CPUs for AI and innovative networking capabilities, while increasing efficiency and reducing total cost of ownership,” said Michelle Johnston, Intel’s interim co-CEO and CEO.

■ AI inference improved by 1.5 times

According to market research firm IDC, as AI adoption accelerates, total spending by enterprises on machine learning and analytics is expected to reach $361 billion (KRW 515.833 trillion) by 2027. Of these, it is expected that up to 153 billion dollars (about 218.622 trillion won) will be invested in generative AI (GenAI).

Intel is working with silicon, software, and solution providers to enable the AI ecosystem and further strengthening Xeon 6 as a CPU for AI systems.

The latest Intel Xeon 6 processors offer performance improvements in data centers and networking. The Intel Xeon 6700/6500 series processors, featuring P-cores, a CPU optimized for the latest data centers, are said to provide an average of 1.4 times the performance over the previous generation and up to 1.5 times the AI inference performance with 1/3 fewer cores compared to the 5th generation AMD EPYC processors.

The Xeon 6 processor delivers performance-per-watt efficiency that enables consolidation of servers that are on average five years old by a ratio of 5:1, and in some use cases up to 10:1, reducing total cost of ownership by up to 68%, it explained.

Intel Xeon 6 for Network and Edge is a high-performance, power-efficient system-on-chip (SoC) that addresses the growing network and edge demands of AI-centric environments by leveraging Intel’s built-in accelerators for virtualized radio access networks (vRAN), media, AI, and network security.

■ Support for building advanced networks

In an era where 5G and AI are revolutionizing connectivity, existing network optimization strategies have their limitations. To maximize the potential of next-generation networks and connectivity, carriers are embracing technologies such as network slicing, AI-based radio controllers, and cloud-native architectures.

The Intel Xeon 6 SoC is up to 4.3x faster than the Intel Xeon D-2899NT for Webroot CSI upload model inference, and up to 3.2x better per-core AI RAN performance over the previous generation with vRAN Boost. The 38-core system can process up to 38 camera streams simultaneously with int8 inference in video edge servers.

Intel also unveiled two new Ethernet controller and network adapter families to meet the growing demands of enterprise, communications, cloud, high-performance computing (HPC), edge and AI applications.

The Intel Ethernet E830 controller and network adapter provide up to 200GbE bandwidth, flexible port configuration, and precision time measurement (PTM), while the Intel Ethernet E610 controller and network adapter support 10GBASE-T connectivity optimized for control plane operations.