Intel bounces back
with new AI chips

Intel yesterday (24 February) launched a range of Xeon 6 processors with Performance-cores (P-cores), promising industry-leading performance for the broadest set of data centre and network infrastructure workloads, as it bounced back from a series of technology and management setbacks.

“The Xeon 6 family delivers the industry’s best CPU for AI and groundbreaking features for networking, while simultaneously driving efficiency and bringing down the total cost of ownership,” said Michelle Johnston Holthaus, interim co-CEO of Intel and CEO of Intel Products.

Early reactions suggest the chips have restored Intel’s competitiveness in the server CPU market, particularly with regard to AI and data centre workloads. However, concern has been expressed that the processors have higher power consumption compared to competitors, potentially leading to increased operational costs. Nevertheless, competitive pricing of the new chips are likely to balance out such concerns.

Intel provided the following information on the Xeon 6 Processors

Intel’s latest Xeon 6 processors feature significant advancements in both data centre and networking portfolios.

·     The Intel Xeon 6700/6500 series processor with P-cores is the ideal CPU for modern data centres, offering the perfect balance between performance and energy efficiency. Delivering an average of 1.4x better performance than the previous generation across a wide range of enterprise workloads, Xeon 6 is also the foundational central processing unit (CPU) for AI systems, pairing exceptionally well with a GPU as a host node CPU. When compared to 5th Generation AMD EPYC processors, Xeon 6 provides up to 1.5x better performance in AI inference on chip using one-third fewer cores. Xeon 6 processors also enable substantial performance-per-watt efficiency, allowing for 5:1 consolidation of a 5-year-old server on average, with potential for up to 10:1 in certain use cases, resulting in up to 68% savings in total cost of ownership (TCO).

·     The Intel Xeon 6 for network and edge is a system-on-chip (SoC) designed for high performance and power efficiency. It leverages Intel’s built-in accelerators for virtualised radio access networks (vRAN), media, AI and network security, addressing the growing demand for network and edge solutions in an AI-driven world. Xeon 6 SoCs deliver up to 2.4x the RAN capacity and a 70% improvement in performance-per-watt compared to previous generations thanks to Intel vRAN Boost. Additionally, Xeon 6 is the industry’s first server SoC with a built-in media accelerator — the Intel Media Transcode Accelerator — enabling up to 14x performance per watt gain versus Intel Xeon 6538N.

The world’sbest CPU for AI

As AI adoption continues to accelerate, organisations are projected to spend up to $153-billion on generative AI (GenAI) by 2027, with total spending for machine learning and analytics reaching $361-billion, according to IDC. Intel Xeon 6 is optimised to capture a significant share of this growing market, delivering leadership performance in traditional machine learning, smaller GenAI models and GPU-accelerated workloads in a host CPU capacity. Intel is collaborating with silicon, software and solution providers to enable the AI ecosystem, reinforcing Xeon 6 as the go-to CPU for AI systems.

With 5G and AI poised to transform connectivity, traditional network optimisation strategies are no longer sufficient. To unlock the full potential of next-gen networks and connectivity, telecom operators are adopting technologies like network slicing, AI-powered radio controllers and cloud-native architectures. By leveraging Intel’s unified Xeon platform, operators can dynamically optimise workloads, reduce costs and build scalable, flexible networks that adapt in real time to changing customer demands, traffic patterns and market shifts.

Key performance highlights of the Intel Xeon 6 SoC include:

• Webroot CSI upload model inference is up to 4.3x faster than Intel Xeon D-2899NT.
• AI RAN performance per core is improved by up to 3.2x compared with the previous generation with vRAN Boost.
• A 38-core system supports int8 inferencing of up to 38 simultaneous camera streams on a video edge server.

New advanced Ethernet solutions

Intel also unveiled two new Ethernet controller and network adapter product lines to address the growing demands of enterprise, telecommunications, cloud, high performance computing (HPC), edge and AI applications. Initial availability includes dual-port 25GbE PCIe and OCP 3.0-compliant adapters, with additional configurations expected this year.

• The Intel Ethernet E830 Controllers and Network Adapters deliver up to 200GbE bandwidth, flexible port configurations and advanced precision time capabilities, including Precision Time Measurement (PTM). These adapters are optimised for high-density virtualised workloads, offering robust security features and performance.
• The Intel Ethernet E610 Controllers and Network Adapters provide 10GBASE-T connectivity optimised for control plane operations. The 610 series offers outstanding power efficiency, advanced manageability and robust security features that simplify network administration and ensure maximum network integrity.
• The combination of Intel Xeon 6 processors and high-performance Ethernet connectivity forms a powerful foundation for businesses to accelerate innovation and gain a competitive advantage.