Intel has officially unveiled its Xeon 6+ processors, codenamed “Clearwater Forest,” boasting an industry-leading core count of up to 288. These fresh processors, showcased at Mobile World Congress (MWC) in Barcelona, represent a significant leap forward in data center technology, targeting demanding workloads in telecom, cloud computing, and edge AI. The launch signals Intel’s commitment to delivering high-performance, energy-efficient solutions for increasingly complex computing needs.
The Xeon 6+ processors are built on Intel’s advanced 18A fabrication process – a 1.8nm-class technology – and are designed to optimize performance for specific applications. Intel aims to position these chips as a key component in the transition towards 6G networks, offering the processing power required for advanced network functions and on-site AI inference. This move comes as network operators increasingly favor server-class CPUs and Systems on Chips (SoCs) for these evolving requirements, according to Intel.
Core Architecture and Design
The “Clearwater Forest” processors utilize a complex chiplet design, combining 12 compute chiplets manufactured on the 18A node with three active base tiles on Intel 3 and two I/O tiles on Intel 7. Each compute tile contains six modules of four “Darkmont” efficiency cores, resulting in 24 E-cores per tile and a maximum of 288 cores per socket. A dual-socket system could therefore approach 576 cores. This modular approach allows Intel to scale performance and optimize power efficiency.
These chips leverage Intel’s Foveros Direct 3D technology for stacking compute tiles onto the base dies, while EMIB bridges facilitate lateral connections between multiple tiles in a 2.5D arrangement. Each “Darkmont” E-core features a 64 KB instruction cache, a wider front end, and a larger out-of-order window to handle more in-flight work. Improvements to execution resources and the number of execution ports aim to boost parallel integer and vector throughput. The package-level last-level cache can exceed 1 gigabyte, reaching approximately 1,152 MB across the entire package.
Key Specifications and Connectivity
The Xeon 6+ processors support existing Xeon server platform sockets and offer 12 memory channels, targeting DDR5-8000 speeds. Connectivity options include 96 PCIe 5.0 lanes and 64 CXL 2.0 lanes, providing ample bandwidth for demanding applications. Intel’s Advanced Matrix Extensions (AMX), QuickAssist Technology (QAT), and Intel vRAN Boost technologies are also integrated into the processors, further enhancing their capabilities for AI and networking workloads.
Target Applications and Performance
Intel is specifically targeting the Xeon 6+ processors for telecom, cloud, and edge AI applications. The processors’ high core count and advanced features produce them well-suited for tasks such as network virtualization, AI-powered analytics, and high-performance computing. A YouTube demonstration showcased the 288-core Clearwater Forest “Xeon 6+” CPU running a 5G workload with over twice the performance of a 144-core Sierra Forest “Xeon 6” processor (YouTube).
The introduction of the Xeon 6+ processors marks a significant milestone for Intel’s 18A process node, which the company considers crucial for its future in the data center market. The combination of high core counts, advanced features, and a new fabrication process positions Intel to compete effectively in the evolving landscape of data center computing.
Looking ahead, Intel’s focus will likely turn to optimizing the 18A process and expanding the availability of the Xeon 6+ processors to a wider range of customers. The success of these processors will be a key indicator of Intel’s ability to regain momentum in the data center market and maintain its position as a leading provider of computing solutions.
What are your thoughts on Intel’s new Xeon 6+ processors? Share your comments below.
