Intel unveils Xeon 6+ Clearwater Forest with 288 efficiency cores on 18A

Intel is introducing its most core-dense Xeon 6+ processor, Clearwater Forest, combining 12 compute chiplets on the Intel 18A node with advanced 2.5D and 3D packaging to deliver up to 288 efficiency cores per socket. The design targets high parallel workloads with expanded cache, high memory bandwidth, and extensive PCIe 5.0 and CXL 2.0 connectivity.

Intel is showcasing its most core-dense Xeon 6+ processor, codenamed ‘Clearwater Forest,’ at MWC in Barcelona, positioning it as one of its most complex chiplet server designs to date. The package combines 12 compute chiplets manufactured on an Intel 18A node with three active base tiles on Intel 3 and two I/O tiles on Intel 7 to maximize scalability within the existing Xeon server platform socket. In this configuration, each compute tile contains six modules of four ‘Darkmont’ efficiency cores, providing 24 E-cores per tile and a maximum of 288 ‘Darkmont’ E-cores on a single socket, while a two-socket system, therefore, approaches 576 cores.

The architecture relies on advanced packaging to interconnect and stack its many components, with clusters linked by a high-bandwidth on-chip fabric for data movement across tiles. Die are stacked using Foveros Direct 3D technology to vertically integrate logic, while EMIB links connect multiple tiles in a 2.5D arrangement to balance bandwidth, latency, and manufacturing complexity. This approach aims to deliver very high core density while preserving flexibility in process node selection for different functional blocks, as compute, base, and I/O tiles each use distinct Intel process technologies.

At the core level, each ‘Darkmont’ E-core comes with a 64 KB instruction cache, a wider front end, and a larger out-of-order window to sustain more in-flight work, while execution resources and the number of execution ports have been increased to improve parallel integer and vector throughput. Physically, clusters are grouped in four-core units sharing about 4 MB of L2 cache per group, and the package-level last-level cache can exceed a gigabyte, with about 1,152 MB of combined last-level cache across the package. ‘Clearwater Forest’ supports the existing Xeon server platform socket, 12 memory channels, and broad I/O, including 96 PCIe 5.0 lanes and 64 CXL 2.0 lanes, and memory speed targets push toward DDR5-8000 to feed the large number of efficiency cores.

58

Impact Score

TSMC N2 node capacity booked through 2027 as customers face long delays

TSMC’s next generation N2 process node is already heavily booked through the end of 2027, forcing customers to lock in allocation early or face extended delays and higher costs. The squeeze is reshaping product roadmaps across the semiconductor industry, including for TSMC’s largest customer, NVIDIA.

OpenAI strikes Pentagon deal that sidelines anthropic’s harder line

OpenAI has reached a rushed agreement to let the US military use its Artificial Intelligence models in classified settings, relying on existing law and internal safety rules rather than hard contractual bans on controversial uses. The deal contrasts sharply with Anthropic’s failed attempt to impose explicit red lines on autonomous weapons and mass surveillance, which triggered a fierce backlash from the Pentagon.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.