Sunday, November 16th 2025

Intel Drops 8-Channel "Diamond Rapids" Xeon 7 in Favour of 16-Channel SKUs

Intel is simplifying its product stack in the upcoming "Diamond Rapids" Xeon 7 family of processors, and the focus is on memory bandwidth. An Intel spokesperson confirmed for ServeTheHome that the company will be exclusively focusing on 16-channel Xeon 7 CPUs, and that the eight-channel designs have been cancelled. "We have removed Diamond Rapids 8CH from our roadmap. We're simplifying the Diamond Rapids platform with a focus on 16-channel processors and extending its benefits down the stack to support a range of unique customers and their use cases." This is a potential massive advantage for Intel customers across the stack as the added memory bandwidth will increase performance across most workloads.

For servers, Diamond Rapids will use "Panther Cove" P-cores and push core density aggressively. It is expected to arrive in the second half of 2026. The Xeon 7 series will feature up to 192 P-cores in the top-end SKU, split across four 48-core tiles. These initial Panther Cove server chips will not include simultaneous multithreading. However, Intel plans to address this with the follow-up "Coral Rapids" family, which will reintroduce Hyper-Threading for datacenter workloads.
Intel has confirmed that its upcoming seventh-generation "Diamond Rapids" Xeon processors will feature the second generation of MRDIMMs (Multiplexer Rank Dual Inline Memory Modules), marking an upgrade from the first-generation MRDIMMs used in the Xeon 6 family. This upgrade will enable Intel to achieve transfer rates of 12,800 MT/s, up from 8,800 MT/s on Xeon 6 with MRDIMMs Gen 1. In addition to this 45% increase in transfer rates, the number of memory channels will rise to 16 from the current 12, providing an additional boost in bandwidth. MRDIMMs work by connecting more memory ranks using a multiplexer, and these modules buffer data and commands, allowing the increased data transfer rate without additional signal degradation.
Source: STH
Add your own comment

8 Comments on Intel Drops 8-Channel "Diamond Rapids" Xeon 7 in Favour of 16-Channel SKUs

#1
dismuter
16 x 64-bit, that's equivalent to a 1024-bit bus. With 12,800 MT/s, if my calculations are correct, that should bring it to a total of 1.638 TB/s of memory bandwidth. That's very close to the RTX 5090's memory bandwidth.
Posted on Reply
#2
kondamin
says who? is this by intel or some other random twitter user?
Posted on Reply
#3
ncrs
kondaminsays who? is this by intel or some other random twitter user?
Says Intel via official communication to ServeTheHome. It's quoted as source for this article.
Posted on Reply
#4
maur0
dismuter16 x 64-bit, that's equivalent to a 1024-bit bus. With 12,800 MT/s, if my calculations are correct, that should bring it to a total of 1.638 TB/s of memory bandwidth. That's very close to the RTX 5090's memory bandwidth.
exactly speed type cache L3 HEHE
Posted on Reply
#5
kondamin
ncrsSays Intel via official communication to ServeTheHome. It's quoted as source for this article.
ah read over that. tnx
Posted on Reply
#6
_roman_
good point
We had multiple sources, enough to publish on, so I sent a note to Intel last night, and they quickly sent the following statement:

We have removed Diamond Rapids 8CH from our roadmap. We’re simplifying the Diamond Rapids platform with a focus on 16 Channel processors and extending its benefits down the stack to support a range of unique customers and their use cases. (Source: Intel Spokesperson to STH)
Posted on Reply
#7
igormp
dismuter16 x 64-bit, that's equivalent to a 1024-bit bus. With 12,800 MT/s, if my calculations are correct, that should bring it to a total of 1.638 TB/s of memory bandwidth. That's very close to the RTX 5090's memory bandwidth.
Math checks out. When we consider capacities of 512GB~8TB (16x 32~512GB sticks), this would be a pretty neat node for larger MoE inference, as well as large scale CFD and in memory databases.

Multiply that by 2~8x depending on how Intel is planning to do multi socket scaling.
Posted on Reply
#8
lambda
Cancellation of the 8ch platform is slightly worrying but I hope we have lower end Xeons with 16ch memory.

Massive progress compared to desktops with puny 2ch memory interfaces.
Posted on Reply
Nov 22nd, 2026 15:49 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts