Mikeserv Support Forum

Arrow Lake specs

Core Ultra 9 models — the 285K and 275 (and possibly one more) — will have eight performance cores, 16 efficiency cores, and four Xe graphics cores. Clock speeds will vary among them, but they should all have the same 24 CPU core configuration. The clock speeds may well be lower than last generation, though, with rumors suggesting they’ll be scaled back by several hundred MHz. Potentially, the top chip will have a maximum boost clock of 5.7GHz, although recent speculation points to 5.5GHz. E core clocks could be on the rise, though, which may compensate.

It’s not clear if this is purely from a design perspective or if it’s somewhat of a response to the recent controversy surrounding 13900K and 14900K instability at higher clock speeds. Intel is also said to be dialing back the TDP to about 80% of Raptor Lake, which could help.

Lower clock speeds aren’t great. It’s worth remembering, though, that clock speed is only a component in how fast a chip is. AMD’s processors have had lower clock speeds than Intel’s for many years, and yet it’s been hotly competitive just about everywhere.

Arrow Lake will be built on a new LGA 1851 socket design, with more pins than 12th- and 13th-generation designs on the LGA 1700 socket. That means you’ll need a new motherboard for Arrow Lake processors and won’t be able to upgrade from 12th-, 13th-, or 14th-generation PCs directly. However, it will be the same physical size as LGA 1700, so coolers should be compatible.

That new socket brings a new generation of motherboard chipsets, too, as well as mandated DDR5 — no more DDR4 on Intel boards with this generation. High-end Z890 boards are also said to support the CAMM2 memory standard, a possible replacement for SO-DIMM. Memory speed support will rise again, and there are rumors of support for Thunderbolt 5, though those haven’t been confirmed. The flagship PCIe technology will remain PCIe 5, too.

One big change that could be coming with Arrow Lake is that Intel will ditch its long-time simultaneous multithreading technology: hyperthreading. It did this with sub-top-tier models in the 9th generation, but this will reportedly be a complete top-to-bottom removal of the feature. The idea is to allow the performance cores to stretch their legs on demanding tasks, while the multithreading capabilities hyperthreading typically enables would be handled by the new E and LP-E cores.

We initially thought that Intel would use its own 20A node for Arrow Lake. However, it revealed during IFA 2024 that it’s shifting gears toward developing a smaller node: 18A. Arrow Lake is therefore said to be manufactured by “external partners,” which most likely means Intel’s rival fab TSMC.

Bartlett Lake

There have also been rumors that Intel will debut a new Bartlett Lake design for low-end processors alongside Arrow Lake in 2024, utilizing the older Intel 4 3nm design. These may be designed to offer competition for AMD’s aging, but certainly not unpopular, Ryzen 5000 generation, which continues to see support with new designs like the 5700X3D, and new 5000 GT series.

Barlett Lake will reportedly based on updated Raptor Lake silicon designs, so it should be compatible with existing LGA 1700 platforms. It should have DDR4 memory, too, potentially making it a cheaper upgrade path for Intel fans.

While it might seem odd for Intel to utilize an older process node and split its next-gen lineup in two, this is to take the pressure off the fabrication of Intel’s 20A silicon, which will be cutting edge at the time of release and in limited supply. Using an older design for more affordable chips should improve availability and pricing come launch time.