Introduction

Intel today announced the Core Ultra 2-series Arrow Lake-S desktop processor family. This marks the first true next-generation of desktop processors for the company since its 13th Gen Core Raptor Lake. With Arrow Lake, Intel aims to bring many of the technological innovations it debuted with the Core Ultra 1-series Meteor Lake on the mobile platform, including on-chip AI acceleration, a completely new set of CPU cores, generational power and thermal improvements, and new platform I/O. The Core Ultra 2-series Arrow Lake-S also debuts a new desktop platform, and a new socket—LGA1851, so you'll need a new motherboard. Intel is also debuting the new 800-series desktop chipset, led by Z890, today.



Today's launch of the Core Ultra 2-series follows a similar processor generation rollout strategy as past generations, with Q3-2024 seeing launches and availability of the unlocked K or KF SKUs targeting PC enthusiasts and gamers, alongside the top Z890 chipset that supports processor overclocking, with an expected 2025 ramp of the rest of the series, with more affordable SKUs, and other chipsets in the 800-series.

Intel's goal with Arrow Lake isn't just to introduce new technology across the board—the CPU cores, NPU, and I/O, but also address the weakest aspect of the preceding processor generations by Intel going all the way back to the 10th Gen Core Comet Lake—power draw. Since its switch to TSMC and 7 nm with the Ryzen 3000 Zen 2, AMD has consistently stayed ahead of Intel in energy efficiency, as Intel had to drag on 14 nm for two more generations (Comet Lake and Rocket Lake), and by the time its 10 nm process (later rebranded Intel 7) arrived, AMD had already developed the efficient Zen 3 on 7 nm, and was preparing to transition to 5 nm and Zen 4. Ryzen is now on 4 nm with Zen 5. Intel held onto its own Intel 4 node (7 nm) with the compute tile of Meteor Lake, but decided on TSMC for Ultra 2-series. Arrow Lake is the first high-performance client Intel architecture to use TSMC for the CPU cores, with its Compute tile being built on TSMC 3 nm—a more advanced node than the 4 nm that Zen 5 uses.



The design philosophy with Arrow Lake hence is to reestablish Intel as a leader in efficient client desktop processors. This goal would be achieved even if the generational performance or IPC gains are modest, as long as there is a significant drop in power draw, because then it opens up other possibilities for users, such as overclocking—something Intel processors tend to be naturally good at.

In this article, we preview what's in store for you when the Core Ultra 2-series Arrow Lake desktop processor family hits the shelves later this month.

The Core Ultra 2-series Arrow Lake Lineup

Intel is launching five processor models today, across three brand extensions—Core Ultra 9, Core Ultra 7, and Core Ultra 5. Intel switched to this nomenclature with its Core Ultra Meteor Lake series on the mobile platform, and they succeed Core i9, Core i7, and Core i5, respectively. The Core Ultra 9 has just one SKU, the flagship Core Ultra 9 285K. There is no variant of this chip without integrated graphics—there's no "285KF" to succeed the i9-14900KF.

The Core Ultra 9 285K maxes out the Arrow Lake-S silicon, with a full 8P+16E core configuration. That's 8 "Lion Cove" performance cores, 16 "Skymont" efficiency cores, and the chip's full 36 MB L3 cache available on the Compute tile. The E-cores come with a base frequency of 3.20 GHz, and boost up to 4.60 GHz. The P-cores have a 3.70 GHz base frequency, a 5.50 GHz Turbo Boost frequency, a 5.60 GHz Turbo Boost Max 3.0 frequency (spreads across up to four cores), and a Thermal Velocity Boost (TVB) frequency of 5.70 GHz (up to two cores, assuming temperature headroom).

Next up, is the Core Ultra 7 265K, and its variant without integrated graphics, the Core Ultra 7 265KF. This comes with an 8P+12E configuration, similar to its predecessor, the Core i7-14700K. Interestingly, the shared L3 cache size has generationally reduced to 30 MB, from the 33 MB that the i7-14700K has. The 265K comes with an E-core base frequency of 3.30 GHz and boost frequency of 4.60 GHz, a P-core base frequency of 3.90 GHz, Turbo Boost frequency of 5.40 GHz, and Turbo Boost Max 3.0 frequency of 5.50 GHz. There's no TVB for this segment.

Lastly, there is the Core Ultra 5 245K, and its iGPU-disabled variant, the Core Ultra 5 265KF. This retains the 6P+8E core configuration of its past two generations of predecessors, and shared 24 MB of L3 cache among the cores. The 265K has the highest E-core base frequency of the series, at 3.60 GHz, but with the same 4.60 GHz maximum E-core boost frequency. It also has the highest P-core base frequency at 4.20 GHz, which boosts up to 5.20 GHz. There's no Turbo Boost Max 3.0 or TVB.

The P-cores on all five models come with 3 MB of dedicated L2 cache. This is higher than the 2.5 MB per core that the "Lion Cove" P-cores come with on Core Ultra 200V Lunar Lake mobile processors. The "Skymont" E-cores are arranged in clusters of 4 cores, each. Each cluster shares a 4 MB L2 cache among the four cores. We will learn more about the two CPU core types in the following pages.

On all SKUs with integrated graphics, the iGPU is branded simply "Intel Graphics." This iGPU is based on the Xe-LPG graphics architecture (same one powering the iGPU of Meteor Lake). It comes with 4 Xe cores or 64 execution units (EU), worth 512 unified shaders. All three processor models with this iGPU come with a GPU base frequency of 300 MHz. The 285K and 265K offer up to 2.00 GHz GPU boost frequency, while the 245K tops out at 1.90 GHz.

The integrated NPU 3 unit is available on all five processor models, it features two Gen 3 NCEs (neural compute engines), and offers a peak performance of 13 TOPS, which falls well short of the 40 TOPS required by Microsoft Copilot+ AI PC certification, and compares closer to the NPU 3 found in Meteor Lake processors.

Intel is sticking with its power ratings system it started with the 12th Gen Core "Alder Lake." All five processor models come with 125 W processor base power. The 285K, 265K, and 265KF come with 250 W maximum turbo power. The 245K and 245KF, on the other hand, have their maximum turbo power value reduced to 159 W. This is interesting, as this value used to be 181 W for its Core i5 predecessors, such as the i5-14600K and i5-13600K. There's more to report on power, so stay tuned.


As for pricing, the Core Ultra 9 285K is positioned on the top, with a suggested price of USD $589. This is followed by the Core Ultra 7 265K at $394. The 265KF can be had for $379. The Core Ultra 5 245K goes for $309, followed by the 245KF at $294. These are the exact same price points Intel launched the previous generation i9-14900K, i7-14700K/KF, and the i5-14600K/KF at, respectively.

All processors being announced today go on sale from October 23, 2024.

The Arrow Lake Microarchitecture

Intel developed the Arrow Lake processor microarchitecture to address two of its biggest client processor market segments, the desktop, and conventional notebooks. For ultraportable notebooks without discrete graphics, the company developed a grounds-up new silicon under the Lunar Lake microarchitecture. Lunar Lake shares many IP blocks with Arrow Lake, especially the CPU cores, but there are several differences. Intel uses the form-factor designation "S" to refer to mainstream desktops, and hence the "Arrow Lake-S" silicon powers all five processor SKUs being announced today. The company plans to launch the "Arrow Lake-H" processor for mainstream notebooks, and the "Arrow Lake-HX" processor for enthusiast notebooks, in Q1-2025. There are no processor model announcements for the H and HX variants today.


With "Arrow Lake," Intel retains the disaggregated tile-based processor design that it embarked upon with "Meteor Lake" in the client segment. The governing idea here is that you needn't built a large monolithic chip based on the latest foundry node, but rather identify specific IP blocks that benefit the most from the node, such as the CPU cores and the iGPU, and build just those components on the latest nodes, as breakout tiles. The rest of the processor that deals mainly with platform I/O interfaces, can be built on a tile with a slightly older foundry node, thereby letting Intel maximize its yields on the latest foundry node, since the tiles it builds on these nodes are smaller.


There are five types of tiles making up "Arrow Lake." It begins with the Foveros base tile. This is a silicon interposer that sits on the fiberglass package substrates, and seats the other tiles on top. An interposer facilitates high-density microscopic wiring between tiles or chiplets stacked on top of it, which would otherwise not be achievable on the fiberglass substrate. This is the key difference between Intel's tile-based processor, and AMD's chiplet-based ones, which rely on the package substrate to connect the CPU complex dies (CCDs) to the I/O die (cIOD). The perks of this interposer include closer physical proximity of the tiles, benefitting latency, and lower power needed to move data around between the tiles.