Intel Announces Core Ultra 200S Series Desktop Processors

Low Boon Shen
By Low Boon Shen 9 Min Read

At long last – Intel has finally unveiled its next generation of desktop processors, codenamed Arrow Lake, officially as the Core Ultra 200S series. This new lineup will adopt the Core Ultra naming for the first time in the consumer desktop segment, and with it, the chipmaker has set different targets this time around when designing this new architecture.

Performance-Per-Watt Is The New Game In Town

From the get go, Intel says the goals of Arrow Lake is not to chase outright performance, but rather, performance-per-watt, something that the chipmaker has faced criticism for years due to its power hungry designs, which also indirectly caused the fiasco affecting several high-end Raptor Lake models with degradation issues.

If you’re looking for better gaming performance, it’s now time to keep your expectations low. Intel has explicitly stated that they intend Arrow Lake to reduce power usage “while maintaining gaming performance,” but you can still expect a small boost in multi-core performance, despite the absence of Hyper-threading feature that has been a mainstay in Intel processors since 2002.

CPU Architecture Overview

The Arrow Lake architecture is built using Intel’s new Foveros Advanced 3D packaging technology, which connects several ’tiles’ into a single silicon. Similar to Lunar Lake, the silicon is split into CPU (Compute), GPU, SOC, I/O, and filler tiles, sitting on top of the base tile.

Arrow Lake uses two new core architectures, and they are Lion Cove (P-Core), and Skymont (E-Core), both saw their debut in the Lunar Lake processors launched just last month. Compared to the Raptor Cove and Gracemont – both featured in the prior desktop generation – the P-Core is 9% faster in IPC (instructions per clock), while the E-Core gets a big 32% boost in the same metric.

Despite the lesser thread count, Intel’s first-party benchmarks claim that its overall multi-core performance (with the Core Ultra 9 285K) is faster than the AMD Ryzen 9 9950X; while the single-thread is comparatively closer to each other, the Core Ultra chip is shown as the slightly faster one. As for comparing against AMD’s top cache-laden chip, the Ryzen 9 7950X3D, gaming benchmarks showed a back-and-forth between the two, while it’s mostly advantage Intel for content creation workloads.

Gaming, Power Draw & Thermals

For gaming benchmarks, the Core Ultra 9 285K’s performance is largely a tie against its predecessor, the Core i9-14900K, as well as its top rival, the AMD Ryzen 9 9950X. However, due to architectural differences the result can vary quite a bit – edge cases on the charts above indicate that the new chip can lead as much as 28% over AMD, but lose up to 13% in performance on the other end of the charts. (Note that Intel Application Optimization, or APO for short, is enabled in all cases.)

One thing Intel is keen to point out this time around is the power savings achieved while maintaining the same level of performance. In light workloads, the E-cores contributes to more than 40% power savings – this should be good news for commercial applications with concerns for energy bills; besides that, Intel also claims that the Core Ultra 9 285K is capable of achieving the full performance of Core i9-14900K with just half the power.

In some parts of the world, energy prices is especially expensive (Malaysia’s energy pricing is a lot cheaper comparatively), so even the gamers will have to consider the power draw of their gaming systems before the bills get too costly. To that end, Intel claims that Arrow Lake is capable of maintaining the same level of gaming performance even if you limit the TDP to just 125W. When paired with NVIDIA’s top RTX 4090 GPU, Intel also claims that the Core Ultra processor can contribute up to 165W in total system power savings, with a geomean of 73 watts.

As a result of reduced power draw, temperatures goes down as well. The chipmaker says the new chip can net an average of 13°C in temperature reduction with a 360mm AIO cooler; it’s not particularly shy about admitting a small loss of performance in exchange for huge temperature and power draw reductions either.

GPU & NPU

Moving on from the CPU side of things, the new Core Ultra lineup will also feature the Xe-LPG based GPU (not the new Xe2-LPG found on Lunar Lake), and for the first time, NPU (neural processing unit) on a desktop processor. Granted, both the GPU and NPU are not as powerful as their laptop counterparts, but they are still capable of enabling better codec support and some lightweight AI functionalities.

Platform & Overclocking

Intel’s new 800 Series motherboard chipset will be the one accompanying the launch of Arrow Lake processors, and the platform combines a total of 24 CPU + 24 chipset PCIe lanes – 20 from the CPU is PCIe 5.0. This, on paper, should allow up to three active PCIe 5.0 SSDs installed on the system when coupled with a PCIe 5.0-compliant GPU running at x8 mode.

The processor themselves will come with integrated Thunderbolt 4 support (along with Thunderbolt Share feature set) and Wi-Fi 6E, though motherboard makers can optionally include up to four discrete Thunderbolt 5 ports, along with Wi-Fi 7 and 2.5GbE LAN connectivity; however, we’re already seeing motherboard makers including even faster 5GbE LAN ports into mainstream high-end models by this point.

For overclockers, Arrow Lake will feature finer granularity with stepping as small as 16.6MHz, while tile-to-tile connections, SoC and BCLK can now be independently overclocked; DLVR allows for per-core or per-cluster voltage control, though this can be bypassed under extreme overclocking conditions. On that note, low-temperature overvolting allows voltage to be bypassed, which may also benefit extreme overclockers.

In terms of memory support, while the modules remain maxed out at 192GB (48GB per module), the official support is now raised to DDR5-6400, and support for CUDIMM (clocked unbuffered DIMM) and CSODIMM (clocked small outline DIMM) should enable speeds beyond DDR5-8000. This speed is considered as the “sweet spot” so long as the memory controller is set on Gear 2, though Intel hinted that more clock speeds can still be extracted, stating that Arrow Lake is “resilient”.

Intel Core Ultra 200S Series: Availability

If you’re keen on getting one of these new chips for yourself, save October 24th on your calendars, as all five Arrow Lake processors – Core Ultra 9 258K, Core Ultra 7 265K/KF, Core Ultra 5 245K/KF – will be available later this month starting at $294 (~RM1,260).

One More Thing

Finally, Intel is also announcing that Arrow Lake will be coming to high performance laptops next year in both H and HX-series, and while it’s not explicitly stated – it’s safe to say more details will be revealed in the coming CES in January 2025.

Pokdepinion: Personally, a tad bit disappointed that gaming performance has largely stayed still, but I think the change of development direction to power efficiency will pay dividends in the future.

Share This Article
Leave a comment

Leave a Reply

Your email address will not be published. Required fields are marked *