Skip to content

Intel unveils Arrow Lake CPUs with increased power efficiency

  • by
  • 4 min read

Source: Intel

Intel has unveiled a new range of desktop processors, Core Ultra 200S, codenamed Arrow Lake. This marks a significant architectural shift for the company, with the introduction of chiplet-based design and a focus on power efficiency.

The new lineup, launching October 24th, offers improvements in multi-core performance and integrated graphics while reducing power consumption by 33% compared to its predecessors. However, Arrow Lake’s AI capabilities fall short of expectations, leaving some key Windows 11 features out of reach, reports Ars Technica.

The new Core Ultra 200S will cost $589 for Core Ultra 9 285K, $394 for Core Ultra 7 265K, and $309 for Core Ultra 5 245K. Furthermore, Intel is also launching KF variants of Ultra 7 265KF at $379 and Ultra 5 245KF at $294 without a built-in GPU.

Intel’s Arrow Lake processors bring many of the same innovations previously seen in its laptop-focused Core Ultra 100- and 200-series chips. The shift to chiplet-based architecture is the most significant change, where multiple silicon tiles are connected via Intel’s Foveros packaging technology.

These include GPU, system-on-chip (SoC), and I/O tiles. Intel uses manufacturing processes from TSMC for most components, while the base tile is produced in-house.

A standout feature is a significant reduction in power consumption, a welcome change from the energy-hungry 13th- and 14th-gen processors. Arrow Lake processors deliver a modest 10% boost in multi-core performance over their predecessors, thanks to Intel’s Lion Cove architecture for performance (P) cores and Skymount architecture for efficiency (E) cores.

Notably, the E-cores have seen a significant uplift, offering up to 72% faster performance in single-core floating-point workloads and 55% faster than multi-threaded floating-point workloads.

One of the key changes in the new processors is Intel’s decision to eliminate Hyper-Threading from the P-cores, which lowers the overall thread count compared to previous generations. Intel has emphasised the increased efficiency of its E-cores, which compensates for this, with the space once reserved for Hyper-Threading now better utilised.

While single-core performance gains are less dramatic, multi-threaded applications are expected to benefit from these architectural adjustments.

In gaming, where single-core performance traditionally plays a major role, Intel’s data suggests that the Arrow Lake chips offer similar frame rates to 14th-gen processors, but with a notable power-saving advantage — using 73 watts less power on average and running cooler by 13 degrees Celsius.

Despite Intel’s decision to include a neural processing unit (NPU) in the Core Ultra 200S processors, these chips won’t be able to support Microsoft’s Windows 11 Copilot+ features. Arrow Lake’s NPU offers just 13 trillions of operations per second (TOPS), far below Microsoft’s minimum requirement of 40 TOPS.

As a result, desktop users will miss out on advanced AI features such as enhanced image generation and webcam effects, leaving Intel playing catch-up with AMD, whose Ryzen 8000G series has set the pace for AI in desktops.

Intel has also improved the integrated GPU, doubling its performance compared to the UHD 770 found in earlier Core chips. While the new GPU, based on Intel’s older Alchemist architecture, supports ray tracing and AV1 video codec encoding, it remains unsuitable for high-end gaming, catering more to low-end gaming and additional display outputs.

Another major change for Arrow Lake is the introduction of the LGA1851 socket, replacing the LGA1700 used by Intel’s 12th-, 13th-, and 14th-gen processors. This change means that users upgrading to Arrow Lake will need a new 800-series motherboard, with manufacturers already promoting the high-end Z890 chipset.

This change means that users upgrading to Arrow Lake will need a new 800-series motherboard. Manufacturers are already promoting the high-end Z80 chipset, which supports up to 24 PCIe 4.0 lanes, Thunderbolt 4 ports, and advanced connectivity options such as WiFi 6E and Bluetooth 5.3.

Intel has also improved memory with Core Ultra 200S series and Z890 motherboards, offering support up to DDR5-6400, up to 48 GB per DIMM, and up to 192 GB max capacity.

The chips will also support Secure Core compliance and include three built-in hardware engines for security.

In the News: Tesla unveils Cybercab, Robovan, and Optimus humanoid robots

Kumar Hemant

Kumar Hemant

Deputy Editor at Candid.Technology. Hemant writes at the intersection of tech and culture and has a keen interest in science, social issues and international relations. You can contact him here: kumarhemant@pm.me

>