This year feels like a rebuilding year for Intel. As AMD’s Ryzen processors continue to get more powerful, and the thermal degradation issues that plagued Intel’s last desktop CPUs, Team Blue launches new tile-based CPUs for mobile CPUs (Lunar Lake) and desktop CPUs (Arrow Lake S) architecture. Lunar Lake arrived to generally positive reviews, with excellent power efficiency, AI hardware support, and reasonable performance, so can the new Core Ultra 200S desktop chip achieve the same goal?
To find out, we tested the Core Ultra 9 285K and Ultra 5 245K, which Intel promises will deliver roughly similar gaming performance and improved content creation performance compared to existing 14th-gen parts, while delivering significantly higher Energy efficiency. Our review focuses on how close the 285K and 245K are on gaming benchmarks – there are 11 games in our emerging new test suite! – But we also tested video production and 3D modeling workloads as well as some basic power analysis.
These new Core 200S processors also come with a new desktop socket, LGA 1851, and a new series of motherboards starting with the Z890 chipset. Both MSI and Asus provided high-end motherboards for testing, which we’ll cover briefly, but the main thing to know is that existing LGA 1700 coolers are compatible with the new motherboards, which have more PCIe bandwidth and support memory overclocking up to DDR5- 9200. Most high-end motherboards require Thunderbolt 4 and WiFi 6E, and are expected to come with WiFi 7 and/or Thunderbolt 5, as well as up to 10 GB of wired networking.
| CPU core | Promote | L3 cache | L2 cache | Maximum power | Suggested retail price | |
|---|---|---|---|---|---|---|
| Super 9 285K | 24(8P+16E) | 5.7GHz | 36MB | 40MB | 250W | $589 |
| Super 7 265K | 20(8P+12E) | 5.5GHz | 30MB | 36MB | 250W | $394 |
| Super 7 265KF | 20(8P+12E) | 5.5GHz | 30MB | 36MB | 250W | $379 |
| Super 5 245K | 14(6P+8E) | 5.2GHz | 24MB | 26MB | 159W | $309 |
| Super 5 245KF | 14(6P+8E) | 5.2GHz | 24MB | 26MB | 159W | $294 |
Before we discuss the results, let us briefly describe the spread in the field. Arrow Lake S currently includes five SKUs, ranging from the top-tier Ultra 9 285K (similar to previous generations of the Core i9 series) to the mid-range Ultra 5 245K (comparable to previous Core i5 offerings). There is the -F variant, which means there is no onboard graphics capabilities, but an NPU is included throughout the stack for accelerating AI workloads.
Interestingly, the 285K has 24 cores, including 8 high-performance P cores and 16 high-efficiency E cores, and only 24 threads because simultaneous multithreading (SMT, aka Hyper-Threading) has been eliminated from the new architecture. This means the 285K is competing with higher-threaded parts in the 32-thread 14900K, as are the Ultra 5 and Ultra 7 models.
Elsewhere, the base P-core (especially the E-core) clock speeds are up across the board compared to 14th gen – although the P-core boost clock is lower, RAM is supported up to DDR5-6400 without overclocking, and the cache is slightly slower Bigger. The main computing module is produced using TSMC’s N3B process rather than Intel’s 14th generation 10nm Intel 7 process, so we generally expect greater power efficiency. However, the memory structure and controller have been moved off-chip, which introduces a latency penalty that may be felt in games.
Our 285K and 245K test systems are based on MSI’s flagship Z890 MEG Ace motherboard, with supplementary testing on ASUS’s beautiful premium ROG Strix Z890-F Gaming WiFi. We used Trident Z5 Neo DDR5-6000 CL30 RAM, NZXT Kraken Elite 360mm AiO and NZXT C1200 Gold ATX 3.1 1200W power supply for testing. Our graphics card is an Nvidia GeForce RTX 4090 Founders Edition, while storage duties are handled by a 2TB Samsung 990 Evo Plus and a 4TB WD Blue SN5000 PCIe 4.0 NVMe SSD.
To put our Intel Core 200S results into context, we also tested the 12th and 14th generation Intel chips on a Gigabyte Aorus Z790 Master with the same RAM, and Ryzen on an ASRock X670E Taichi motherboard with the same RAM 7000 and 9000 chips, as well as Ryzen 5000 chips, tested using an Asus ROG Crosshair 8 Hero with Trident Z Royal DDR4-3600 CL16 RAM.
All tests were conducted on a fresh install of Windows 24H2 with the latest chipset drivers, BIOS updates (1A14U6 for MSI MEG Z890 Ace) and graphics card drivers (Nvidia 565.90). Core isolation is always enabled. “Intel Default Settings” was used for most of our 245K and 285K tests, with “MSI Unlimited Settings” results marked as appropriate. On our 12th generation motherboard, ASUS’s “Intel Default” option is set to “Disabled” instead of “Performance” or “Extreme”. Since all the data is as of the recently released 24H2, the performance here isn’t comparable to our Ryzen 9000X/9950X reviews.
We’ll start with the results of the content creation benchmark. These should map to the biggest increases in single-core and multi-core tasks, and are certainly good indicators of performance for other non-gaming workloads.
The 285K and 245K performed well in Cinebench, a popular synthetic benchmark designed to simulate the workload of 3D modeling and animation program Cinema4D. In our 2024 version of the test, the 285K achieved the highest single-core (144) and multi-core (2386) scores we’ve ever recorded, narrowly beating out the 9950X (138/2235) and 14900K (133/2107). Going to the “MSI Unlimited” power profile slightly improves these results.
The 245K also ranks quite high, with a Cinebench R2024 score of 128/1435, just ahead of the Core i5 14600K (120/1400) and Ryzen 7 9700X (130/1172).
| movie station | 2024 (1 ton) | 2024 (MT) | R20(1T) | R20 (metric tons) |
|---|---|---|---|---|
| Core i5 14600K | 120 | 1400 | Chapter 777 | 9420 |
| Core i7 14700K | 127 | 1987 | 818 | 13614 |
| Core i9 14900K | 133 | 2107 | Chapter 875 | 15297 |
| Super 5 245K | 128 | Chapter 1435 | Chapter 841 | 9864 |
| Super 9 285K | 144 | 2386 | 895 | 16055 |
| Ultra 9 285K (unlimited) | 145 | 2416 | 896 | 16478 |
| Ryzen 5 3600X | 77 | Chapter 578 | Chapter 485 | 3654 |
| Ryzen 7 5800X3D | 95 | 915 | Chapter 546 | 5746 |
| Ryzen 9 5900X | 98 | Chapter 1171 | 610 | 8393 |
| Ryzen 5 7600X | 114 | Chapter 845 | Chapter 744 | 5814 |
| Ryzen 7 7700X | 118 | 1127 | Chapter 758 | 7609 |
| Ryzen 7 7800X3D | 112 | 1074 | Chapter 688 | 6988 |
| Ryzen 9 7900X | 116 | 1605 | Chapter 776 | 11196 |
| Ryzen 9 7950X | 121 | 2004 | Chapter 784 | 14272 |
| Ryzen 5 9600X | 132 | Chapter 935 | 850 | 6358 |
| Ryzen 7 9700X | 130 | Chapter 1172 | Chapter 862 | 7851 |
| Ryzen 9 9900X | 135 | 1784 | Chapter 879 | 12617 |
| Ryzen 9 9950X | 138 | 2235 | Chapter 866 | 15850 |
Handbrake was a slightly more realistic test because we used its video transcoding feature for every video we produced. We’re actually transcoding the Rise of the Tomb Raider Patreon sample here, but the only thing you really need to know is that it’s an 823MB file using the x264 and x265 (HEVC) presets Transcoded to production quality standards (CRF 18) format.
Here, the 9950X tops the list, but the Ultra 9 285K in MSI Unlimited power mode is effectively on par with H264 at 103 fps, closely behind at H265 at 40 fps, and the 9950X trailing behind at 45 fps .
At default power levels, results were more modest – 97fps on the H264 and 38fps on the H265, slightly behind the Ryzen 9 7950X but ahead of the 14900K at 85/35. The 286K consumes significantly less power – 362W max, compared to the 14900K’s 476W – although the 285K’s unlimited mode does consume up to 423W.
| handbrake | H264 (frames/second) | HEVC (frames/second) |
|---|---|---|
| Intel Core i5 14600K | 59.42 | 25.39 |
| Intel Core i7 14700K | 80.26 | 31.07 |
| Intel Core i9 14900K (476W max) | 85.06 | 35.08 |
| Intel Ultra 5 245K (286W max) | 61.05 | 26.88 |
| Intel Ultra 9 285K (362W max) | 97.17 | 38.44 |
| Intel Ultra 9 285K (unlimited, 423W max) | 103.15 | 39.81 |
| Ryzen 5 3600X | 26.66 | 10.80 |
| Ryzen 7 5800X3D | 42.00 | 18.71 |
| Ryzen 9 5900X | 57.59 | 23.83 |
| Ryzen 5 7600X | 41.29 | 18.31 |
| Ryzen 7 7700X | 53.27 | 23.65 |
| Ryzen 7 7800X3D | 49.63 | 21.54 |
| Ryzen 9 7900X | 78.35 | 32.59 |
| Ryzen 9 7950X | 98.58 | 41.68 |
| Ryzen 5 9600X | 42.51 | 19.77 |
| Ryzen 7 9700X | 51.80 | 23.79 |
| Ryzen 9 9900X | 82.96 | 35.33 |
| Ryzen 9 9950X | 103.25 | 44.97 |
For 285K in particular, it’s a promising start, with clear potential for content creators who need 3D rendering and video transcoding at maximum speed while consuming less power than the Core i9 14900K.
Now let’s look at the soy protein we tested, the gaming benchmark. We have 11 games in total, plus additional testing examining power consumption and the impact of DDR5 speed and core isolation.
Intel Core Ultra 9 285K and Ultra 5 245K Analysis
- Introduction, test equipment, and content creation benchmarks [this page]
- Game benchmarks: Dragon’s Dogma 2, Baldur’s Gate 3, Starfield
- Game Benchmarks: Flight Simulator 2020, F1 24, Forza Horizon 5
- Game benchmark: “Counter-Strike 2”, “Cyberpunk 2077”, “Crysis 3 Remastered”
- Game Benchmarks: Far Cry 6, Hitman
- Strength analysis: “Counter-Strike 2”, “Far Cry 6”, “Forza Horizon 5”
- RAM Game Benchmark: “Cyberpunk 2077”, “Far Cry 6”, “Flight Simulator 2020”
- Intel Core Ultra 9 285K and Ultra 5 245K: Digital Foundry’s Conclusion
