After the disappointing launch of Intel Core Ultra 200S (test), Intel has refined the platform and promises more performance for Arrow Lake-S through microcode, Windows and game updates. ComputerBase measured and ran Microcode 110, 112, 113 and 114 (“January patch”) with current Windows in the benchmark test.
Model AMD Ryzen 9000?
After AMD Ryzen 9000 (test) only delivered an average picture at the launch in August and the following weeks, the picture improved significantly in the following weeks – the test two months later at the launch of the new 800 motherboards for socket AM5 confirmed this also with hard facts. The gaming performance patch for Windows 11 is the crowning achievement of AMD's work in the BIOS.
With Intel Arrow Lake aka Core Ultra 200S (test), things didn't go well at the start either: presented in mid-October, the picture in the test was good in some areas, but unexpectedly weak in others. In this case, too, the platform will be looked at again after two months and in this case too, BIOS (microcode) and Windows updates are said to have unlocked a lot of performance.
Intel Arrow Lake-S: New microcodes and updates
The main focus of the test is on the four microcodes that have been in circulation since the start in mid-October:
- Microcode 110 (the launch version)
- Microcode 112
- Microcode 113
- Microcode 114 (officially announced for January 2025)
There are also some small and large updates when playing (Cyberpunk 2077 was in the media a few days ago) as well as the latest December edition of Windows 11 24H2.
The status “January 2025” is already being tested
This means that the test also looked at what Intel officially announced with MCU 114 (microcode update 114) for January, when non-K-Core Ultra 200S CPUs and cheaper mainboards with B860 chipsets are available market should come.
The final solution planned for the start of the new year is a BIOS with MCU 114 and Intel firmware 1854 v2.2. The latter looks like a real last-minute update, as board partners don't have it yet, although there was recently a big jump from version 1753 v4 to 1827.
But the mainboard manufacturers also have a solution for this: Intel ME Build 1854 is offered separately as a download. Whether this is already v2.2 remains unclear, but according to analyzes of the last ME firmware, there were never multiple variants of a specific version number.
This corresponds to 99.9 percent of the final product, there is only a slight chance that there could be a bit more performance, but realistically you should say goodbye to it. On the other hand, Windows 11 24H2 is already further along than Intel stated: Build 2605 is already available today.
That's what Intel promised
“Single-digit performance gains“Should be possible on this MCU version Windows build basis, Intel reported only two days ago as the result of an “error analysis” on Arrow Lake.
- Arrow Lake Performance Update: Four issues fixed, “more” gaming performance starting in January
Although Intel also spoke of some significantly larger increases, they referred to tangible problems at the start:
- The “Balanced” energy saving profile costs a lot of power
- Intel APO was not active in supported gaming
Neither had any effect on ComputerBase's benchmarks: The editorial team tested in “Highest Performance” and none of the games in the course offered APO support.
So far there is no “miracle update” or “over-microcode”. Primarily, these updates are aimed at stability and compatibility. There are really a lot of little things that have been corrected, both by Intel and especially by Microsoft via Windows updates.
Current benchmarks Core Ultra 9 285K vs. Core i9-14900K
The same test setup as the first test in October was used for the benchmarks in this test. The performance increase achieved by Intel was determined incrementally.
The different test levels include four microcodes (MCU 110, 112, 113 and 114) paired with Windows updates (Build 2033 in the original, Build 2314 as an October update and Build 2605 as a December update). Various game updates were also installed and updated to the status in mid-December – an influence from this aspect alone cannot be ruled out for any of the games.
In order to reflect the latest performance of an Intel Core i9 285K compared to the Intel Core i9-14900K, this CPU also had to be updated to the latest BIOS and a Windows 24H2 Build 2605, and of course all the latest game patches had to be used. And the BIOS has also been updated: from version 1602 to version 1801 at the beginning of December.
“Balanced” no longer eats up performance
The first finding in the test: The “Balanced” energy saving profile no longer consumes 20 percent of power – measured and confirmed.
Benchmark tests in gaming
The previous six percent lead in FPS and four percent lead in frame times for the Intel Core i9-14900K now remains five to three percent ahead in the updated test setup. The Intel Core Ultra 9 285K has also improved somewhat, but still lags behind its predecessor.
That's not nothing, but it's still not enough to dethrone the Intel Core i9-14900K in our own company. At least Intel's pre-launch statement that both processors should offer similar gaming performance is closer to the truth. But even today there are still titles in which an Intel Core i9-14900K has a 20 percent advantage in the worst case scenario for the newcomer. Curiously, the latest Windows is also a little slower here and there than the older one. All good things didn't exist for Arrow Lake yet.
The power is still provided with almost half the consumption. The jump in efficiency in gaming remains more than huge with Arrow Lake compared to Raptor Lake.
Benchmark testing in applications
There is also a small increase in performance in the tested applications: both with Intel Core Ultra 9 and with Intel Core i9. Here you can play your cards with the latest Windows and slightly more up-to-date software. Two tests become more reproducible, while others do not react at all.
The test journey through the microcodes shows that the performance even with build 2033 in October in the multi-core area corresponded to what is also delivered in today's new update package. The fact that performance has been achieved, especially since Microcode Update 113 together with the December update of Windows 24H2 to Build 2605, can be seen in single-core tests.
In single-core scenarios, an Arrow Lake processor can gain up to six percent on average. This is considerable, but is also largely due to Windows 11 – because an Intel Core i9 is also four percent higher than in October. In the WebXPRT browser benchmark, the CPUs all increased significantly, in the extreme case there was a 17 percent difference in the 285K from the very old microcode and old Windows 11 to the latest version – the browser was the same in each case.
However, there is less happening in full load multi-core use. There are power limits here, the CPU is used to its maximum. There are no major shifts here; in the end, Windows is the influencing factor here too.
Interim conclusion in applications: The distance between the Intel Core i9-14900K and the Intel Core Ultra 9 285K changes very little on average and often nothing in detail, but positive outliers do make a difference in the end: in single-core scenarios there are now seven instead of five percent advantage for the newcomer 285K, in the multi-core area it is twelve instead of ten percent.
Conclusion
Intel Core Ultra 200S aka Arrow Lake (test) has become faster thanks to microcode and Windows updates since its launch in October at the end of the year. However, the big bang that AMD experienced with Ryzen 9000 after two months of BIOS and Windows updates has not yet happened.
Only those who tested with the “wrong” energy profile in Windows at the start, or who heavily relied on games with incorrect APO support at the time, will find a completely different CPU today. But that wasn't the case with ComputerBase. The progress is also visible in this test, but the weakness remains, especially in the gaming environment: An Intel Core i9-14900K is still faster than a Core Ultra 9 285K, even if only slightly.
In applications that don't fully utilize the CPU, there is slightly more noticeable progress due to optimized Turbo mode and proper scheduling. Not only the update by Intel helps, but also the work on the BIOS by the partners and Microsoft with the latest Windows 11 update (version 2605). In applications in which the load is always on all cores, hardly anything changes.
The bottom line is that the Intel Core Ultra 9 285K is now pretty much on the same level as an AMD Ryzen 9 9950X across all areas. In applications it is slightly ahead and slightly behind in games.
The overall Arrow Lake package is now a little more rounded and the development suggests that Intel can still leverage performance potential in the future, and Windows 11 also has to do its part. But the steps are likely to remain small. The company's own error analysis promises significant double-digit increases in some cases, but this always referred to cases in which things didn't work at all at start due to a faulty standard energy saving profile in Windows or because of faulty APO support in selected games.
With the mature platform, many new Core Ultra 200S processors below the K class and new, cheaper B860 mainboards, Intel will start a new attempt to position the Core i successor in the desktop market in early 2025. In the DIY segment, which the K CPUs are aimed at, the generation has so far been lying on the shelves like lead. However, the OEM market works differently and if Intel changes the prices a bit as availability increases, hobbyists could also take a second look.
Was this article interesting, helpful or both? The editorial team appreciates any support from ComputerBase Pro and deactivated ad blockers. More about ads on ComputerBase.
