Intel Core Ultra 9 285K VS AMD Ryzen 9 9950X3D

When it first launched in late October 2024, the Intel Core Ultra 9 285K was praised for its power efficiency improvements, yet the performance uplift over the previous gen i9-14900K was almost non-existent, even looking like a downgrade in gaming scenarios. Over the course of 2025, Intel has released quite a few microcode improvements, which translated into ~9% performance increase and better power efficiency in Linux. Is this true for Windows, as well? This is the question that inspired this test. Due to the GPU differences between our original test system (RTX 4090) and the XMG systems used for this test (RTX 5090), 3D rendering performance including games was not be explored.

We decided to pit the Core Ultra 9 285K against AMD’s Ryzen 9 9950X3D, as they represent the pinnacle for the current desktop generation from both companies. Granted, the two models are not in the same price bracket (the Intel CPU costs $100 less) and the core configuration differs. Nonetheless, the Ryzen 9 9950X3D has received some stability updates since its launch in March 2025 as well, so a retest seemed appropriate.

Despite not exactly comparing apples to apples, as the Intel Core Ultra 9 285K integrates a hybrid core cluster with 24 cores and no hyperthreading and the AMD Ryzen 9 9950X3D features 16 cores, the two processors trade blows in most workloads. We can see the latest Intel microcode offers noticeably better performance in GeekBench 6.5 multi-core (6% boost) and PCMark 10 (5% boost in Digital Content / 61% boost in Productivity / 4% boost in Essentials) over the initial microcode, but the two versions are mostly on par otherwise.

On the AMD side, the Ryzen 9 9950X3D dominates the AIDA CPU tests, even though the RAM Read and Copy speeds are 15-20% slower compared to the Intel platform. There are some notable gains on the latest revision code in the FPU Julia and Mandel AIDA tests, but overall the improvements are in the single digits over the original revision.

While both systems support PCIe 5.0 SSDs, it looks like the Intel motherboard is somehow limiting the PCIe 5.0 X4 slot as CrystalDiskMark reports that the maximum read speed reaches just above 12 GB/s instead of 14 GB/s. This limitation does not prove to be a problem in tests simulating real world use cases, however. In fact, the Intel system is 13% faster in the 3DMark storage test and 4% faster in the DirectStorage test.

When not under heavy loads, the Intel processor appears to be a bit more power efficient than its AMD competitor. Thus, in idle and single core scenarios, and even in most gaming tests where the GPU is doing most of the work, the Intel CPU consumes 20-30 W less on average. The AMD power draw can be improved to the point where it mostly matches the Intel CPU if the Balanced power profile is activated in Windows 11, and its performance in games does not seem to be affected, but this may have some negative impact on productivity and content creation workloads.

In scenarios where all cores are utilized such as the multi core stress test of Cinebench 2024 or the Prime 95 large FTT test, the AMD CPU emerges less power hungry even under the High Performance power profile. In Prime 95, the AMD CPU consumes 65 less W on average compared to the Intel one, while the Intel CPU also enters thermal throttling state and almost hits TJMAX in the first minute of the test, with the temperatures gradually getting lower afterwards. These extreme cases are not very likely to happen in real world scenarios, though.