An overclocked Core Ultra 9 285K vs an overclocked Core Ultra 7 270K Plus

El_Capitan

2[H]4U
2FA
Joined
Nov 11, 2012
Messages
2,143
Updated with I think my final stable overclocks and scores (11/10/2026).

I finally got my hands on a Core Ultra 9 285K, but it's an engineering sample. I didn't know this until I got it (I should look at the product photos better). I overclocked and optimized it, and put it against my best overclocking and optimized Core Ultra 7 270K Plus (I've gone through 5 270K Plus's). The test setup is exactly the same, the only difference is the CPU.

Core Ultra 9 285K:
P-Core = 5.7 (2) - 5.6 (6) - 5.5 (8)
E-Core = 4.8
CPU Ratio = 42
NGU = 35
D2D = 28

Core Ultra 7 270K Plus:
P-Core = 5.8 (all core)
E-Core = 5.0
CPU Ratio = 43
NGU = 31
D2D = 30

On paper, it looks like the Core Ultra 7 270K Plus should perform a lot better.

CPU-Z 3.02 Single:
Core Ultra 9 285K = 943 (+0.71%)
Core Ultra 7 270K Plus = 936.3

CPU-Z 3.02 Multi:
Core Ultra 9 285K = 19518.4 (+0.6%)
Core Ultra 7 270K Plus = 19401.5

Geekbench 7.1 Single:
Core Ultra 9 285K = 3057 (-0.1%)
Core Ultra 7 270K Plus = 3060

Geekbench 7.1 Multi:
Core Ultra 9 285K = 28739 (-1.18%)
Core Ultra 7 270K Plus = 29079
2026_11_10_285K_SP87_8000MHz_CL38_57(2)-56(6)-55(8)_48_42_35_28_Geekbench7_CPU-Z.png
2026_11_04_270K_Plus_SP86_8000MHz_CL38_58_1.30V_50_1.26V_43_34_30_Adaptive_Geekbench7_CPU-Z_2.png

Y-Cruncher (0, 1, 8):
Core Ultra 9 285K = 166.061 (-6.88%)
Core Ultra 7 270K Plus = 155.009
2026_11_10_285K_SP87_8000MHz_CL38_57(2)-56(6)-55(8)_48_42_35_28_Y-Cruncher.png
2026_11_04_270K_Plus_SP86_8000MHz_CL38_58_1.30V_50_1.26V_43_34_30_Adaptive_Y-Cruncher.png

Cyberpunk2077 2K High Native with an overclocked RTX 5070 Ti:
Core Ultra 9 285K = 164.24 (+1.02%)
Core Ultra 7 270K Plus = 162.57

2026_11_04_Gigabyte_RTX_5070Ti_Aero_OC_380_Core_i7-270K_Plus_58_50_43_34_30_8000MHz_CL38_Cyber...png

Geekbench 7 CUDA with an overclocked RTX 5070 Ti:
Core Ultra 9 285K = 382498 (+7.19%)
Core Ultra 7 270K Plus = 355964

Geekbench 7 OpenCL with an overclocked RTX 5070 Ti:
Core Ultra 9 285K = 271430 (+2.47%)
Core Ultra 7 270K Plus = 264817

Geekbench 7 Vulkan with an overclocked RTX 5070 Ti:
Core Ultra 9 285K = 249056 (+7.11%)
Core Ultra 7 270K Plus = 231955
2026_11_10_Gigabyte_RTX_5070Ti_Aero_OC_380_Core_i9-285K_57(2)-56(6)-55(8)_48_42_35_28_8000MHz_...png
2026_11_04_Gigabyte_RTX_5070Ti_Aero_OC_380_Core_i7-270K_Plus_58_50_43_34_30_8000MHz_CL38_Geekb...png

I may retake some of these tests and update the scores. The Core Ultra 7 270K Plus is obviously the better value, and all the ones I've gone through all overclock better than the sole Core Ultra 9 285K ES sample I got my hands on. Y-Cruncher also performs better with the 270K Plus. However, the Core Ultra 9 285K seems to perform a hair better in gaming at 2K High Native in Cyberpunk2077, and much more with the Geekbench 7 GPU benchmarks, which was interesting to see. The Core Ultra 9 is also much better at memory overclocking (which I'm not going to do with the price of DDR5 RAM these days).

While seemingly identical as far as specs, the 270K Plus seems to have better potential for overclocking, while also being easier to cool. On paper, the 270K Plus used here should really outperform the 285K, but it only thoroughly did so in Y-Cruncher, and barely in Geekbench7.
 

Attachments

  • 2026_11_10_Gigabyte_RTX_5070Ti_Aero_OC_380_Core_i9-285K_57(2)-56(6)-55(8)_48_42_35_28_8000MHz_...png
    2026_11_10_Gigabyte_RTX_5070Ti_Aero_OC_380_Core_i9-285K_57(2)-56(6)-55(8)_48_42_35_28_8000MHz_...png
    957.2 KB · Views: 0
Last edited:
So, what you're telling us is two nearly identical chips perform nearly identically? Jokes aside, the geekbench thing is pretty strange as they are so similar, I wonder if its some ES sauce allowing that gap.

Since you seem to like doing this crap have you tried single channel? I am getting very strange results on my 250K. CP2077 and Crimson Desert both show better results 🤷‍♂️

4.png5.png

1.png2.png
 
Last edited:
So, what you're telling us is two nearly identical chips perform nearly identically? Jokes aside, the geekbench thing is pretty strange as they are so similar, I wonder if its some ES sauce allowing that gap.

Since you seem to like doing this crap have you tried single channel? I am getting very strange results on my 250K. CP2077 and Crimson Desert both show better results 🤷‍♂️

View attachment 835862View attachment 835863

View attachment 835864View attachment 835865
Not for the 270K Plus or the 285K, but I have for the 9850X3D and 14600K. The 9850X3D gained FPS performance with single channel in Cyberpunk2077, whereas the 14600K didn't.
 
Back
Top