Testing ARROW lake - power limited

[BoggledBeagle]

I got myself a testing PC with 265K, Strix E board and ordinary RAM (32GB, 6000 MHz, CL36) and tested the CPU with two power limits 100 and 160W, and compared it to 13900KS with 4 E cores disabled, both with HT on and off.





As you can see, the results are pretty consistent, both apps in both power limits mean 24% performance improvent of Arrow lake compared to Raptor lake with HT disabled and 11% with HT enabled.

Most of the power inefficiency of Raptor lake is gone when it is so much power limited - Raptor lake is extremely efficient with 100W power limit. I believe that the result is pretty positive for Arrow lake with 20 threads against Raptor lake with 28 threads.

13900KS could be a little bit better than 14900K, but that is down to silicon lottery, I do not think it has a significant advantage compared to standard 14th gen CPUs.

 

[Onasi]

I mean, even any architectural changes aside, I would kind of hope for this being the case considering, you know, the massive node jump to TSMC 3N.

 

[Beginner Macro Device]

the massive node jump to TSMC 3N.

Yes, it's absurd.

NVIDIA, 2014: same node, give or take 40% more power efficiency (GTX 670 VS GTX 960).
Intel, 2024: 3x node shrink, up to 24% more power efficiency.

 

[Tralalak]

Could I ask you to test the CPU-Z 2.11 benchmark? The benchmark (2.09, 2.10, 2.11) can now be ran on a single coreset/cluster.

• Compare P-cores Golden/Raptor Cove vs P-cores Lion Cove
• Compare E-cores Gracemont vs E-cores Skymont

source: https://www.techpowerup.com/forums/threads/share-your-cpuz-benchmarks.216765/post-5286485
source:

Share your CPUZ Benchmarks!

Hi, Can you tested also the CPU-Z 2.09 benchmark on a single coreset/cluster ? Single-Thread / Multi-Thread • All Cores : 901 / 10 793 • P-Cores : ??? / ???? • E-Cores : ??? / ???? Thanks

www.techpowerup.com

Thank you in advance.

 

[N/A]

Intel, 2024: 3x node shrink, up to 24% more power efficiency.

That is if the CPU has the ideal voltage curve. But it might be able to get a bit more performance per watt with some fine tuning.

Also, the input voltage DLVR is 1.5 volts and must be stepped down to 1.2 as a resistor, so there is a 33% power loss in voltage conversion, as explained by Derbauer.