Ideas for 13600k optimization

[MrFrank]

Hello everyone, I hope you're doing great!
So I built a new PC this week and I was testing my CPU with cinebench r23

My build:
CPU: i5 13600k
MOBO: Gigabyte Aorus x690 Elite DDR4
COOLER: Arctic Freezer II 280 Rev 7

My BIOS settings:
XMP profile enabled
Power Limit set from "Auto" to "Intel POR"

My score was: 20526 and here are my Hwinfo64 logs during a 10 minute CPU Multi core test:


I would be happy to see your suggestions in order to improve/optimize my CPU!
Thank you in advance!

 

[dejo]

I would start with trying to get the voltage in bios for the cpu lowered some. I use auto or adaptive voltage for the cpu and then under "dynamic voltage" start using a -.01 and test and watch heat under a load and stability. If all works well- try -.015 and work your way down to see where you can end up. I expect you can lower temps by quite a few degrees. @5.3ghz, my 13700K I can get to -.11 without stability issue

 

[yoadknux]

I would start with trying to get the voltage in bios for the cpu lowered some. I use auto or adaptive voltage for the cpu and then under "dynamic voltage" start using a -.01 and test and watch heat under a load and stability. If all works well- try -.015 and work your way down to see where you can end up. I expect you can lower temps by quite a few degrees. @5.3ghz, my 13700K I can get to -.11 without stability issue

Do you have a special LLC setting? I'm just starting to overclock my 13600k on an Aorus Z790 and I'm looking to have maybe a "ballpark" starting point in terms of voltage and core clocks before I tweak things. Also, when you stress tested it, did you use P95? AVX on or off? I hear several opinions.

 

[ShrimpBrime]

Do you have a special LLC setting? I'm just starting to overclock my 13600k on an Aorus Z790 and I'm looking to have maybe a "ballpark" starting point in terms of voltage and core clocks before I tweak things. Also, when you stress tested it, did you use P95? AVX on or off? I hear several opinions.

If you can pass yCruncher 2.5b and keep her under 99c, you're doing good in my opinion. Good memory and IMC test also.

 

[Woomack]

I agree here with ShrimpBrime. Whatever is below 99°C is already good.
I would start lowering the voltage to the point it will be ~95°C under full load (in the worst scenario). When you are sure it's stable and below the throttling point, then you will see your max voltage, and you can bump the frequency and check stability as the frequency itself is not raising the wattage (unless something else is changing). However, I recommend checking the max CPU voltage together with already overclocked RAM as SA/IO and other voltages are raising the CPU temp too. So in, short check the max voltage at ~95°C CPU with already overclocked RAM, test the stability of both, and next, check how high the CPU clock will go on the voltage that you already checked.

LLC for overclocking is almost always 2-3 spots below the maximum, so the voltage is not going above standard under a higher load. Typically the value is at 3-4 for ASUS and 2 for brands with 5 level scale. Of course, it can be different, so the best is to check which one is close to the desired voltage under a high CPU load.

 

[Bumslap]

Had a feeling i needed to post this. Sold my 13600K because of the horrible temps and Asus MB. Bought 13600KF and Z690 Tomahawk MB. I'm absolutely stunned by the results. First of all, stock it took quite a long time(2-3min) to reach 100°C which was already amazing. And here is the thing... I am using 50€ aircooler !

P0 5.6
P1 5.6
P2 5.4
P3 5.4
P4 5.6
P5 5.4

E-Cores
0-3 4.2
4-7 4.4

Ring 4.7

1.25V
Lite Load 4
HT-off
148W power draw


Almost the same score as stock HT-on !

13600KF
MSi Z690 Tomahawk
Kingston Fury Beast DDR4 3200 CL16 16Gb
Cooler NZXT T120(1 Fan)
Case NZXT H7 Flow(2 case fans)
Thermalright CPU Contact Frame
Grizzly Kryonaut

My advise: Never ever ever go K. There's a 0.0001% chance the iGPU turns out useful.

E: Turns out i get few extra frames in games with default ring 4.5GHz and 1.24V. 2-3°C lower temps also.

 

[yoadknux]

Isn't 21969 a bit low? I have a very conservative OC on my 13600k (5.5 + 4.2) and I score 25905. Aren't the KF and K versions identical either than iGPU?


Regardless, I think you made the right call going with MSI. Gigabyte and ASUS have issues for Z690+Z790 from what I read.

 

[Bumslap]

Isn't 21969 a bit low? I have a very conservative OC on my 13600k (5.5 + 4.2) and I score 25905. Aren't the KF and K versions identical either than iGPU?


Regardless, I think you made the right call going with MSI. Gigabyte and ASUS have issues for Z690+Z790 from what I read.

It's pretty damn good score HT(hyperthreading) OFF. HT enabled the score will be around same as yours. I will test it tonight. You get more FPS in games HT disabled and lower temps. KF just lacks the iGPU and isn't as "stacked".

 

[yoadknux]

It's pretty damn good score HT(hyperthreading) OFF. HT enabled the score will be around same as yours. I will test it tonight. You get more FPS in games HT disabled and lower temps. KF just lacks the iGPU and isn't as "stacked".

Ok, I didn't get the point that HT is off. So I'm sure you'll score as high as me or higher.