Question intel 9900k 5.0 ghz high temp with h100 pro

[hwmachine]

Hey!

I'm quite new to OC but very interested! So if you have any tips or tricks i would appreciate it.

Before i start explaining what I've tried to do I will post a youtube title on the guide which i followed., and most of my settings are the same as his (This is not an AD)
The Complete i9 9900K Overclocking Guide - Maximus XI Z390 and Others




My hardware:
CPU Intel i9 9900k
Motherboard Asus rog strix 390-f
Cooler: Corsair H100 pro

Here's the thing. I've got an i9 9900k which I've tried to overclock to 5.0 ghz without any real progress for some time. My current settings in BIOS are as followings.

LLC: level 7
DRAM Freq: 3600mhz
CPU Core/Cache Limit 255,75
Min. CPU cache ratio 45
Max -||- 45

CPU Core/Cache voltage: Manual
CPU Core voltage override 1.295 (tried it all the way up to 1.325, temps high all the way to 1.325)
DRAM Voltage 1.35
CPU VCCIO Voltage 1.15000



The issue
Now I have tried to benchmark around the cpu core voltage ratio which has been changing from 1.295 to 1.325 the higher i go the faster i gain temp, which is logical.

Either way, during my tests on cinebench r23 i have a temperatre after 2-3 min run time around 92-96 degrees celcius. I have tried changing my values through different kind of tutorials but its mostly the same with few temp differences. Now my question is what may be the cause that my temps are this high? MY current idle temp with 5.0ghz and current settings with vram voltage 1.295 is around 40degrees. Can it be my cooler which is not working properly or can it be that im just unlucky with the CPU and it cant handle the 5.0 ghz?

Thanks for all the answers, and i hope i clarified some information.

 

[Phaaze88]

Oh, I remember that guide. Made me facepalm a few times...

Spoiler

1)Please note that some 9900Ks can, and some can't, reliably do 5.0ghz. The latter either just can't do it at all(constantly crashes regardless of settings), or require a butt-load more Vcore to be stable, going into unsafe territory, leading to accelerated voltage degradation if run 24/7.
The voltage degradation will cause the OC to become unstable after an unspecified period of time.
Later revisions of the cpu have better bins, but the samples that can't do it is lower.

2)High LLC. That is just not practical for most users, and is really for liquid cooled, or LN2 cooled VRMs.
Why the heck so many of these 'guides' suggest/recommend ridiculously high LLC is beyond me... like, they didn't put in enough time to test it thoroughly or something...
Asus boards have been pretty good with LLC - use Level 1 for starters, and go no higher than Level 4.
Choose to skip to Level 4 anyway, and things getting too hot? Turn it down to 3. 3 too hot? Turn it down, and so on.

3)The H100i Pro is really not adequate for this cpu at the target frequency. There are scenarios that will see 80C or more.
You've not skimped on airflow, at least? Have a high end Ampere gpu beneath it too? That'd only further solidify the inadequate statement.

4)AVX does crank up the power use, as well as temperatures, when it is in use.
Best drop that offset up to -3 if temperatures are still subpar.

5)There's little reason to disable Cpu Svid.
It at least lets you see how much power the cpu's pulling.

6)Set a maximum cpu core temperature.
Yeah, the cpu default thermal throttle limit is 100C, but there's no reason to let it get that high in the first place.
Set it to 86 or 90C.

7)This guy ran large data set in OCCT, and not small. This might be the biggest joke about this guide.
Large data set does not heavily stress the cpu; if it's anything like Prime95's small and large sets, large mainly stresses the memory controller and the ram. It's not that hard on the cpu cores.

Short version of the stuff in spoiler, because I have a bad habit of dragging on:
-some settings are too aggressive.
-current cooler isn't sufficient for the performance target.
-video guide fails at the end with stability testing.