M13H + i9-11900K - intel specs

[sl910]

Hi,

I have just built a new PC based on a M13H, an intel 11900K and a Nvidia RTX 3090 FE GC. It is mainly used for Microsoft Flight simulator 2020. The result is really amazing, i.e about 40 FPS in the Ultra mode (1440x3440) and in very dense area (photogravimetry) at low altitude.

My Bios setting are those by default + Intel ABT on + resize bar on. When idling, the Vcore voltage permanently oscillates between 0.7 V and 1.45V, even when idle. This last value sounds too high in my ears.

My previous PC was a M9C + i7-7700K. As a result I am not yet familiar with the options of the M13H Bios. My question is the following:

1. If I want to use very conservative values corresponding strictly to intel specs, which Bios options should I change
2. How to limit the Vcore voltage in order to be sure to be in the official and healthy range.

By the way, I was obliged to turn off HT otherwise I get many BSOD's (clock.watchdog.timeout), even when idling. Switching off HT makes the system rock solid for several days now (no more BSOD).

When making an Aida64 stress test (CPU+FPU) with defaut preferences, i.e. AVX 512 ON, there are several thermal throtteling events. Is this normal ? I use a Noctua NH-D15S cooler

Best regards

sl91

[thomas_yiu]

Hello sl9,

I suggest to reinsert the thermal paste on the processor.
The range of voltage is normal paramters, because when it reach 5.2Ghz, it is usually 1.42V.

[sl910]

thomas_yiu@ROG wrote:
Hello sl9,

I suggest to reinsert the thermal paste on the processor.
The range of voltage is normal paramters, because when it reach 5.2Ghz, it is usually 1.42V.

Thank you for the idea. In the Bios, i can read A prediction of SP=82 and a cooling performance index ranging between 166 abnd 176. Are these values compatible with the idea that the cooler is not correctly mounted with respect to the thermal considerations, i.e. not enought thermal paste. I have applied a round dot of about 3 mm on the center of the CPU as advised by Noctua.

Best regards

sl91

[sdmf74]

The vf point 7 frequency on my chip for 52x is 1.468v

I just love it when people give a reply & dont explain why or give any context but yes that voltage is fine, Rocket lake has bigger cores & can handle more voltage, I just came from an I7-7700k also and 1.35v- 1.4v was high on that chip but not on RL.

Also you should never have to disable hyper threading to stop bsod's! If you are getting them you have other settings wrong or other issues, I would fix them rather than handicapping your expensive processor.

Another thing Im just learning about RL is the better value you select for svid support (closest to best-case scenario) will provide much lower temps. On my chip the difference between best-case scenario & Intels fail safe is about 16C+ @ LLC4!
Mine couldnt pass P95 though @ LLC4 & best-case scenario so I needed slightly more than .01mOHM
AC/DC LL or select typical-case scenario (although it looks like Asus chose .600mOHM AC/DC LL for TC scenario which is uneccessarily high.

Also use good thermal paste & spread it evenly across the chip. Ive never used Noctua TP but Gelid GC extreme, TG kryonaught, or K|INGP|N KPx are all excellent choices!

[sl910]

Thank you for reporting your experience. Finally, in my case I found that all the issues were generated by the USB Audio. When I diseabled it in the BIOS, all the issues disappeared. I use tne Nvidia sound (Display port). I will try the Best scenario setting.

Regards

sl91

[sdmf74]

Ahh yes of course Asus armory crate and/or audio drivers you are correct they are nothing more than bloatware and defective as hell! Sadly we need audio drivers to function its a shame that Asus despite being on their 13th generation of ROG motherboards has NOT figured out how to develop a working driver yet. Its pathetic!

Actually unless you have a golden top 10% 11900k switch svid behaviour to "TYPICAL SCENARIO" (make sure that you have the IA AC/DC LOAD LINE settings both set to "AUTO" otherwise svid behavior wont do anything.
Turn hyperthreading back on, then
set your fans to max and run Prime95 small FFT's with all the AVX options disabled & see if you can pass that for 30 minutes without getting any errors, meanwhile
monitoring your temps & your voltage closely with HWmonitor64 (or something similar).

What LLC setting are you using? You should be able to do the above with LLC4

[sl910]

Hi,

Starting from the Intel's safe conservative setting, I have tried the following just for testing purpose and in order to try to lower the CPU temperature:

1. Best case + CPU LLC auto (i.e. 2): Windows 10 load freeze - BSOD Machine_Check_Exception
2. Best case + CPU LLC 3: Normal windows boot - Aida 64 FPU (with AVX) stress test OK - Cinebench freeze and BSOD Machine_Check_Exception
3. Best case + CPU LLC 4: Normal windows boot - Cinebench freeze and BSOD Machine_Check_Exception
4. Typical case + CPU LLC auto (i.e. 2): Normal windows boot - Aida 64 FPU (with AVX) stress test OK - Cinebench OK - P95 (small + AVX's) OK - MSFS OK

IA AC/DC LOAD LINE settings both set to "AUTO" all the time

Regards

sl910

[sdmf74]

Yes that is correct that the lower your LLC setting (I wouldn't use auto) the lower the temperature & voltage overshoot but what multiplier are you using? I would imagine you will bsod at LLC2 with any sort of overclock. Even just with turbo boost clocks.

There is a reason Asus recommends LLC4 for any type of overclock, sure every CPU is different but they have tested prolly hundreds of CPUs and professional overclockers have verified this setting through thorough testing.

I have tried LLC4 and LLC5 but I only used LLC5 before I was familiar with the SVID behavior setting. I determined it was better to switch to typical scenario from best-case and lower LLC back down to LLC4 (my chip is an SP86 IIRC).

I've been busy playing with my new RTX3080 FTW3 but once I do some air overclocking on it im gonna slap my waterblock on it and then I'm gonna come back to the 11900k and dial in a manual AC/DC LL setting rather than using one of the SVID behaviors. I figure since instability is .01mohm with best-case and stability is around .300mohm which is typical-case I should be able to dial in a much lower setting somewhere in between while lowering temps in the process

Btw Im almost positive your bsods were related to having too low of an LLC setting (IE LLC up to level 3) rather than hyper threading causing it

[sl910]

My i9-11900K SP is 82

sl91