MCE explanations and others

[RobertoSampaio]

Look this SP 48...
V/F8=1.603V

[theriffsthatkill]

Look this SP 48...
V/F8=1.603V

Wow, terrible at 52 and 53, but 1.29v @ 51 is not bad...

[rfkicker2000]

I'm new to overclocking. Running an i9-10900K with ASUS Rog Strix Z490-e motherboard.

I noticed setting SVID behavior to "Auto" results in an AC/DC LL of 0.66mOhms/0.66mOhms. Setting SVID behavior to "Trained" results in 0.66mOhms/1.1mOhms. If I understand Shamino's post correctly, I think the latter setting would result in higher voltages being requested from the controller, and possibly lead to more cases of VmaxStress. Seems like it is best to have the AC DC LL values the same and thus I should keep SVID behavior set to Auto?

I have VmaxStress and TVB Voltage Optmizations enabled with +1 boost profile.

Max vCore I saw in HWInfo64 was 1.359v for 5.4Ghz. Max VID was 1.421V for 5.4Ghz.

[Nate152]

Hi rfkicker2000,

Setting SVID Behavior to Best Case Scenario might yield better results, but what you have actually looks quite good.

[rfkicker2000]

Hi rfkicker2000,

Setting SVID Behavior to Best Case Scenario might yield better results, but what you have actually looks quite good.

Thanks Natel52! Forgot to mention the 10900k I have is SP 79 and I'm using LLC4 as recommended by the BIOS. All loadline settings are on Auto. Also, those VID and vCore readings were at idle. Any particular stress test I should run to get voltages under load? I'd prefer not to run Prime95 for very long since temps hit 90-95C during the test. I have a CoolerMaster ML360R cooler which gets a cooler score of about 160 in BIOS.

If not seeing or only occasionally seeing vmaxStress as an IA limit reason, is there reason to be concerned with that?

Windows will not boot with SVID Behavior set to 'Best Case Scenario', likely because the silicon quality isn't good enough for that low of a voltage.

[Nate152]

I'm not familiar with your newer cpu or motherboard so I can't really comment on VmaxStress. I also use Hwinfo64 and from the author he says, " VmaxStress may clip the frequency if the effective maximum voltage on the silicon is too high."

I thought maybe setting SVID Behavior to Best Case Scenario might lower the VID value. Since it wouldn't boot, auto looks to be working well for you. Vcore of 1.359v as you know is the voltage supplied to the cpu, VID of 1.42v is the requested voltage, this looks fine and nothing to worry about.

LLC helps with Vdroop or voltage overshoot. I use Hwinfo to check the maximum Vcore under load, and compare it to what cpu voltage I set in the bios to see they are at or very close the same, if overshooting too much I'll lower LLC. I use LLC 6 for my 7700k, I set 1.44v in the bios and it will fluctuate from 1.440v (current) up to 1.456v (maximum), I'm ok with that.

Your cpu doesn't seem to be throttling at idle, not sure if this is intended. You have a good cooler and it seems to be doing well. I like ROG Realbench for a stress test, it loads the cpu to 100% without hitting it as hard as Prime95 and should give you lower temps. Here's the link if you want to give it a try, click Download Realbench in the center of the page.

What does maximum Vcore/VID and temp go to?

https://rog.asus.com/rog-pro/realbench-v2-leaderboard/

Here's my Hwinfo, click the picture to make it bigger.