Voltage drops below manual setting

[Somethingwhatever]

I have my 8700k set to 4.5ghz 1.26v with an LLC of 6. It reads 1.264v in Hwinfo and normally under load it hits 1.28v, usually averages in the 1.27 range when I'm gaming. What just started happening is that the voltage sometimes will drop to 1.248v and there's no explanation as to why. I tried gaming for an hour and an hour of OCCT just to see and if it's steady at 1.28 the entire time and it was. I was just watching a youtube video and it dropped at some point to 1.248v.

Lately, the +12v psu readings on HWinfo have dropped from almost a constant 12v to 11.904 - 11.808. I know these are within the +-5% acceptable range for my EVGA 1000 GQ. It's just odd I start seeing this drop. I believe it happens when the +12v drops to 11.808. The BIOS reads +3 +5 +12 no deviation and the +3v and +5v are also fine in HWINFO. +3v sits about 3.3v, +5v sits at +5v dips to 4.98v rarely.

There's also a heat smell coming from my PC. Not burning, just like a component somewhere is either overheating or just being worked a little harder under load. It never used to drop to 11.808v

Full Specs:

8700k 4.5 1.26 LLC6
Maximus X Formula latest bios
evga 1000w GQ
evga 980 sc while my 1080 is being RMA'd (this card pulls 1.2v Maximum in hwinfo)
970 evo 500gb
Toshiba HDD + 16gb intel optane
Thermaltake 240mm AIO

[Japion]

Have you using different PSU?

[Somethingwhatever]

Have you using different PSU?

No. The only other one I have is like 5 years old. 1000watts still but I'm not risking it. What's odd is the readings in the bios are normal. Hwinfo is showing fluctuations but they're not drastic or frequent. Not sure which is more accurate. I would think the bios.

[NemesisChild]

Have you adjusted the LLC setting in your bios?

[MrAgapiGC]

Have you adjusted the LLC setting in your bios?

good point

I detect, on my 9900k codexi these.

I have desable the AVX since windows use avx instruction. No explanation since i avx there is o drop

Now. windows also make these droop. have being playing with the LLC and is consistent

I have now level7 and still drops from 1.30 to 1.28/1.27 volts (5.0ghz)

Try these with a normal oc and with S3 disable. Odd but expected.

[Somethingwhatever]

Have you adjusted the LLC setting in your bios?

I adjusted a bunch of times while I was dialing in my OC. The issue is that it drops below the manual setting of 1.264v to 1.248v with LLC6. That's Vdroop and LLC6 overvolts so I'm not sure what the issue is. Going the other way, with LLC5 and setting manual voltage to 1.3v, it will idle at 1.312v and it will drop to 1.28v under load. The only other time I've seen it droop to 1.248v is if I keep the LLC at 5 and set the manual vcore to 1.29. It then idles at 1.28v and drops to 1.248v underload which is unstable. That's the only time i've seen it drop that low other then when it's on stock at 1.25v

Before any of this I had it at 4.6ghz stable at 1.32v LLC5 -> 1.280v load 1.304v average. I set AVX to -2 to try it out and saw that the CPU was dropping to AVX speed even when there's no AVX being processed, so I turned it back to auto. I then read that there is some sort of bug with AVX and they're going to release a bios update at some point. The 4.6 OC was a little too warm for my taste so I backed it off. It's really after I tried the AVX thing that I saw this start happening. AVX only drops the clock speed though, not the voltage as far as I know. This also doesn't happen all the time. It's just ever now and then I'll do something trivial and the voltage will hit 1.248v

I've been trying to figure out where I can safely OC for about a month and a half now and I'm just wondering if all this stress testing maybe has borked something either with the bios or the PSU. I have a mid tier AIO so it doesn't have a high enough TDP for anything above 4.6ghz 1.3v with LLC5 (vdroop) I can't get an answer out of the HWinfo guy about how accurate the program readings are and ASUS refuses to answer questions about overclocking which is hilarious. I just really hope someone has a definitive answer because I'm reluctant to do anything right now until I figure this out.

Thanks for your replies.

[Somethingwhatever]

good point

I detect, on my 9900k codexi these.

I have desable the AVX since windows use avx instruction. No explanation since i avx there is o drop

Now. windows also make these droop. have being playing with the LLC and is consistent

I have now level7 and still drops from 1.30 to 1.28/1.27 volts (5.0ghz)

Try these with a normal oc and with S3 disable. Odd but expected.

I don't understand what you're saying. If you have an ASUS board set to LLC7 it should be overvolting. That's what it does, or would do if I set it to LLC7 for my 8700k. Maybe 9 series is different but I know that lower LLC means more vdroop. LLC5 is where vdroop starts and LLC1 is the greatest. I don't know what S3 is either. I feel like you're trying to relate to this somehow and I don't understand. I'm sorry and thank you for your reply =D

[Falkentyne]

I adjusted a bunch of times while I was dialing in my OC. The issue is that it drops below the manual setting of 1.264v to 1.248v with LLC6. That's Vdroop and LLC6 overvolts so I'm not sure what the issue is. Going the other way, with LLC5 and setting manual voltage to 1.3v, it will idle at 1.312v and it will drop to 1.28v under load. The only other time I've seen it droop to 1.248v is if I keep the LLC at 5 and set the manual vcore to 1.29. It then idles at 1.28v and drops to 1.248v underload which is unstable. That's the only time i've seen it drop that low other then when it's on stock at 1.25v

Before any of this I had it at 4.6ghz stable at 1.32v LLC5 -> 1.280v load 1.304v average. I set AVX to -2 to try it out and saw that the CPU was dropping to AVX speed even when there's no AVX being processed, so I turned it back to auto. I then read that there is some sort of bug with AVX and they're going to release a bios update at some point. The 4.6 OC was a little too warm for my taste so I backed it off. It's really after I tried the AVX thing that I saw this start happening. AVX only drops the clock speed though, not the voltage as far as I know. This also doesn't happen all the time. It's just ever now and then I'll do something trivial and the voltage will hit 1.248v

I've been trying to figure out where I can safely OC for about a month and a half now and I'm just wondering if all this stress testing maybe has borked something either with the bios or the PSU. I have a mid tier AIO so it doesn't have a high enough TDP for anything above 4.6ghz 1.3v with LLC5 (vdroop) I can't get an answer out of the HWinfo guy about how accurate the program readings are and ASUS refuses to answer questions about overclocking which is hilarious. I just really hope someone has a definitive answer because I'm reluctant to do anything right now until I figure this out.

Thanks for your replies.

The last two posts are wrong.
NO LLC levels overvolt.
Transient spikes isn't the same as what you call overvolting from reading vcore from a badly calibrated SIO chip (Pre Maximus XI boards).
LLC8 does not overvolt. It's a 0 mOhm loadline.
LLC7 is about a 0.2 mOhms loadline.
LLC6 is about a 0.4 mOhms loadline.
overvolting would mean a negative resistance which is not possible.
You can "temporarily" exceed the BIOS set voltage even on LLC3 ! Those are known as transients swings. Not sustained voltage.

https://www.overclock.net/forum/6-in...l#post28022104
https://www.overclock.net/forum/27686004-post2664.html

[Somethingwhatever]

None of that answers my question. You're hung up on the words I'm using. There's no LLC8 on my board. I don't have a 9900k or a maximus XI genie, which has a different bios, so that first post didn't really help me. He just says you want a flat load line. I know that already. I can get a flat loadline with LLC6 4.5ghz 1.25v Hits 1.26v under heavy load but averages 1.251v with my everyday use. Before this post, I had it at 4.5 1.26v which would hit 1.28v with LLC6 but then sometimes drop below 1.26v to 1.248v and I don't know why. I have no idea what the state of my SIO chip is. I never had these drops in vcore before.

I just wanted to know why that would happen. No LLC overvolts? MY BAD DUDE. I didn't know we were splitting hairs today. Obviously I meant 'overshoots' which causes more heat. Again, I knew that already. You're also saying I can exceed the bios setting with LLC3. If I set 4.5ghz at 1.25v to LLC3 my computer probably wouldn't even boot to windows. With LLC3 4.5ghz I set it to 1.3v just to test how far down it goes. It hits 1.232v which is unstable even for stock all core. I don't know what voltage you would have to set to make LLC3 'overshoot' the manual vcore setting and it doesn't matter because my question, again, is why would it 'undershoot' the manual setting with an LLC that is designed to reduce the vdroop with the least amount of overshoot.

It's like pulling teeth.

[Falkentyne]

I already told you why it happens and you completely ignored me. Maybe it's all over your head. Do you know how power plane impedance works?
First let's explain voltage loadline resistance. Resistance causes a drop in voltage, and resistance is measured in milliohms. Increase the resistance, the more vdroop you have. Decrease it, the less vdroop you have, but the more transient spikes and dips you have due to 'feedback loop'.

This is explained here.

https://en.wikichip.org/wiki/load-line_calibration

Now that helps you understand the maximus XI oscilloscope shots here.
https://www.overclock.net/forum/6-in...l#post28022104

However, voltage *sensors* that encounter IMPEDANCE across the power plane behave differently than voltage *itself*. Note the terms impedance vs resistance.
Resistance operates on the AC circuit and impedance on the DC circuit. But it's basically the same term.

When a voltage sensor encounters impedance, there is a *RISE* in the measured voltage. This is what causes the vcore sensor to report erratic voltages or to report a vcore that is higher than you set in the bios. The Super I/O sensor on your board and on many boards released in the last 12 years suffer from issues like this.

You're basically asking "why is the super i/o sensor inaccurate?". Well I just answered that for you.
If you really want more information, you need to either contact an Asus hardware engineer or take an EE class and learn about IC's and motherboards on your own. You are not going to get the answers you want by posting on a basic computer forum. I'm sorry if you don't like that answer, but that's all I can tell you. Good luck.