Having trouble switching to Adaptive voltage mode on Z390 Maximus XI Formula + 9900k

[Zammin]

Hi everyone

I've just recently finished dialing in my stable 5Ghz overclock in manual mode with the following settings:

- AI Overclock Tuner XMP I (clicked Yes at the prompt)
- Sync all cores
- AVX Offset 0
- Manual Voltage Override 1.33V
- LLC7
- IA/DC Loadlines 0.01
- Loadline Synch diabled
- CPU Current Capability 170%
- Long and Short Duration Package Power Limit 4095
- Package Power Time Window 127
- CPU Core/Cache Current Limite Max 255.75
- Cache Ratios Auto
- VCCIO 1.15V
- System Agent Voltage Auto (1.2V)
- BCLCK Aware Adaptive Voltage disabled
- MCE disabled
- SVID disabled
- Intel SpeedShift/Step disabled
- C-States disabled

Now that I'm stable in manual mode I'm currently trying to switch to adaptive voltage mode. I've done this before on Z370 and my 8700k with success, however with the 9900k and my ASUS Z390 motherboard it is exhibiting some strange behavior.

In manual mode it sits at 1.314V idle and drops to 1.27V under RealBench (or AIDA64) types of loads. When I re-enable C SpeedShift/Step, C-States and SVID, then switch to Adaptive voltage and set 1.33V as the additional turbo voltage it idles between 1.33-1.35V and sits around 1.33V under load, causing it to run much hotter than before. The only way I can get it to come down to 1.33V idle and sag to 1.27V under load again is to set LLC to level 6. This is all very strange to me as I would expect the LLC behavior as well as the idle voltage and load voltages to be the same as before until I set the windows power plan to balanced to allow the CPU to clock down when at idle.

This is not in line with my experience overclocking my 8700k on my ASUS Z370 Maximus X Code. In this case on the Z390 motherboard LLC6 in adaptive mode behaves closer to the way LLC7 behaves in manual mode, although the voltage level is not as stable under load, it fluctuates up and down a lot instead of sitting at 1.27V for the duration of the test. This appears to cause more heat and definitely isn't stable as I get BSOD's in RealBench with these settings in adaptive mode.

I'm not really sure what to do, I want to use adaptive mode to allow my CPU to clock down (and bring the voltage down) at idle but I just can't get it to work. I had no trouble doing it on Z370 following Raja's Z270 overclocking guide but it just doesn't want to work on Z390 for some reason.

If anyone has any advice for me I would greatly appreciate it. I've listed my settings above but if anyone requires BIOS screenshots I can get them for you as well.

[jpmboy]

Did you try this without changing the AC and DC load lines... leave them on Auto - changing these only if necessary. Also, enable all c-states up to C6, disable speedSTEP and enable speedSHIFT in bios. Adaptive is working fine on my Max11E. NOt all that different from your Formula, and I'm using LLC6, 1.325V in bios - AVX loads droop to 1.279V.

[Zammin]

jpmboy wrote:
Did you try this without changing the AC and DC load lines... leave them on Auto - changing these only if necessary. Also, enable all c-states up to C6, disable speedSTEP and enable speedSHIFT in bios. Adaptive is working fine on my Max11E. NOt all that different from your Formula, and I'm using LLC6, 1.325V in bios - AVX loads droop to 1.279V.

Hey man thanks for the reply.

When I was dialing in my manual overclock I couldn't get it stable at a reasonable voltage until I set IA/DC loadlines to 0.01. Then I was able to get stability at a good voltage. In Raja's adaptive OC guide he suggests setting these to 0.01 as well. I did this on my 8700k adaptive 5Ghz OC as well.

I could try setting them to auto again but if last time was anything to go by, it will want more voltage and I'm already hitting peak individual core temps of 92-95C in 1 hour stress tests of RealBench or AIDA64. So I don't really want to add much more voltage when I'm stable at 1.27V load with these settings.

All C states should have been enabled with my adaptive settings, I didn't disable or enable them individually, there is just one setting that say "C-States" that you can either enable or disable. That's all I did regarding C-states.

If I disable speedstep and speed shift, will that not prevent the CPU from downclocking/volting when idle on balanced windows power plan? Or will it just prevent the CPU from downclocking and still enable to voltage to drop at idle? Sorry for not fully understanding.

The part that I'm most confused about is why the LLC behavior would be different between manual and adaptive, because that seems to be what's preventing me from achieving stability in adaptive mode. 😕

[jpmboy]

Zammin wrote:
Hey man thanks for the reply.

When I was dialing in my manual overclock I couldn't get it stable at a reasonable voltage until I set IA/DC loadlines to 0.01. Then I was able to get stability at a good voltage. In Raja's adaptive OC guide he suggests setting these to 0.01 as well. I did this on my 8700k adaptive 5Ghz OC as well.

I could try setting them to auto again but if last time was anything to go by, it will want more voltage and I'm already hitting peak individual core temps of 92-95C in 1 hour stress tests of RealBench or AIDA64. So I don't really want to add much more voltage when I'm stable at 1.27V load with these settings.

All C states should have been enabled with my adaptive settings, I didn't disable or enable them individually, there is just one setting that say "C-States" that you can either enable or disable. That's all I did regarding C-states.

If I disable speedstep and speed shift, will that not prevent the CPU from downclocking/volting when idle on balanced windows power plan? Or will it just prevent the CPU from downclocking and still enable to voltage to drop at idle? Sorry for not fully understanding.

The part that I'm most confused about is why the LLC behavior would be different between manual and adaptive, because that seems to be what's preventing me from achieving stability in adaptive mode. 😕

actually that's : disable STEP and enable SHIFT as I posted. Shamino laid this out for you - follow his recommendations. As with any recent architecture, Adaptive only applies to multipliers higher than the stock max turbo multi, and then it can only ADD(itional) turbo voltage to the VID at that multiplier.

[Zammin]

jpmboy wrote:
actually that's : disable STEP and enable SHIFT as I posted.

Ah yes that was my bad, I misread it.

jpmboy wrote:
Shamino laid this out for you - follow his recommendations. As with any recent architecture, Adaptive only applies to multipliers higher than the stock max turbo multi, and then it can only ADD(itional) turbo voltage to the VID at that multiplier.

Yeah I understand that now. As I said I will try his suggestion and post back with the results. All being well I should have some time tomorrow to play around with it more.

[Zammin]

I just tried the usual adaptive mode settings with LLC6 and IA/DC loadlines on auto and it did seem like the voltage was pretty steady at 1.27V in realbench but it detected instability at 6m (blender) and stopped.

[Shamino]

welcome to adaptive voltage

understand the criteria for adaptive voltage to take effect :

per intel design:
1) adaptive voltage only takes effect at ratios > max single core boost ratio which is 50x in 9900ks case (so if you set 1.5v at 50x for eg, it gets ignored till you go up to 51x)

2) any volt set below the cpu's default vid gets ignored (eg setting 51x ratio with 1.1v when cpu's vid is 1.25v at 50x will get ignored)

on client you still have AC/DC loadline to manipulate to undervolt proc to a certain extent (you have no room to undervolt when AC/DC LL is already at minimum)
to undervolt further you will have to rely on VRM Loadline.

so in view of the above:

comparing a different cpu on different boards does not shed much information, number 1 (specially with 8700k when you will be way above the single core boost of 47x making adaptive easier to use)

number 2 : the volt sensing is different on maximus Xi than on other boards , its sensing on-die volt compared to traditional off-die sense
https://www.youtube.com/watch?v=4x3A6V2kSiU


in your case the straightforward and very logical way to run adaptive is to run the minimum LLC that is stable at 5g (since any v setting gets ignored by proc till it goes to 51x)
EVEN IF YOU COULD run a vid lower and a higher LLC to match, it typically runs cooler with a higher vid and a lower LLC. (you can verify this just by being on manual mode)

[Zammin]

Shamino wrote:
welcome to adaptive voltage

understand the criteria for adaptive voltage to take effect :

per intel design:
1) adaptive voltage only takes effect at ratios > max single core boost ratio which is 50x in 9900ks case (so if you set 1.5v at 50x for eg, it gets ignored till you go up to 51x)

2) any volt set below the cpu's default vid gets ignored (eg setting 51x ratio with 1.1v when cpu's vid is 1.25v at 50x will get ignored)

on client you still have AC/DC loadline to manipulate to undervolt proc to a certain extent (you have no room to undervolt when AC/DC LL is already at minimum)
to undervolt further you will have to rely on VRM Loadline.

so in view of the above:

comparing a different cpu on different boards does not shed much information, number 1 (specially with 8700k when you will be way above the single core boost of 47x making adaptive easier to use)

Hi, thanks for your reply

So if I understand you correctly, I'm seeing this strange behavior in adaptive mode because I am not exceeding the maximum single core boost speed, even though all cores are sync'd? I suppose that would explain why I did not see this behavior on my 8700k.

Shamino wrote:
number 2 : the volt sensing is different on maximus Xi than on other boards , its sensing on-die volt compared to traditional off-die sense
https://www.youtube.com/watch?v=4x3A6V2kSiU

Yes I am aware of this already, I've seen elmors post about it on OCN along with the graph he posted. Thanks for the link to that video though.

Shamino wrote:
in your case the straightforward and very logical way to run adaptive is to run the minimum LLC that is stable at 5g (since any v setting gets ignored by proc till it goes to 51x)
EVEN IF YOU COULD run a vid lower and a higher LLC to match, it typically runs cooler with a higher vid and a lower LLC. (you can verify this just by being on manual mode)

Sorry but I don't fully understand this part. I know that LLC7 is the minimum stable LLC for manual mode, I've tried LLC6 and increased target voltage to compensate but it just won't be stable. I spent a few hours trying today. Someone on OCN said the same thing about lower LLC having better transient response and requiring a lower VMIN but I can't seem to replicate that. Even at the same voltage under load with LLC6 it's not stable, where as LLC7 is.

So knowing that LLC7 appears to be the most stable for my CPU in manual mode, what should I do for adaptive mode? When I run LLC7 in adaptive mode I get overshoot at idle and at most, only a very small drop in voltage under load (from target 1.33V to 1.32V). When I run LLC6 it idles at target voltage and drops about the right amount under load but is unstable. The only thing I haven't tried is maintaining LLC7 in adaptive mode and lowering the voltage target. Is that worth trying?

Thanks for your advice.

[Shamino]

Zammin wrote:
Hi, thanks for your reply

Sorry but I don't fully understand this part. I know that LLC7 is the minimum stable LLC for manual mode, I've tried LLC6 and increased target voltage to compensate but it just won't be stable. I spent a few hours trying today. Someone on OCN said the same thing about lower LLC having better transient response and requiring a lower VMIN but I can't seem to replicate that. Even at the same voltage under load with LLC6 it's not stable, where as LLC7 is.

So knowing that LLC7 appears to be the most stable for my CPU in manual mode, what should I do for adaptive mode? When I run LLC7 in adaptive mode I get overshoot at idle and at most, only a very small drop in voltage under load (from target 1.33V to 1.32V). When I run LLC6 it idles at target voltage and drops about the right amount under load but is unstable. The only thing I haven't tried is maintaining LLC7 in adaptive mode and lowering the voltage target. Is that worth trying?

Thanks for your advice.

I know that LLC7 is the minimum stable LLC for manual mode, I've tried LLC6 and increased target voltage to compensate but it just won't be stable.
=> you mean adaptive right? and at 50x ratio right? now wouldnt that still mean that increasing target voltage does nothing becos its not > 50x ratio and hence gets ignored? so bascially the vid doesnt change becos criteria 2 is not fulfilled.

The only thing I haven't tried is maintaining LLC7 in adaptive mode and lowering the voltage target. Is that worth trying?
=> that still wouldnt work becos criteria 2 is not fulfilled. your target gets ignored

your last paragraph is basically saying level 7 is too shallow and level 6 is too steep.
=> you want just the right amount of v with adaptive , only way is to do this:
remain level 6 50x adaptive auto.
manually tweak ac/dc ll. 0.01/0.01 value is where you're at and u are unstable
increase this to 0.03/0.03, etc etc till you get what you need with VRM LLC 6

[Zammin]

Shamino wrote:
I know that LLC7 is the minimum stable LLC for manual mode, I've tried LLC6 and increased target voltage to compensate but it just won't be stable.
=> you mean adaptive right? and at 50x ratio right? now wouldnt that still mean that increasing target voltage does nothing becos its not > 50x ratio and hence gets ignored? so bascially the vid doesnt change becos criteria 2 is not fulfilled.

No I meant manual, I've tested LLC7 and LLC6 in manual mode to see if I could get it stable at the same or lower load voltage with LLC6, but it isn't stable. People say that lower LLC has better transient response and can achieve lower VMIN and better temperatures, but I just can't seem to replicate it. By lowering the LLC to level 6 (in manual mode still) and increasing the target voltage in order to achieve the same 1.27V load it is unstable. However at LLC7 (in manual mode) it is stable at that same load voltage.

Shamino wrote:
The only thing I haven't tried is maintaining LLC7 in adaptive mode and lowering the voltage target. Is that worth trying?
=> that still wouldnt work becos criteria 2 is not fulfilled. your target gets ignored

your last paragraph is basically saying level 7 is too shallow and level 6 is too steep.
=> you want just the right amount of v with adaptive , only way is to do this:
remain level 6 50x adaptive auto.
manually tweak ac/dc ll. 0.01/0.01 value is where you're at and u are unstable
increase this to 0.03/0.03, etc etc till you get what you need with VRM LLC 6

So you are suggesting to leave the Adaptive Mode Additional Turbo Voltage setting on auto and simply adjust the IA/DC LoadLines instead? Okay I will give that a shot. I've never tried IA/DC LoadLines at anything other than Auto or 0.01. I'll let you know what the results are.