Skylake standby issue with Vcore adaptive mode

SilverSliver said:
I don't know why guys are using adaptive and auto voltage for overclocking. Manual voltage, LLC to 5 - all problems solved.
Click to expand...

Actually Dan-D-man doesn't seem to agree;

http://hardforum.com/showpost.php?p=1041825145&postcount=97

But I'm guessing they were saying something similar while writing their sugar coated reviews which often omit things others may be concerned about...Or probably saying "I don't know why guys use standby, what do you need standby for anyway".
 
SilverSliver said:
I don't know why guys are using adaptive and auto voltage for overclocking. Manual voltage, LLC to 5 - all problems solved.
Click to expand...

Finally got some time to answer you properly, well at least indirectly. Here's a post left in the ROG forum where this issue has garnered more interest;

"Adaptive mode (AM) combines the attributes of manual (MM) and offset (OM) modes. AM behaves the same way as MM under heavy load and the same way as OM under light load. It wears two hats and is Jack and master of both modes. That's until one resumes from standby and the equivalent manual component setting is reset to stock.

Nobody seems to be concerned about high Vcore at light load so they just continue to combine MM with LLC settings to limit Vcore under heavy load. Either their computers never idle or Vcore > 1.4 when the computer is idling means little to them. Heck, the Gamer's preset that comes with the Gene sets MM Vcore to 1.425 and LLC to max.

What's frustrating about the adaptive/standby issue is that one does not know who to believe. It's probably in Asus' interest to say as little as possible....Is it confined only to certain Asus motherboards, certain motherboard revisions or BIOS level? I see where the Z170-A BIOS is now at 901 while the Gene is still at 801 so was the OP's issue resolved? The OP is nowhere to be found. Tracked him (or her) to the overclocker.net Skylake forum but he disappeared again."


Just to clarify...The OP in that thread had (or still has) the same issue with his Z170-A board. Also, my "return window" is still open until Thursday and Micro Center is only a 3 mile drive from here. Furthermore, I haven't even activated Windows yet.

The last thing I want to see happen is being saddled with an issue that needs a new MB revision and/or CPU stepping to correct. Haswell early adopters can probably see where I'm coming from.
 
SilverSliver said:
I don't know why guys are using adaptive and auto voltage for overclocking. Manual voltage, LLC to 5 - all problems solved.
Click to expand...

adaptive is not the same as Auto

I use adaptive as well, if you use Manual the board will keep the voltage solid not matter if it is idle or not which doesn't make any sense.

On Auto it just does whatever it wants and that is crazy

I use adaptive set to the exact voltage I want to max out at with LLC 4 and it will boost slightly above what I have set like .02v at max load, but it still drops down to .8v at idle

If I try to use Offset it will let the voltage go way higher than I want and I was not able to control it.

My PC is set to sleep=never so I am not going to have the OPs problem.

If I am understanding this issue correctly the manual component of adaptive is somehow being set to AUTO when returning from sleep, if that is true I agree that is a major issue.
 
zaniix said:
adaptive is not the same as Auto

I use adaptive as well, if you use Manual the board will keep the voltage solid not matter if it is idle or not which doesn't make any sense.

On Auto it just does whatever it wants and that is crazy

I use adaptive set to the exact voltage I want to max out at with LLC 4 and it will boost slightly above what I have set like .02v at max load, but it still drops down to .8v at idle

If I try to use Offset it will let the voltage go way higher than I want and I was not able to control it.

My PC is set to sleep=never so I am not going to have the OPs problem.

If I am understanding this issue correctly the manual component of adaptive is somehow being set to AUTO when returning from sleep, if that is true I agree that is a major issue.
Click to expand...

"adaptive is not the same as Auto"

I think SS meant "I don't know why guys are using anything other than manual".

"If I try to use Offset it will let the voltage go way higher than I want and I was not able to control it."

This is especially true if LLC isn't set to auto when using Offset (+) mode.

"My PC is set to sleep=never so I am not going to have the OPs problem."

What happens if you manually put it to sleep and wake it a few seconds later then run something like Realbench while monitoring Vcore? You can run just the Realbench "H.254 Video Encoding" benchmark instead of the stress test. Just moving the mouse will stop it before a crash if you see Vcore is too low.
 
Eldata said:
"adaptive is not the same as Auto"

I think SS meant "I don't know why guys are using anything other than manual".

"If I try to use Offset it will let the voltage go way higher than I want and I was not able to control it."

This is especially true if LLC isn't set to auto when using Offset (+) mode.

"My PC is set to sleep=never so I am not going to have the OPs problem."

What happens if you manually put it to sleep and wake it a few seconds later then run something like Realbench while monitoring Vcore? You can run just the Realbench "H.254 Video Encoding" benchmark instead of the stress test. Just moving the mouse will stop it before a crash if you see Vcore is too low.
Click to expand...

I set adaptive at 1.20v at stock speed , LLC set to 4 put it to sleep brought it back and ran real bench while monitoring voltage. Everything worked as expected, ramped up from .8 to 1.24v at full load no crashes. I only ran it for like 15 minutes.

I am using ASUS Maximus Ranger with the newest bios 0801

maybe this is only an issue on specific boards or maybe I am not understanding the issue.

edit: well right after I posted I rebooted from windows and on reboot it stopped seeing my SSD boot drive, I reloaded again went into bios and it was just gone.
I shut it down and removed power then it booted it fine, so yeah screw sleep mode. rebooted several times after that and it seems fine.

and I should also make it clear all bios changes were done in the BIOS, I never use AI suite for anything other than the fan expert settings as I do not trust it.
 
Last edited:
zaniix said:
I set adaptive at 1.20v at stock speed , LLC set to 4 put it to sleep brought it back and ran real bench while monitoring voltage. Everything worked as expected, ramped up from .8 to 1.24v at full load no crashes. I only ran it for like 15 minutes.

I am using ASUS Maximus Ranger with the newest bios 0801

maybe this is only an issue on specific boards or maybe I am not understanding the issue.

edit: well right after I posted I rebooted from windows and on reboot it stopped seeing my SSD boot drive, I reloaded again went into bios and it was just gone.
I shut it down and removed power then it booted it fine, so yeah screw sleep mode. rebooted several times after that and it seems fine.

and I should also make it clear all bios changes were done in the BIOS, I never use AI suite for anything other than the fan expert settings as I do not trust it.
Click to expand...

Well, thanks for going through with it and sorry about the SSD issue. I haven't had any issues with sleep other than with the adaptive voltage reverting to stock. Anyway, your adaptive voltage setting is too low for the results to be conclusive in my case.

If I set the adaptive voltage to 1.3V and LLC to 6, for a 4.3GHz thru 4.5GHz OC, the Vcore voltage is 1.296V when running IBT. Repeating the test, after going to sleep for 3 seconds or longer, results in Vcore limited to 1.248V. I have assumed that this is some stock value (or thereabout). I chose IBT only because it is stable at this voltage with a 4.5GHz OC. Realbench would crash in a few seconds with [email protected].
 
I misunderstood your issue. So you are getting too little voltage to support the overclock after return from sleep.
 
zaniix said:
I misunderstood your issue. So you are getting too little voltage to support the overclock after return from sleep.
Click to expand...

Yes, that's the problem. The CPU appears to be ignoring what's in the BIOS and reverting to some stock value for 4.2GHz after waking from sleep. Nothing like this happens with manual or offset mode so the workaround is to just use manual mode.

The only input I've seen from Asus comes from Raja;

"This can happen with Adaptive. I think Intel need to revise the microcode to change the voltage step changes with core ratio."

Link to the thread where he states this;

https://rog.asus.com/forum/showthread.php?74685-OC-Adaptive-Voltage-Mode-and-Win-10-Sleep-Problems!

A BIOS update is the only solution I'm prepared to accept. Hence the investigation before returning both board and CPU.
 
Last edited:
zaniix said:
adaptive is not the same as Auto

I use adaptive as well, if you use Manual the board will keep the voltage solid not matter if it is idle or not which doesn't make any sense.
Click to expand...

My 2600k was overclocked using manual voltage and all sleep/c states disabled for almost 5 years. If temps are good and voltage is at a reasonable level, I don't see why it doesn't make sense. It likely cost me a couple bucks extra a year.

For me, I want to know exactly what the voltage is and pin it there. I was getting very unstable clocks and wild voltage swings using adaptive on the 6700k - 500mv above what I was calling for at load - and the overclocks were still unstable using the Asus/ROG guide. Manual let me dial the voltage way down, and overclock higher at the same time.
 
SilverSliver said:
My 2600k was overclocked using manual voltage and all sleep/c states disabled for almost 5 years. If temps are good and voltage is at a reasonable level, I don't see why it doesn't make sense. It likely cost me a couple bucks extra a year.

For me, I want to know exactly what the voltage is and pin it there. I was getting very unstable clocks and wild voltage swings using adaptive on the 6700k - 500mv above what I was calling for at load - and the overclocks were still unstable using the Asus/ROG guide. Manual let me dial the voltage way down, and overclock higher at the same time.
Click to expand...

My 2600K build isn't going anywhere anytime soon. It's right beside me next to the 6700K one. KVM switch to it when stress testing the 6700K. Use offset Vcore there for the same reason I don't want to use manual with the new build.

"I was getting very unstable clocks and wild voltage swings using adaptive on the 6700k - 500mv above what I was calling for at load - and the overclocks were still unstable using the Asus/ROG guide."

Useful info. I see the Hero BIOS is at 802. FWIW, I don't have this issue. Manual and adaptive modes behave identically under heavy load, with the same Vcore setting, as they should. Except of course after waking from standby. There should be no denying that adaptive is potentially the best mode when working properly.
 
I just checked if i have this issue with my 4970k and Asus Sabertooth Mark 2 and everything looks normal when waking up from sleep.

I have it set Adoptive 1.275v, ran tests before and after sleep, its all good.
 
TroyX said:
I just checked if i have this issue with my 4970k and Asus Sabertooth Mark 2 and everything looks normal when waking up from sleep.

I have it set Adoptive 1.275v, ran tests before and after sleep, its all good.
Click to expand...

No surprises there. It's a Skylake issue as the thread subject indicates. Haswell doesn't have this problem.
 
Well shit, this explains a lot.

I was wondering why I keep coming back to a BSOD whenever my PC was left for awhile....
 
Blue Falcon said:
Well shit, this explains a lot.

I was wondering why I keep coming back to a BSOD whenever my PC was left for awhile....
Click to expand...

A quote from the HARDOCP Z170-A review;

"The Z170-A was no different. It felt very mature while working with it. With the CPU Vcore in adaptive mode with an offset equaling 1.35v I had no trouble achieving a clock speed of 4.7GHz which seems to be the maximum this Core i7 6700K can hit without over-heating."

A quote from the HARDOCP Z170-Deluxe review;

"Another factor in my overclocking adventures with the Z170-Deluxe had to do with how voltage was applied. In offset or adaptive mode I found that getting results was very easy. Hard coding the voltage manually did not achieve the same stability. This is a switch from Haswell. Haswell handled adaptive voltage just fine, but it was often easier for us reviewers to lock down the voltage manually and run our stress tests."

A forum quote from Dan-D-Man-Wid-D-Plan;

http://hardforum.com/showpost.php?p=1041825145&postcount=97

So we use adaptive mode instead of manual, only to find out that between Intel and Asus, the darn thing is broken and driving people up the wall.
 
You must log in or register to reply here.
Back
Top