• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Newbie finding best settings for 8086k, Noctua NH-U14s and Prime Z390A

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

granolamonster

Registered
Joined
Nov 6, 2018
Hi all,

First of all hello, I'm new to messing with OC'ing, please feel free to throw me pointers!

I was fortunate enough to win one of the 8086k's in intels comp a few months back and have finally found enough $$$ to buy a motherboard and ram (corsair lpx 3200) to get it going. I've paired it with my previous Noctua NH-U14s which seems to cope fairly happily.

I'm now trying to find the best settings for my system. I'm not really interested in truly OC'ing this CPU as I cannae afford a new one or a new cooler to cope. At the moment I'm just trying to get my head around the classic with/without asus MCE debate.

I've tried a few combo's so far:

1) XMP enabled and MCE enabled (all other settings default):

Max CPU = 5 ghz
Avg CPU under load = 4.3 ghz

Max Core temp = 66-70oC
Avg Core temp under load = 55-60 oC

Max Core voltage = 1.399v
Avg Core voltage under load = 1.175v

2) XMP enabled and MCE disabled (all other settings default)

Max CPU = 4.6 ghz
Avg CPU under load = 4.1 ghz

Max Core temp = 65oC
Avg Core temp under load = 60 oC

Max Core voltage = 1.25v
Avg Core voltage under load = 1.175v

3) XMP enabled and AI O'Cing enabled

Max CPU = 5.2 ghz
Avg CPU under load = 4.8 ghz

Max Core temp = 91oC
Avg Core temp under load = 70 oC

Max Core voltage = 1.399v
Avg Core voltage under load = 1.375v

My question is really which of 1 and 2 should I really be running with? I realise option 3 is pushing things too far with my current cooling. Option 2 seems very safe but I would prefer it if I could hit the 5ghz turbo once in a while. What would you all advise? Is running with MCE enabled going to put too much of a strain on CPU and cooler?

All advice much appreciated!

Cheers!
 
Last edited:
If you're going to stick with relatively "stock" settings then use the MCE enabled. Will give you more all core performance and should stay within your cooling limits. You never did explain what you used to test the stability and thermals though.
 
You seem to be contradicting yourself when you say, "I'm new to messing with OC'ing" in your first sentence and then state in your third sentence that "you are not interested in truly overclocking this CPU" because you can't afford to buy a new one or a cooler to cope.

And then under 3) you say you have AI O'Cing enabled.

Which is it?

First, it is highly unlikely that you will harm the CPU by overclocking if you do it even close to responsibly.

Second, your current cooling should be adequate for a modest overclock. Have you done any temp monitoring and stress testing? You don't say. How do you know your cooler won't handle overclocking?

Third, if you allow AI to automatically overclock your CPU that would be the worst possible overclock option as those bios auto overclock genies generally apply way more voltage than is needed and typically underclock the RAM.

Fourth, your hardware information is very limited. We know you have an 8086k, Corsair lpx 3200 RAM and a Noctua NH U-14s cooler. What motherboard do you have? What PSU? What case? What video card? Please create a Signature with more complete information. You can do that by clicking on Settings in the upper right corner of any page and then clicking on "Edit Signature" along the left side.

By the way, Welcome to OCF!
 
Last edited:
Hi,

Thanks for responses - hopefully new signature will help with additional info.

I've been stress testing the CPU using Prime95 and testing whilst playing a few games.

I've only used the AI Oc function on the board just to see what it could do but temps of 90+oC make me think it is too much.

I guess what I am really trying to gauge is what a sensible max temperature really looks like and what an acceptable vcore is? For example, if I just use MCE (I know it isn't the most efficient approach) then I end up with these sort of figures - are these alright? Max CPU = 5 ghz
Avg CPU under load = 4.3 ghz

Max Core temp = 66-70oC
Avg Core temp under load = 55-60 oC

Max Core voltage = 1.399v
Avg Core voltage under load = 1.175v

If I go about manually changing the settings do you think I would be able to achieve similar performance but with lower vCore/temp? If so, can you give me a few pointers in the settings to get started with?

Thanks very much :D

PS, just realised sig doesn't include:
Storage: Samsung Evo 850 SSD 256
Secondary Storage: WD Black 1 tb * 2, WD Green 2tb
OS: Windows 10 pro 64

- - - Auto-Merged Double Post - - -

Oh, and using HWMonitor to get vCore/Temp figures.
 
Thanks for the additional info.

Nice system. I might suggest adding some additional case fans to the Fractal as currently it only has the two that it ships with. Especially since you are cooling the CPU with air.

Honestly, I would agree with Johan45 and just run the system at stock settings with MCE enabled. The reason I say that is the 8086k is already overclocked from the factory as it were since it is just a cherry-picked 8700k that has been juiced a little and given a higher binning. If you had good water cooling you could probably get another couple hundred mhz of speed out of it but the small additional performance increase would not be worth the expense and hassle to make it happen. Overclocking is more rewarding when you start with a low binned chip and clock it up quite a bit.
 
And yes, your current max stress test temps are quite acceptable. Anything up to about 95c is okay when stress testing or heavy loading. Just keep in mind that temps tend to slowly climb with extended intense use like in gaming. That's why I'd recommend some more case fans.
 
Thanks for the feedback, much appreciated.

I've had a go at tweaking a few options for myself. I've found that it'll run at 50x and only 1.25v under load but it doesn't half get hot and I don't need a hoover in my office :D. Now stress testing 45x at 1.20v under load achieving a more reasonable max temp of 80oC so far.

However, I'm assuming this setup means that my CPU will always run at 45x, even at idle - is there some middle ground I've missed where you can specify the voltage and MCE? Or am I just over thinking things?
 
You can use offset voltage and leave all power savings active, that should allow the CPU to ramp down volts and speed. Set Windows to balanced power as well.
 
Ah, I can't get my head around offset voltage. I've tried a few times but got it really wrong. Last time tried to put in an offset of -.7 and ended up with the pc trying to boot at 1.7v which it obviously failed with. What am I doing wrong?
 
You were one decimal point off and going in the wrong direction with you offset. Offset is how much voltage you want the bios to add or subtract to baseline under load. If you are overclocking you want a positive offset, not a negative one. For instance, if your stock, baseline voltage is 1.22 and you want to try 1.25 you would enter a value of +.03. From experience I can tell you that voltage increments need to be added in multiples of .01 to really make any difference.
 
????? (Also thanks :D)

So.......if I set the cores to sync at 50x, I would use offset to add a certain amount to the baseline underload, i.e., I could say 50x will run fine at 1.25 but under load will need to be 1.3? Does this mean I would need to set both the manual vcore of 1.25v and the offset of .05 in the asus bios?

Apologies for the many qs, the asus bios didn't make this connection apparent, it keeps the manual (I'm assuming this is what you mean by baseline) and the offset under the same set of drop downs making them look like they are mutually exclusive.

Thanks once again!
 
I've done a bit more testing, I think the following does what I'm after:

Core's sync'ed: x47
Adaptive Voltage: 1.2v
Offset: Auto
MCE: Disabled

In practice running Aidas stress test gives a max core voltage of about 1.26v, generally 1.15 under load, running at 4700 with a max temp of about 80oC, again generally 60-70oC under load.

Does that sound sensible or have I made some terrible noob mistakes?
 
The power/voltage management of modern processors, both AMD and Intel is complex and hard to nail down. Experimentation is in order.

Have you experimented with LLC (Load Line Calibration)? It might do what you are trying to do better than offset or adaptive voltage adjustments. Especially since you seem to be getting a lot of voltage drop under load. That's exactly what LLC was designed for. A lot of us here at OCF just set the vcore manually and allow LLC to help keep the vcore up to snuff under load. No, you haven't made any terrible noob mistakes. You are doing very well actually. That CPU should be capable of 5.0 ghz easy if your cooling is up to it, which I know is a question mark.



Make sure your run AIDA64 extreme overnight if that is going to be your stress tester.
 
Are we sure that is a drop under load or an effect of auto voltage and offset with varying aida loads?

It is difficult to tell with auto voltage what the vdroop is considering the voltage varies for other reasons. To me, there isnt much point to use LLC with an offset.

Seeing vdroop of over .1v is quite a difference and not typically seen. The vast majority of motherboards, both z370 and x299, showed quite little in fact. I expect to see similar results with most z390 boards. :)
 
I've done a bit more testing, I think the following does what I'm after:

Core's sync'ed: x47
Adaptive Voltage: 1.2v
Offset: Auto
MCE: Disabled

In practice running Aidas stress test gives a max core voltage of about 1.26v, generally 1.15 under load, running at 4700 with a max temp of about 80oC, again generally 60-70oC under load.

Does that sound sensible or have I made some terrible noob mistakes?

So, this lot was run with CPU SVID enabled - does that explain the variation?
 
Ok, I've had another go with the following settings, not all of which I understand.......:eek:

Cores Synced: 46x (I'm still trying to get the temp to top out at 80oC)
MCE: Disabled
CPU SVID: Disabled
Manual vCore: 1.15v (is this too low? are there any advantages to trying to get it lower?)
AVX Offset: 0
Max vCore under load: 1.154v
Min vCore no load: 1.146v
LLC: Level 6 (again not sure if this might not be better set to auto as I haven't fully got my head around LLC)
Max Temp: 80oC
Running at 4600 mhz consistently under AIDA for 40 minutes

However, I'm beginning to wonder if these temps are reliable from HWMonitor - got any suggested alternatives?

Are there any other settings I can explore to try getting this temp under control (other than just lowering the multiplier)?

I think it is going to be worth finding this sweet spot myself rather than use MCE, this way I top out a 4.6 rather than 4.3 and within my thermal limits :D
 
The goal with overclocking is to use the least amount of voltage for a given clockspeed and be stable... so no, the lower the better!

SVID/auto certainly explains the voltage difference you saw earlier. Notice the difference now is almost non existant?

Try coretemp or realtemp.

Multiplier itself doesn't do much... voltage is the big difference maker in temps.

Manual overclocking is better, yes.

Just an FYI, to type the actual degree symbol instead of typing the letter 'o', hold down alt and type 0176. Or, just put the C next to the number...we know its degrees.:)
 
You can also use HWInfo64 it's free and gives a ton of info.
E_D I had to check that for the degree symbol I use Alt and 248 which also works
 
Thanks both, I'll try HWInfo64 then when I get home later.

I'll keep going on finding the lowest stable vCore then - I'm assuming the advice then is to stick with a manually defined voltage. If you are manually applying the voltage constantly does this mean the CPU never spins down when doing nothing?

What do you think I should do in terms of LLC? Should I specify it (if so is level 6 alright?) or leave it to auto?

Also, assume an AVX offset would help with stability, is an offset of 1 enough?

Finally - assume a stable max of 80ºC (apparently just alt and 0 on a mac :p) is a pretty safe target given that I'm using air cooling?

Cheers :D
 
If you are manually applying the voltage constantly does this mean the CPU never spins down when doing nothing?
The voltage, yes.

To add to this, now that you know X.xx volts is stable, you can then use an offset if you choose. For example. If 1.15V is stable at your clockspeed. Reboot with the voltage on auto and see what the LOAD voltage is. Then use an offset to reach your known voltage value. If your CPU needs 1.25V on auto, use a -100mv offset to reach 1.15V under load.

What do you think I should do in terms of LLC? Should I specify it (if so is level 6 alright?) or leave it to auto?
Whatever it needs. LLC is an OPTIONAL function. If you do not have a lot of vdroop (like in your case), I leave it on auto. If you have a lot, my goal is to have the load voltage match what I set in the BIOS. All too often I see advice to just use it, but, as often as I see it on these platforms, 9/10 times, like this one, it doesn't seem to be needed.
 
Back