AI suite 3 configuratoin (CPU temperature)

[RGAlix]

Hi everyone !

I'm new on the forum, and very inexperienced. I just finished my first build with an Asus Maximus VIII Hero and an Intel i7 7700K, and when I use the 5-Way optimisation in AI Suite 3, it seems to be automatically overclocking my CPU to crazy numbers. When I restarted my computer after the process, it wouldn't boot because the CPU temperature was too high. I had to do a Clear CMOS to be able to boot again. AI Suite never asked me for my preferences in terms of overclocking when it ran its tests, it simply chose insane settings for me (I don't remember the precise number).

How can I run it (or some other tool) to optimize my system without compromising its stability ?

Thank you very much !
Alix

 

[Johan45]

Personally I try to avoid using AISuite for anything. Quite often it's more trouble than it's worth. Their OC optimization is terrible IMO. I have tried it. Drops memory down and you're right increases CPU beyond stability. I do all my work in the BIOS

 

[RGAlix]

Ok, I don't have enough knowledge to set it up manually. Thank you for the answer, though !

 

[Johan45]

What are you trying to set up. There's a lot of good people here that could help you through it. Once you start it just gets easier.

 

[RGAlix]

I actually don't have a specific goal in mind. I just thought that AI Suite would optimize my components so that they would work best together, but I don't have a set performance to hit. I also have no idea how you're supposed to know how much overclocking is too much : is there a way to know what my cooler will support ? Are there other limiting factors ? Are there other operations to be done to optimize a build ?

Thanks for your help.

 

[Johan45]

A good start would be for you to add your components to a post or better yet your signature. Then we know what you're working with.
When overclocking you'll need CPUz, HWInfo64, and a stability test like AIDA64, Prime 95 or IBT/linpak. This will test for stability and show you when things are getting too warm.
Then it's a matter of testing, making changes and testing again until you find a good balance.

 

[RGAlix]

Ok, great. So, I can change settings in the BIOS, stress-test it with a program while monitoring the temperature, and go on until it's unstable ?

Two questions : how do I know if my components will take the new voltage ? And how do I choose between changing the multiplier or the base clock ?

I've added my components in my signature.

- - - Updated - - -

Woops, here is my signature. And just in case it doesn't show :

Motherboard : ASUS-MAXIMUS-VIII-HERO
CPU : Intel Core Kabylake i7-7700K
Cooler : Cooler Master Hyper 212 Evo
GPU : MSI Radeon RX 460
Memory : G.Skill Ripjaws V 16GB DDR4-3200Mhz
PSU : EVGA 550 G3
Case : Antec P9 Window

 

[Johan45]

To start with, just use the multiplier. No need to adjust the BCLK. Your cooler is OK but not a 5.0 GHz cooler, we'll see how far it can take you. You're good on the CPU up to ~ 1.4V but the cooler may not take you that far temps under stress high 80s low 90s is a good place to stop.
It's best to get a baseline before you start testing. Go to BIOS and hit F5 then save and reboot back to BIOS. Set the XMP and when it askes about multi core enhancement say yes. Save and boot to Windows. Test and monitor then post some pics of the system while it's under load so we can see voltages and temperatures.

 

[RGAlix]

Okay, thanks ! I will replace my PSU for a 850W, which I will receive tomorrow or the day after, so I will wait until it's installed. I have looked at the apps you mentioned, and I don't really understand the difference between them. I don't see any that tests the stability, or monitors the CPU temperature, but all of them seem to stress-test the CPU. How would you use them together ?
Also, when I enable XMP, in MultiCore enhancement option, I can only choose Auto or Disabled.

 

[EarthDog]

Aida64, p95, ibt/linpack are stress test apps. A64 also monitors temps. Hwinfo64 displays temps and a ton of other data. Youd run a stress test with hwinfo running to see the temps and other info.

 

[Johan45]

Leave MCE at auto then

 

[RGAlix]

Hi !

I've changed my PSU for an EVGA 850 BQ and I have started playing with OC settings, and I'm a bit puzzled. I started with the default settings (42x), and measured 1,27V / 72°C. I upped the multiplier to 46x and got 1,27V / 82°C. I bumped the voltage to 1.3V and measured 1,21V / 65°C. After that, I kept changing the multiplier (48x, 52x), and always measured 1,21V / 65°C. Am I missing something ? Also, should I modify the baseclock ?

Another weird fact : I tried to change the DRAM frequency from Auto to 3200 (my memory frequency), and the computer wouldn't boot. I had to revert it back to Auto.

Here are some pics with settings as captions :

https://imgur.com/a/e02W58B

Thanks a lot for your help !

 

[EarthDog]

Set xmp profile for your memory, it should do it automatically.

Set voltage to 1.25v manually and see how far it goes. No need to touch bclk.

 

[RGAlix]

I've done that, and the temperature reaches 93°C at 53x, so I'm staying with 52x (60°C). How did you come up with 1.25V ? And why do I get the same actual voltage when I set it to 1.25 or 1.3V ? Have I reached a certain cap limit ?

Is there more tweaking left to do ?

Also, I've done what you said about the memory, and I got the same problem : no boot.

 

[EarthDog]

30c drops from 100mhz isnt possible without a significant voltage change. That makes zero sense.

Re the ram, manually set vccsa and vccio .1v over stock amd see if that helps.

Is the bios updated to the latest version on your motherboard?

I came up with 1.25V as a starting place for your cooling.

And why do I get the same actual voltage when I set it to 1.25 or 1.3V ? Have I reached a certain cap limit ?

Not sure what you mean here... are you saying if you set it higher it only reaches 1.25/1.3v? Setting it at 1.25V and seeing 1.25V under load is a good thing.

 

[RGAlix]

My BIOS was indeed not up to date, I just updated it. I'm following your suggestions now. What do you mean, by 'Re the ram' ? And are VCCSA and CPU System Agent Voltage the same thing ?

I have started to do some tests with default voltages again after the BIOS update and a question comes up : when I stress-test at 49x with CPU-Z, I get 78°C. But at the same multiplier with Prime95, I go all the way up to 93°C. Which one should I care about ?

And why do I get the same actual voltage when I set it to 1.25 or 1.3V ? Have I reached a certain cap limit ?

I meant that with both 1.25V and 1.3V in the BIOS, I measured 1.211V under stress.