6600k OC Problem

Boosted7M

Commendable
Jul 21, 2016
18
0
1,510
Currently I am running an MSI Z170A-Pro motherboard and a 6600k. I wanted to try my hand a manually overclocking and was able to achieve a result of 4.6Ghz at 1.35v. I stress tested with Prime95 and Aida64 for 15min a piece, both passed with core 2 reaching a max temp of 66°C. I used the computer to watch some movies in a marathon and left it on overnight. I came back this morning, reset the computer to try to get to 4.7ghz, but was getting some artifacting on the desktop when I tried to run aida64. I tried to drop the multiplier in 100mhz increments until it went away, not changing the voltage. It went away at 4.4ghz. So my questions are this could my VRMs have just gotten tired? should I be worried? I also overclocked my GPU by 70mhz base and 70mhz mem, could that be the culprit? Any help would be appreciated, I might just go back to using the OC Genie present if I am at risk.
 
If you were at 4.6 @1.35v, then you scored a good chip.
The motherboard automated oc apps are often not optimal and perhaps over aggressive. Stick with bios overclocking.
It might be worth a check to see if your motherboard has bios updates germane to your issue.
DO NOT update just on speculation; it is always risky.

Here is my simple skylake overclocking guide.
----------simple skylake overclocking----------------

I suggest the following:
Reset all including ram to default.
Install a stress tester like OCCT. Not prime95 or IBT which are unrealistic loads.
Then, leaving all voltages on auto, gradually raise the multiplier from the stock 40 to a higher number.
Stress test with occt. It will shut down the test if temperatures reach 85c.
You will run out of safe vcore before you run into thermal limits.
Monitor vcore. CPU-Z will tell you.
While Intel spec says 1.5v is a limit, more like 1.4 would be max.
When you reach your comfortable multiplier limit, stop.
use only the maximum ram xmp that will run with the stock 1.2v voltage.
Overclocking ram to 1.35v to get faster ram speeds will reduce your overclock.
Faster ram is not that helpful since it is accompanied by higher latency.
Here is a study:
http://www.silentpcreview.com/article1478-page1.html

When you are done, implement speedstep and adaptive voltage.
That will reduce your multiplier and vcore when there is little to do.

As of 10/11/2016
What percent can get an overclock at a somewhat sane 1.40v Vcore.

5-6600K

4.9 14%
4.8 38%
4.7 67%
4.6 87%





 

Boosted7M

Commendable
Jul 21, 2016
18
0
1,510
I am willing to try what geofelt suggested. I looked at the bios updates, but none of the notes address overclocking issues, just microcode updates and memory & storage compatibility updates. What I really want answered is why would the computer work perfectly well at 4.6 for about 18 hours straight then turn half my screen black when I try and open a program. Could it be that the system had just been running too long? Is it possible that is was the GPU OC that I did. Could it be that I haven't updated my GPU driver? Additionally my bios does not want me to go past 1.355v vcore, possibly due to the limited power phases the motherboard has.
 

Themastererr

Respectable
May 22, 2016
1,101
1
2,660
It's hard to say exactly why. The system is just unstable.

If you want to make sure it's 100% stable, overclock, then run prime95 for 12-24 hours. If you can past this test you're probably stable. If you can't, you're not.

 

Themastererr

Respectable
May 22, 2016
1,101
1
2,660
Nah, the CPU is fine. Just make your temps are in check.

The VRM's like you said before aren't the best. However when you compare it to AMD FX systems that were pushing 1.55Vcore @ 125watt TDP - 1.35V doesn't sound too bad. Make sure there is airflow around the VRM's though. Whether it be from your CPU cooler or case fans.

The GPU is a little harder to test for stability. I found Unigine heaven/valley to be a decent test of stability. I use to leave a game like GTA V open overnight to test that stability when I knew my CPU was good.

 

Boosted7M

Commendable
Jul 21, 2016
18
0
1,510
I just wanted to jump on and say that when the vcore setting is left on auto my motherboard doesn't increase voltage to match clockspeed. It just targets 1.2v and leaves it. So when I got to running occt at 4.4ghz it would crash after 1minute of test because the vcore wouldn't go over 1.21v. Trying 4.4 at 1.3v with occt to see what happens.
 

Boosted7M

Commendable
Jul 21, 2016
18
0
1,510
Alrighty, after a day and a half of testing (still testing rightnow) I was able to achieve 4.5Ghz at 1.3v confirmed after prime95 30min, Aida64 for 30min, and occt:cpu for a little over 3 hours. Now going for 4.6Ghz at 1.35v again (prime95 for 30min, working on a 2hour test of aida64 right now) but I backed down my OC on my GPU. Which more and more I think is the culprit for the display artifacting. Funny thing I have hwmonitor open during my Aida test. While looking at both, my cpu core volts on Aida says it is 1.385v, but vcore on hwmonitor says the max has been 1.352v. Both say my max VID has been 1.385v. Can anyone explain that?